Professional Documents
Culture Documents
Social Research Methods (Alan Bryman Edward Bell)
Social Research Methods (Alan Bryman Edward Bell)
Published in Canada by
Oxford University Press
8 Sampson Mews, Suite 204,
Don Mills, Ontario M3C 0H5 Canada
www.oupcanada.com
Every effort has been made to determine and contact copyright holders.
In the case of any omissions, the publisher will be pleased to make
suitable acknowledgement in future editions.
Glossary 403
References 410
Index 426
Strategies for observing behaviour 137 Questions for Review and Creative Application 214
Questions for Review and Creative Application 268 Quantitative research and constructionism 328
Interactive Classroom Activities 269 Research methods and epistemological and
Relevant Websites 270 ontological considerations 328
Problems with the quantitative/qualitative contrast 329
12 Content Analysis 271 Mutual analysis 331
Chapter overview 271 Quantification in qualitative research 332
Introduction 272 Multi-strategy research 332
Personal documents 273 Two positions in the debate over quantitative and
Government documents 277 qualitative research 334
Official documents from private sources 279 Approaches to multi-strategy research 334
PART II consists of five chapters concerned with semi-structured or unstructured) and focus
quantitative research. groups, in which groups of individuals are
interviewed on a specific topic.
• Chapter 4 presents the fundamentals of quan- • Chapter 12 applies qualitative approaches to
titative research and provides the context for content analysis, a method used in the study
later chapters. of “documents” ranging from books, letters,
• Chapter 5 focuses on structured interviewing and newspapers to movies, chat lines, and
and the design of questionnaires. It delves television shows. It also examines two ways to
into how to write questions for both question- analyze language: conversation analysis and
naires and interviews. It also discusses how discourse analysis.
to compose a self-completion questionnaire, • Chapter 13 explores some approaches to
using data from already-completed question- the analysis of qualitative data, including
naires and interviews. grounded theory and coding.
• Chapter 6 covers structured observation, a
method developed for the systematic observa- PART IV moves beyond the quantitative/quali-
tion of behaviour. tative division to explore what the two approaches
• Chapter 7 deals with quantitative sampling: have in common, how they may complement each
how to select a sample and the considerations other, and how they may be combined in the same
involved in assessing what can be inferred research project.
from different kinds of samples.
• Chapter 8 presents a range of basic non- • Chapter 14 proposes that the distinction be-
technical tools for quantitative data analysis. tween quantitative and qualitative research
The emphasis is on how to choose a method of may be less fixed than is sometimes supposed,
analysis and how to interpret findings. In order and presents some ways in which they can be
to keep the focus on methodological concepts combined to produce multi-strategy research.
and interpretations, formulae are not discussed. • Chapter 15 provides guidance on writing up
research, an often-neglected area in the teach-
PART III presents five chapters on aspects of ing of the research process.
qualitative research. • Chapter 16 offers advice on conducting a re-
search project, taking readers through the
• Chapter 9 plays the same role for Part III that main steps involved.
Chapter 3 does for Part II. It provides an over-
view of the nature of qualitative research and Finally, the Appendix presents an easy-to-access
hence the context for the other chapters in this resource for successful research.
part.
• Chapter 10 discusses ethnography and partici- • The appendix explains how to use IBM
pant observation. The two terms are often used SPSS Statistics Software (SPSS) and NVivo
interchangeably to refer to the immersion of software to perform, respectively, the quan-
the researcher in a social setting, a technique titative data analyses described in Chapter 8
that is the source of some of the best-known and the qualitative data analyses discussed
studies in social research. in Chapter 13. The SPSS material has been
• Chapter 11 examines the kinds of interview updated to the latest version of IBM SPSS
that qualitative researchers conduct (typically (released in 2018).
Brief Contents
Guide to the Book
Special Features of the Book
vii
ix NEW! Organization. This fifth edition has
Acknowledgments xviii
Preface xix
been reorganized to better reflect how social
PART I Fundamental Issues in Social Research
1 General Research Orientations 2 research as a discipline is taught across
2 Research Designs 27
3 Research Ethics 50 Canada today.
Appendix to Part I The Ideal Stages of Research 69
outline a research story that has appeared in a MacEwan University, conducted qualitative inter-
views with 100 Indigenous and non-Indigenous
also noted that some teachers did not feel confi-
dent enough to address topics relating to Indigen-
parents and teachers in southern Ontario. The ous people, and were wary of giving offence. “The
major media outlet, illustrating how social research purpose of the study was to document the inter-
viewees’ perceptions of Ontario government
problem is that when you have people that are un-
comfortable and intimidated, the result is that we
policy directives designed to introduce Indigen- have educators that may not be doing it at all,” she
can have real impacts on our everyday lives. ous history, culture, and experiences into the cur-
riculum (Canadian Press, 2017).
said (Canadian Press, 2017). Milne recommended
that “Indigenous coaches” be used by teachers as
Milne found that the teachers she spoke to a learning resource. Some of the challenges she
were generally quite willing to incorporate In- identified included how to use appropriate, cul-
digenous perspectives into their classroom turally sensitive terminology when discussing In-
activities, and she observed that Indigenous par- digenous issues, and how to present the history of
ents were in favour of non-Indigenous teachers residential schools.
48 PARTIFundamentalIssuesinSocialResearch
KeyPoints
• Thereisanimportantdistinctionbetweenagen- • Replicability,validity(measurementandexternal),
eral research orientation (quantitative versus andtheabilitytoestablishcausationareimport-
qualitative)andaresearchdesign. antcriteriaforevaluatingthequalityofquantita-
• Thenomotheticapproachtoexplanationinvolves tivesocialresearch.
discoveringgenerallawsandprinciples. • Four key research designs are experimental,
• Nomothetic explanations must satisfy three cri- cross-sectional,longitudinal,andcasestudy.
teria of causation: correlation, time order, and • Threats to the establishment of causation are of
non-spuriousness. particularimportanceinnon-experimental,quan-
• Qualitative researchers usually take the idio-
graphic approach to explanation, which entails
titativeresearch.
• External validity is a concern with case studies
NEW! Expanded end-of-chapter
questions. Questions designed to
creating a rich description of a person or group (generalizability) and laboratory experiments
based on the perceptions and feelings of the (findings may not be applicable outside the re-
peoplestudied. searchenvironment).
10
Ethnography and Participant Observation
Chapter Overview
Ethnography and participant observation require extended involvement in the activities of the people
under study. This chapter explores:
• the problems of gaining access to different settings and ways of overcoming them;
• whether covert research is practicable and acceptable;
• the role of key informants;
• the different roles that ethnographers can assume in the course of their fieldwork;
• the function of field notes and the forms they can take;
• the role of visual materials in ethnography;
• bringing an ethnographic study to an end; and
• the issue of feminist ethnography.
Chapter-opening vignettes. At the Do you like to travel to places you’ve never been to
before? Have you ever observed a group of people
Ethnography and participant observation involve
placing yourself in a social environment that may be
you don’t know very well and wondered what it would foreign to you, and staying there for an extended period
beginning of each chapter, the topics to be be like to be a member of their group? Have you ever
witnessed profound human suffering and asked your-
of time. What kinds of groups or social settings intrigue
you? Non-governmental organizations? Political move-
addressed are introduced in an informal self how things ended up that way and how the people
suffering managed to endure? Would you like to give
ments? Sports teams? Criminal gangs? Hospital emer-
gency rooms? Women’s shelters? All of these can be
such people a voice or expose the hardships that they subjects of ethnographic and participant observation
and provocative way to help students grasp face? If so, doing or at least reading about ethnography
and participant observation should interest you.
research.
discussed in the chapter have been describe, it definitely feels different than prior to
[marriage]” (Green, 2010, p. 411). The people stud-
society and in the process lose their oppositional
tenor. Like other predictions for social change,
ied also mentioned that being married bestowed those regarding gay marriage can be tested only
used to study Canadian society. a sense of legitimacy on their relationship and with the passage of time.
1
bry29440_ch09_197-215.indd 12/17/18 06:55 PM
Chapter Overview
The aim of this chapter is to examine the fundamental assumptions upon which social research is based.
An important distinction commonly drawn by practitioners of social research—between the quantitative
and qualitative approaches—is explored in relation to those considerations. We will consider:
• the relationship between theory and research—in particular, whether theories and the hypotheses
derived from them are tested by gathering data (a deductive approach) or whether data gathering
is used as a means to create theory (an inductive approach);
• epistemological issues, such as whether a natural science model like the one used in chemistry
or biology is suitable for the study of the social world;
Chapter overviews. Each chapter opener
• ontological issues, such as whether the social world should be regarded as a reality external to
individuals over which they have little or no control, or as something that social actors may fashion
includes an overview that serves as a route
into their personal realities;
• how values and practical issues impinge on the research process; and map, alerting readers to what they can
• how these issues relate to both quantitative and qualitative research; a preliminary discussion,
followed up in Chapter 14, suggests that although the quantitative and qualitative orientations are expect to learn.
different, they complement each other.
Soon-Yi wants to find out why Indigenous people in The list of topics she could collect information on
Canada are more likely to live in poverty than other seems endless.
Canadians, but doesn’t know where to begin. Should Maybe rather than beginning her study by accumu-
she start by examining the history of colonialism and lating data, it would it be better to start out with some
conflict between Indigenous peoples and settler- hunches and then gather information to see whether
colonizers, such as disputes over land claims and treat- they are supported by evidence. For example, perhaps
ies, residential schools, or anti- Indigenous prejudice? the discrepancy in economic conditions is a manifesta-
Or how about gathering aggregate data on present tion of a centuries-old system of international domin-
conditions like residence patterns, economic activ- ance and exploitation. Similarly, it could have arisen
ities, the age structure, or educational trajectories? through a clash of civilizations and cultures. Then there
▲ LeonWang/Shutterstock
3 | Research Ethics 55
A key question asked in evaluation research is at its completion. The researchers also did a quali-
whether a new policy initiative or organizational tative analysis of the project by conducting five
change achieved its goals. Ideally, to answer that focus groups at the conclusion of the program. The
question the design would have one group that is quantitative results indicated that the participants
exposed to the treatment—the new initiative—and had higher levels of perceived overall health and
80 groupPart II Quantitative Research
a control that is not. Since it is often not feas- sense of community, and lower levels of physical
ible or ethical to randomly assign research partici- pain, when the program was over. The themes
pants to the two groups, such studies are usually that emerged from the focus groups included the
quasi-experimental. For instance, data gathered conclusions that the program provided the seniors
Boxes. Special feature boxes from people before a change may be compared
BOX 4.2 A multiple-indicator with structure and discipline, facilitated coping, re-
measure of another concept
with data acquired after; the “before” people quired hard work and effort, brought out their art-
provide in-depth examples of how become the control group, the “after” people the istic side, promoted social involvement, and made
In Hay’s (2014) study of secularization (see Box 1.3), frequency of religious attendance, was measured
experimental group. This approach has the added a positive contribution to the community.
the various research methods religious pluralism was measured using a single,
advantage that the two groups are basically the
with the question, “Do you currently attend church
Quantitative quasi-experimental designs in
five-point Likert item that formed part of an exten- temple or mosque?” Respondents who answered
discussed in the book have been same, making random assignment unnecessary. evaluation research go back a long way, but as the
sive survey of Canadians’ value systems. However, “yes” were then asked: “How often?” The response
Such a design was used to evaluate the effect Phinney et al. (2014) study indicates, evaluations
used in real research situations. secularization (the dependent variable in Hay’s
of a community arts program on the well-being of
choices were: “once a week or more” (given a code
based on qualitative research have also emerged.
analysis) was measured with several different indi- of 5 after reverse-coding), “monthly” (4), “every
older adults in the Vancouver area (Phinney et al., Although there are differences of opinion about
The boxes also list the advantages cators in order to tap into different dimensions of
2014). Over three years, four groups of participants
few months” (3), “once or twice a year” (2), and
how qualitative evaluation should be carried
the concept. One dimension, religiosity, was meas- “never” (1). A third dimension, concerned with the
and disadvantages of a particular took part in the collective creation of a physical out, there is consensus on the importance of,
ured by averaging the responses to three 10-point participants’ belief in the religion of their par-
work of art or a performance that was presented first, understanding the context in which an inter-
method, summarize important items indicating the importance respondents
to the public. Baseline quantitative measures of
ents, had the response categories “believe all of
vention occurs and, second, hearing the diverse
placed on: (a) “believing in God;” (b) “obeying God, it” (4), “believe most of it but not all” (3), “believe
well-being were taken in the first year of the pro- viewpoints of the stakeholders (Greene, 2000).
points, discuss methodological doing what he wishes;” and (c) “relating to God in
gram, with the same measures administered again
some parts but disbelieve others” (2), and “don’t
For example, Pawson and Tilley (1997) advocate
a personal way.” A second dimension, relating to believe any of it” (1).
controversies, and offer practical
advice. dimension of the concept (e.g., respecting confi- simply tags to allow the material to be stored quanti-
In quantitative studies, data are collected on to difficulties in holding down a job and thus
dentiality) may not necessarily score high on other tatively. Then it is necessary to go through the infor-
two or more variables, which are then examined poverty? Or is it a bit of both? To take another
dimensions (e.g., fiscal honesty or continuing edu- mation again to look for incidences of the theme or
to detect patterns of association. This approach example, a study of 1000 men found that those
cation), so that for each respondent one can have a category, and to record the appropriate numbers on
sometimes makes it difficult to show cause and who had two or more orgasms a week exhibited
multidimensional “profile.” a computer spreadsheet. This approach is sometimes
effect because the independent and dependent a 50 per cent lower mortality risk compared with
However, in much quantitative research, there called post-coding. Post-coding can be an unreliable
variables are measured simultaneously, making men who had on average less than one orgasm
is a tendency to rely on a single indicator for each procedure because there may be inconsistencies in
any demonstration of time order (showing that per week. It may be tempting to conclude that
Special Features of the Book xiii
Checklist
Checklist of issues to consider for a structured interview schedule or questionnaire
☐☐ Is a clear and comprehensive introduction to ☐☐ Are questions relating to the research topic
the research provided for respondents? asked near the beginning of the interview or
☐☐ Are there any questions used by other re- questionnaire?
searchers that would be useful? ☐☐ Have the following been avoided?
☐☐ Will the questions provide answers to all the • ambiguous terms in questions or response
research questions? choices
☐☐ Are there any questions not strictly relevant • long questions
Checklists. Most chapters also to the research questions that could be • double-barrelled questions
• very general questions
include checklists of points to dropped?
• leading questions
☐☐ Has the questionnaire been pre-tested with
keep in mind when engaging in some appropriate respondents?
• questions that include negatives
• questions using technical terms
a particular activity, whether ☐☐ If a structured interview schedule is used, are
the instructions clear? For example, with filter ☐☐ Do respondents have the knowledge required
devising a structured interview questions, is it clear which question(s) should to answer the questions?
be omitted? ☐☐ Is there an appropriate match between ques-
schedule, conducting a focus tions and response choices?
☐☐ Are instructions about how to record re-
group, or doing a literature sponses clear (for example, whether to tick ☐☐ Are the response choices properly balanced?
or circle; whether more than one response is ☐☐ Do any of the questions depend too much on
review. Checklists reinforce allowable)? respondents’ memories?
key points and remind students ☐☐ Has the number of open questions been
If using a Likert scale approach:
limited?
of things they need to consider ☐☐ Can respondents indicate levels of intensity ☐☐ Are some items that have to be
in their replies, or are they forced into “yes or
when doing their own research. no” answers?
reverse-scored included, in order to identify
response sets?
☐☐ Have questions and their answers been kept ☐☐ Is there evidence that the items really do
on the same page? relate to the same underlying cluster of atti-
☐☐ Have socio-demographic questions been tudes, so that the items can be aggregated?
left until near the end of the interview or ☐☐ Are the response choices exhaustive and not
questionnaire? overlapping?
However, there are several possible departures from as well, and to administer a structured interview to a
this pattern. focus group would be very unusual. In most survey
research a specific individual is the object of ques-
xiv Special Features of the Book
A Online focus groups are appropriate for research Qualitative interviewing alone versus
involving sensitive issues. Identify three issues ethnography
that, because of their sensitive nature, would R Outline the advantages and disadvantages of
be better researched with online rather than in- qualitative interviewing (without immersion in a
person focus groups, then explain why the online social setting) compared to ethnography.
technique would be more appropriate. A Is one method more in tune with the research
needs of qualitative researchers than the other?
Feminist research and interviewing in Explain, using the topic of intimate partner vio-
qualitative research lence to illustrate your answer.
R Why are qualitative interviews so prominent in
feminist research?
A Explain why focus groups may be superior to other
methods of inquiry for giving a voice to highly mar-
ginalized women.
the classroom.
general discussion of:
they developed;
b. whether the moderator’s control of the discus- a. difficulties in getting the interview to flow
sion was excessive, about right, or too weak, smoothly, and how those difficulties may be
and what the consequences of that were; resolved;
c. to explain the advantages of the focus group b. illustrations of how topics that were not on the
method compared to one-on-one qualitative interview guide made their way into the inter-
interviews for researching the topic chosen; view anyway, and how that helped or hindered
d. to explain the disadvantages of the focus group the investigation of the topic; and
method compared to one-on-one qualitative c. the sorts of topics that could be usefully re-
interviews for researching the topic chosen; and searched using semi-structured interviews,
e. to explain how the focus group method may be and which topics would be better pursued
better than structured interviews for exploring using ethnography or structured interviews.
Key Points
• It’s important not to exaggerate the differences methods can be used to analyze the rhetoric of
between quantitative and qualitative research. quantitative researchers.
• Connections between358 epistemology and ontology
PART IV Transcending • Some qualitative
the Quantitative/Qualitative Divideresearchers employ
and Some Practical quantifica-
Advice Key points. Each chapter concludes
on the one hand, and research methods on the tion in their work.
other, are not fixed or absolute.
• Qualitative research can exhibit featuresHypothetical
approached.
• Although the practice of multi-strategy research
normally results hasandincreased, not all writers
conclusions support
café. it. is shown the picture or clip, and each
The class
with a summary of its most
associated with a natural science model. • The view that there are epistemological and onto-
are given for points “f” and “g.”
• Quantitative research can incorporate an interpre-
student then writes up a description of the physi-
logical impediments to the combination of quan-
Once that is done, students leave their groups cal setting that is depicted (maximum 250 words).
significant points.
tivist stance. titative and qualitative research is a barrier to
and the class is reconvened. Each person in the Three volunteers then read their descriptions to the
• The artificial/natural contrast used to distinguish multi-strategy research.
class is given 15–20 minutes to write a brief Intro- rest of the class. Next, the instructor facilitates a
quantitative and qualitative research is often • There are several different ways of combining
exaggerated. duction to their group’s article. The Introduction
quantitative and qualitative discussion based
research; someon the
canfollowing questions:
should not
• A quantitative research approach canexceed
be used250to words and should be
be planned writ-
in advance, others cannot.
tenqualitative
in such a manner a. How are the three descriptions similar? How
analyze qualitative data, and researchthat it attracts the reader’s
attention, gives a clear indication of the article’s are they different?
b. Why are the three descriptions not identical?
Questions for Review ( ) and Creative Application ( )
focus, and R highlights the significance of the find- A
How would you account for the differences?
ings. When completed, the instructor asks for three
The natural science model and qualitative
volunteers who are willing Research
to give a brief methods
verbal and epistemological
c. What is “interpretiveand omnipotence”? Do
research synopsis of their group’s study ontological
and have considerations
their any of the descriptions assume interpretive
R Under what circumstancesIntroduction
can some qualitative
read by the rest How
R of theclosely tied are research
class. Volun- omnipotence?
methods to epis-
research use a natural science
teersmodel?
with good keyboard skillstemological
are then asked and ontological
to d. positions?
Which of the descriptions is most realistic?
Explain.
A A qualitative researcher finds that many of the A You decide to do a secondary e. Can analysis
one everof quanti-
determine whether a particu-
transcribe the three Introductions into electronic
homeless people she encounters in her fieldwork tative data taken from the General lar Social Survey,
description is more realistic than another?
format so they can be shown on the classroom
have addiction issues. How might she use that find- Victimization Study. You have Does nothe
preconceived
term “realistic” even have a definite
screen. The first of the three volunteers then gives
ing to launch a quantitative study? theoretical position when youmeaning? begin, but decide to
a synopsis of their group’s study, and their Introduc-
see if there is any association between gender and
Quantitative research and tion is shown on the screen. Thefear
interpretivism instructor
of crime. fa- does f.your
thenHow Would the person
approach deviate
or persons depicted in the
cilitates a discussion of the Introduction. The other picture or clip describe the setting in a way
R Under what circumstances can some quantitative from a strictly positivist orientation to research?
two Introductions are presented in the same way. that differs from how the three volunteers de-
research exhibit characteristics of interpretivism? scribed it? How so? Why would the person or
A A quantitative researcher finds Inthat
an alternative
25 per cent version
of Problems withthat
of this exercise the quantitative/qualitative
persons depicted have a different description?
the people aged 18–25 in awouldnationalbe sample
appropriate contrast
have for small classes, the in-
g. How could you learn about how the people in
no intention of voting in thestructor collects
next federal all the Introductions
election, R Outline some thatofthethe ways in which the quantita-
class produced and provides written comments on the picture or clip view the setting?
while the figure for people aged 65 and over is tive/qualitative contrast is not as hard and fast as
only 10 per cent. Explain this
eachdifference, making
one, returning is often
them in a later supposed.
class. h. If multiple interpretations of a physical setting
reference to how people at different ages may per- A Explain how grounded theory are possible,could
methods wouldbeinterpretations of people’s
ceive the political process differently.
2. Prior If you
to class, were
the instructor used to
selects develop
a picture or aa theory of actions and experiences
the relationship be- in the setting be even
to write up your answer in a brief
research
videoreport,
clip ofwould tween body
a person or persons shaming and the more
in a physical use ofnumerous and variegated? Explain.
social media,
it be appropriate to describe it asthat
setting qualitative
would beinappropriate
then fordescribe how this sort
qualitative i. What are the can
of research implications
be of the responses to
nature? Explain. research—for example, a market, thought
a sportsof facility,
as a forma of theorythe testing.
previous question for qualitative research?
bry29440_ch14_323-345.indd
2013_
11/20/18 11:57 AM
each chapter.
Glossary
Terms in italic type are defined elsewhere in the Glossary. are broken down into component parts, which are then
Online resources
Textbooks today do not stand on their own: they are only the central elements in a complete learning and
teaching package. Social Research Methods is no exception. This fifth Canadian edition is supported by an
outstanding array of ancillary materials for both students and instructors.
Additional Materials
In addition to the above materials, OUP Canada offers these resources free to everyone using the textbook:
www.oupcanada.com/SocialResearch5e
Special Features of the Book xvii
For Students
Student Study Guide
A comprehensive online study guide provides detailed chapter summaries, learning objectives, lists of key
terms and concepts, self-assessment quizzes, and links to useful media resources.
For Instructors
Online Instructor’s Manual
This revised online resource includes comprehensive outlines of the text’s chapters, additional assignments,
classroom activities designed to encourage student engagement, and teaching aids that will enhance the
learning experience.
PowerPoint Slides
Hundreds of slides for classroom presentation—newly updated and enhanced for this edition—summarize key
points from each chapter and can be edited to suit individual instructors’ needs.
• Wider philosophical and methodological exposure to them will reinforce the textual
issues are discussed at some length in Chapter 1. material and may be useful for later work.
Instructors who do not wish to use this contextual • The quantitative/qualitative distinction is
material can largely ignore the chapter, except used in two ways: to organize the discussion of
for the section on formulating a research ques- research methods and data analysis, and to intro-
tion. Those who do want to emphasize issues of duce some wider philosophical issues that have
context should also see Chapter 14. a bearing on social research. Chapter 1 reviews
• Practical issues involved in doing quantitative the main areas of difference between quantita-
research are the subject of Part II. Chapter 2 is tive and qualitative research, while Chapter 14
a useful introduction to this topic because it explores ways of integrating the two. If time is a
maps out the main research designs used in both concern, the latter chapter can be skimmed.
quantitative and qualitative research. • Writing up research is as much a part of the
• Practical issues involved in doing qualitative research process as data collection and analysis.
research are the subject of Part III. Again, Chapter 15 discusses a variety of issues related
Chapter 2 outlines the most common designs. to writing and should be drawn to students’
• Data analysis is covered in Chapters 8 (quanti- attention even if it is not discussed in class.
tative) and 13 (qualitative), and a guide to the • Specific advice on doing a research project.
use of computer software for these purposes can As we have already noted, the whole book is
be found in the Appendix. There is an additional relevant to student projects, but Chapter 16
guide to using Excel for data analysis on the addresses this subject directly. If time is a factor,
accompanying Dashboard. Even if the module the Appendix to Part I briefly overviews the
is taught without actual computer applications, stages of research for students.
Acknowledgments
Normally, writing the Acknowledgments for a book I’ll miss Alan Bryman, but at the same time I’m
is a rather pleasant task. One can look back on all the sustained by knowing that working with him has
hard work that was done by a network of people, and changed my outlook on the world for the better.
reflect on one’s good fortune that the immense task And of course he touched many others as well. Alan
of producing a book was made possible by the kind- was given a very fitting tribute in the International
ness of others. I feel that gratitude now in writing Journal of Social Research Methodology (2018, 21[3],
this, but it is intermingled with a sense of sadness 267–274), which attested to the tremendous impact
because while this edition of the book was being pre- his work had on a broad range of people, in many
pared, Alan Bryman passed away. fields. As a further tribute, I’d like to dedicate this
Alan died on 20 July 2017. I was in Europe at the edition of the book to his memory.
time, biding my time between conferences. Even Others who contributed to this new edition in-
though we had never met in person, his death was still clude developmental editor Amy Gordon, who got
quite touching to me. We had exchanged emails and the ball rolling and always took an optimistic view
other electronic materials over the years, and he had of challenging circumstances. Her patience, intelli-
given me a standing invitation to visit him in England gence, organizational skills, and attention to detail
whenever I had the opportunity. In the communica- were invaluable. Leslie Saffrey did an amazing job as
tions we shared Alan was always friendly and collegial, copy editor. Her diligence and professionalism shine
and never hesitated to devote his considerable talents through on every page. Jayne Baker, University of
to whatever I asked of him. It is with considerable Toronto Mississauga; Meridith Burles, U niversity
regret that I never made my way to England to see him. of Saskatchewan; Alicia Horton, University of the
Still, as co-authors of this book a special kind of Fraser Valley; Neda Maghbouleh, University of
bond developed between us. Alan greatly expanded my Toronto; Heather Mair, University of Waterloo; Oral
knowledge of research methodology and of the social Robinson, University of British Columbia; and sev-
sciences in general. One thing about him that I greatly eral anonymous reviewers provided detailed feed-
admired was how he dealt with the tensions that some- back on this edition, and in so doing vastly improved
time arise between people taking different epistemo- the quality of the book and saved me from serious
logical, ontological, and methodological approaches errors and omissions. Thank you. To be sure, the
to social research. For Alan there was no blind loyalty shortcomings that remain are mine alone.
to a particular way of doing research; he simply took My family was a constant source of comfort and
the position that what really mattered was finding the inspiration as I worked on this manuscript, and they
best method or methods to address a particular re- continue to nurture me. Much is owed to my wife
search question. And he was a master at showing how Jennifer, to my children Ted and Angelica, and to
seemingly incompatible approaches actually had many Brooke.
things in common and could complement each other. Edward Bell
Preface
There are some answers to the “Why are you
Why Study Research Methods? here?” question that I seldom get from students.
At this point in your life you probably haven’t devoted Consider the following. The acquisition of know-
much thought to social research methods. In all like- ledge through research always goes beyond coming
lihood you are reading this book in conjunction with up with new ideas or getting more information or
the first methods course you have ever taken. That developing novel ways of looking at society, how-
means you are about to experience something new. ever important those things may be. To produce or
Since you are going to expend considerable time and acquire knowledge is a political act. It is tied in with
effort doing something you’ve never done before, it the exercise of power. Social research always has pol-
would make sense to pause for a moment and ask itical ramifications, because it always implies, subtly
yourself the following existential questions: Why am or not so subtly, that some ways of organizing society
I here? Why should I read this book? Why should I are better than others. Some researchers are happy
study social research methods? What’s the point of it to leave their political involvement at that level: an
all? Pondering those questions will make what is to implication in their work that certain social struc-
come much more meaningful for you. tures or practices should change or be preserved.
So why are you here? When I ask my students Others go so far as to engage in advocacy, taking part
that question (usually on the first day of classes) in public campaigns to persuade governments and
they often tell me that a solid background in social the public at large that some sort of action should be
research methods is indispensable if they are to de- taken. Whichever position the researcher takes, the
velop a sophisticated understanding of the topics politics of research cannot be avoided. What varies is
they are passionate about. how loudly the trumpet is sounded, which is a matter
What are you passionate about? Does it matter to for the individual researcher to decide. Debates about
you that literally billions of human beings struggle to the wisdom of social researchers becoming advocates
survive on less than $2.50 per day, while others are are part of the storied history of the social sciences,
so wealthy that they can’t relate to ordinary people? some of which will be told in this book.
Do you have an interest in gender equity, environ- We often hear that “knowledge is power,” and in
mental sustainability, crime, single motherhood, many ways that is true. Dictators know this best;
ethnic tension, labour relations, racism, the living they all try to limit access to knowledge in order to
conditions of Indigenous people, changing notions preserve their control over others. In fact, the power-
of the family, sexual mores? In order to come up with ful in any society may try to inhibit free inquiry
informed, thoughtful analyses of these and other and the flow of ideas. But the “knowledge is power”
social issues, and to be capable of evaluating the maxim leaves several questions unanswered. What
claims made by others on these topics, it is crucial is knowledge? How can it be acquired? How are we
to be familiar with the various research methodol- to tell the difference between a sound idea and one
ogies used in the social sciences to investigate these that should be ignored? How can we gather informa-
subjects. Just as learning how clothes are made can tion that will help us understand our subject matter,
help you tell the difference between a good pair of and what sorts of information should we seek? How
jeans and a shoddy pair, understanding where social should that information be analyzed and evaluated?
knowledge comes from will help you distinguish be- Are there some things that we will never fully under-
tween valid claims and fatuous ones. stand, regardless of how hard we try?
xx Preface
This book explores a variety of answers to those takes in ideas and evaluates them, but also as some-
questions, which will make it somewhat different one who projects ideas onto the world. Start think-
from your other readings and textbooks. In most ing of yourself as a researcher: someone who has the
courses in the social sciences, students spend their potential to examine the human condition and con-
time absorbing, and sometimes challenging, claims tribute to our understanding of it. That’s far more
made about the social world. For example, you may interesting and impactful than just reading other
learn that men are more likely to commit violent people’s work, although reading is an essential part
crimes than women. You may come to understand of the process.
what it feels like to be homeless. Or perhaps you’ll When can you start being a researcher? At vari-
encounter a theory explaining why some countries ous points in your university career you will prob-
are rich and powerful while others are not. That ably be called on to do some small research projects.
sort of endeavour—learning about and reflecting The knowledge you will gain by learning about social
on claims to knowledge—is clearly worthwhile, but research methods can put you at an advantage with
it will not be our central concern. Instead, we will that sort of assignment, and it could even be an
focus on how those types of claims are generated entrée to more sophisticated work if you decide to go
and how they gain acceptance as knowledge. In on and develop your research capabilities in gradu-
other words, instead of exploring the body of facts, ate school or in some other way.
concepts, theories, and interpretations that social Although it is important to acknowledge that
scientists use to make sense of the world, we’ll look social research and its attendant methodologies
at how all those things are acquired or created. One have political ramifications, we shouldn’t lose sight
of the most profound questions that can be asked of of the fact that it is the height of folly to pursue one’s
someone making a claim to knowledge is, “How do political goals with bad social science. To reduce
you know that?” That simple question is at the heart suffering and promote human flourishing we need
of this book. And like so many simple questions, it rigorous social research, which is dependent on
has no simple answer. sound research methods. Learning how to take
To produce knowledge, we need a m ethodology— good field notes, getting instruction on how to con-
a way to get it. And if you accept the premise that struct effective survey questions, using appropriate
knowledge is power, then learning about research sampling procedures, and a host of other methodo-
methods means learning about how to become logical issues are not trivial matters. In fact, doing
powerful, or at least more powerful. You could even those sorts of things properly are prerequisites to
say that “methodology is power,” and you wouldn’t acquiring the knowledge needed to achieve one’s
be far off the mark. broader social goals. Without good methodolo-
How could methodology be power? For one thing, gies we have very little to offer that is not available
if you know how knowledge is generated, you won’t elsewhere.
be fooled as easily as someone who doesn’t know. If you want to learn about how knowledge is cre-
Another way to put this is that learning about re- ated in the social sciences, or if you aspire to make an
search methodologies is an excellent way to enhance impact on the world by doing social research your-
your critical thinking skills—skills that are essential self, reading this book is a good way to start. I hope
in any situation where power is exercised, which is it will at least pique your interest in how social re-
everywhere. searchers do what they do, and what the implications
Another way in which methodology can be power of social research are. To paraphrase Gore Vidal: the
is that it can help to make you a producer of ideas. more one learns, the more interesting consciousness
Start thinking of yourself not only as someone who becomes.
The first two chapters of the text will acquaint
you with a number of fundamental concepts
that feature prominently throughout the
book, in particular the notion of a general re-
search orientation and the idea of a research
design. In Chapter 1, two general orientations
are identified—the quantitative and qualita-
tive approaches—along with a variety of con-
siderations that affect the practice of social
research. Chapter 2 presents various kinds of
research design and identifies some criteria
used to evaluate social research. Together
these chapters provide the basic conceptual
building blocks that we will return to later in
the book. Chapter 3 provides a discussion of
research ethics, which are a primary concern
for every kind of social research and affect
every stage of the inquiry. Part I concludes
with an appendix that provides an overview
of the research process.
PART I
Fundamental Issues in Social Research
© kendo_OK/Shutterstock.com
1
General Research Orientations
Chapter Overview
The aim of this chapter is to examine the fundamental assumptions upon which social research is based.
An important distinction commonly drawn by practitioners of social research—between the quantitative
and qualitative approaches—is explored in relation to those considerations. We will consider:
• the relationship between theory and research—in particular, whether theories and the hypotheses
derived from them are tested by gathering data (a deductive approach) or whether data gathering
is used as a means to create theory (an inductive approach);
• epistemological issues, such as whether a natural science model like the one used in chemistry
or biology is suitable for the study of the social world;
• ontological issues, such as whether the social world should be regarded as a reality external to
individuals over which they have little or no control, or as something that social actors may fashion
into their personal realities;
• how values and practical issues impinge on the research process; and
• how these issues relate to both quantitative and qualitative research; a preliminary discussion,
followed up in Chapter 14, suggests that although the quantitative and qualitative orientations are
different, they complement each other.
Soon-Yi wants to find out why Indigenous people in The list of topics she could collect information on
Canada are more likely to live in poverty than other seems endless.
Canadians, but doesn’t know where to begin. Should Maybe rather than beginning her study by accumu-
she start by examining the history of colonialism and lating data, it would it be better to start out with some
conflict between Indigenous peoples and settler- hunches and then gather information to see whether
colonizers, such as disputes over land claims and treat- they are supported by evidence. For example, perhaps
ies, residential schools, or anti-Indigenous prejudice? the discrepancy in economic conditions is a manifesta-
Or how about gathering aggregate data on present tion of a centuries-old system of international domin-
conditions like residence patterns, e conomic activ- ance and exploitation. Similarly, it could have arisen
ities, the age structure, or educational trajectories? through a clash of civilizations and cultures. Then there
▲ LeonWang/Shutterstock
1 | General Research Orientations 3
is the question of how to gather the information to many social researchers put it, is reality “out there,” in-
address those theories. Should she consult mainly his- dependent of the person who perceives it, or is it so-
torical sources? How about living in an Indigenous com- cially constructed?
munity for a while and recording her interactions and Then there are the broader questions: What, if any-
observations? Or would it be better to stay where she is thing, can be done about the issue of poverty among
and select a few individuals for interviews in which they Indigenous Canadians? Can society be changed so that
could talk about their life stories? Or maybe she should Indigenous poverty is eliminated, or at least reduced?
design a detailed questionnaire on prejudice and send Or are the social forces that affect our lives so strong
it out to large random samples of Indigenous and non- that we have no choice but to conform to them, like
Indigenous people? leaves in a windstorm?
Soon-Yi wonders about the ethical implications of A great many social thinkers and researchers have
each of these approaches. She also puzzles over what confronted the same issues that Soon-Yi is pondering.
she should do if it turns out that Indigenous people In fact there are a number of rich intellectual traditions
take positions on her research topic that differ from that can be drawn on to make sense of these matters
her own. Should she try to determine whose perspec- and to assist people in making the kinds of methodo-
tive better reflects reality, and if so, how would she do logical and practical decisions that all researchers have
that? Can those sorts of issues ever be resolved? Are to make. In this chapter you will be introduced to that
there “right” and “wrong” answers on such matters? As body of ideas and the debates surrounding them.
hearing about the personal histories of her family 15–25 decreases, the crime rate also decreases.”
and friends. Relational statements come in two forms:
Regardless of the motivation for doing research, a. deterministic, which means the two vari-
the data gathered are usually viewed in relation to ables always go together in a particular way;
theories of some kind. That’s because theories are an for example, “as the number of people aged
attempt to “make sense of it all,” to find order and 15–25 decreases by 1 per cent, the crime
meaning in a seemingly infinite mass of informa- rate also shrinks by 1 per cent.” If research
tion. How is that done? uncovers an instance in which the variables
are not related in this way, the relational
statement must be modified.
Theory and research b. probabilistic, which means the two vari-
The connection between theory and research is ables go together with some degree of regu-
not straightforward. There are several issues at larity, but the relationship is not inevitable;
stake here, but two stand out: first, the form of the for example, “regions of the country where
theory; and second, the relationship between data the number of people aged 15–25 is decreas-
and theory. ing are more likely to experience a decline
in the crime rate than regions in which the
Degree of abstraction number of people in that age category is
The term “theory” is used in a variety of ways, but its increasing.” Here, finding a case that does
most common meaning is an explanation of observed not fit the pattern does not mean that the
regularities or patterns. For example, one could try to theory must be modified; this could simply
come up with a theory to explain why schizophre- be one of the times when the variables are
nia is more common in the working class than in not related in the usual way.
the middle class, or why more men than women are
alcoholics. Theories are composed of interrelated There are different types of theories. One distinc-
and usually verifiable statements or propositions. tion that is sometimes made is between theories of
The statements and propositions come in varying the middle range (Merton, 1967) and grand theories.
forms, and different types may be combined in the The former are more limited in scope, and can be
same theory. Here are three common components of tested directly by gathering empirical evidence. For
a theory: instance, Durkheim’s (1897/1952) theory of suicide,
which maintains that suicide is a function of the level
1. Definitions specify what the key terms in the of social integration, is a theory of the middle range.
theory mean; for example, “Attachments are One way to test it would be to compare suicide rates
stable bonds between people who are fond of for married people with those for single, divorced,
each other.” or widowed individuals. Merton’s (1938) anomie
2. Descriptions outline the characteristics of the theory, which suggests that crime is more common
phenomena of interest; for example, “Attach- when a society instills a desire for wealth in everyone
ments to parents and other family members but provides insufficient means for all to achieve it,
tend to decline in adolescence. Later, often is another theory of the middle range. Such theories
after age 25, new attachments develop as a represent attempts to understand and explain a lim-
person marries or lives with a romantic part- ited aspect of social life.
ner, becomes a parent, or gets a steady job.” Grand theories, by contrast, are general and
3. Relational statements connect two or more abstract. They include theories such as structural-
variables, so that knowing the value of one functionalism, symbolic interactionism, critical
variable conveys information about the other; theory, post-structuralism, feminism, and so on.
for example, “As the proportion of people aged Grand theories generally offer few direct indications
1 | General Research Orientations 5
of how to collect evidence to test them, but they pro- how an abstract theory like Giddens’s structuration
vide ways of looking at the world that can be the theory (1984) can be applied to a specific situa
inspiration for a wide variety of research programs. tion and yield some important insights.
For example, standpoint theory was developed by Although theory plays a crucial role in the social
Dorothy Smith (2004; 2005) and others from a gen- sciences, not all studies make reference to it. For ex-
eral feminist perspective. This theory maintains that ample, some qualitative writers focus on providing
the way we view the world and make our way in it is a rich description of the experiences of a group of
largely determined by our placement in various hier- people without trying to come up with a compre-
archies of status and power. One offshoot of stand- hensive theory that would explain those experiences.
point theory has been an increased willingness to Nonetheless, as Box 1.2 shows, social scientists are
examine the views and perceptions of poor or mar- often under pressure from their peers to relate their
ginalized groups in society. Similarly, Box 1.1 shows work to theories of some kind.
As Box 1.2 indicates, some social scientists will
reject research that has no direct connection to
theory in either the grand or the middle-range
sense of the term. However, non-theoretical work
BOX 1.1 Grand theory and social can provide insights that are useful or revealing in
research their own right. McKeganey and Barnard’s (1996)
research on British sex workers and their clients is
Giddens’s (1984) structuration theory attempts a case in point. The authors related their research
to bridge the gulf between notions of structure findings to investigations of people in the sex
and agency in social life. This theoretical issue industry in several other countries, and what they
is explored in empirical research by Dinovitzer describe offers good illustrations of ideas that form
et al. (2003) on the educational attainment of an important part of the sociologist’s conceptual
immigrant youth. The specific focus of their re- toolkit. Although it is not possible to tell whether
search was suburban Toronto immigrants in the the authors had the concept in mind when they col-
years 1976 to 1995. The data were quantitative, lected their data, their book offers real-life examples
generated through structured interviews. The of Goffman’s (1963) notion of “stigma” and the way
goal of the researchers was to tease out the stigmatized individuals, in this case sex workers and
relative influence on educational attainment their clients, manage a spoiled identity. Their analy-
of structural variables (such as class, gender, sis also sheds light on Hochschild’s (1983) concept
and the youths’ family background) and indi- of “emotional labour,” a term she coined to refer to
vidual variables over which the young people what flight attendants do when they feign friendli-
had more control (such as studying and cutting ness in order to deal with difficult passengers. Sim-
classes). ilarly, several other non-theoretical studies provide
The authors found that educational achieve- data that could eventually be used to evaluate or
ment is affected by two structural variables devise a theory.
(gender and father’s occupation) and one indi- Our discussion of what theory is and its im-
vidual factor (intellectual investment), and that portance invites consideration of another question:
bilingual ESL students do better than young im- What is the relationship between theory and re-
migrants for whom English is a first language. search? Up to this point we have focused primarily
They are not brighter, nor do they work harder, on how theory can guide research, in particular on
but they have greater parental supervision, how the collection and analysis of data can be used
and—perhaps partly as a consequence—they to test theories. But this notion of research as essen-
plan better. tially “theory testing” does not provide a complete
picture of what social scientists do. Theory may also
6 Part I Fundamental Issues in Social Research
sunflowerey/123RF
it art. What social theories might you invoke to ex-
plain what this person is doing? How might the young
person explain it? Do you think his explanation would
have anything in common with your social theories?
follow upon or arise from the collection and analysis Note that the researcher starts by coming up with
of data. Here we begin to see two different ways to go a theory that seeks to explain a particular phenom-
about acquiring knowledge: the deductive and induc- enon, and then deduces specific hypotheses from it
tive approaches. that are tested with empirical data (see Box 1.3 for
a concrete example) and then are either confirmed
or rejected. In the case of the research described in
Deductive and inductive Box 1.3, which examined several theories simul-
approaches taneously, some hypotheses were supported while
The deductive method is the most common ap- others were not. If the data gathered do not support
proach to social research. The sequence of steps the researchers’ hypotheses, the theory may have to
taken in deductive research is depicted in Figure 1.1. be revised or rejected.
1 | General Research Orientations 7
Guenter Guni/iStockphoto
has occurred in which fewer and fewer people
embrace religious beliefs and practices, a pro-
cess that became especially marked after the
middle of the twentieth century. To be sure,
religion has not disappeared in Western coun-
tries, including Canada. In fact about 26 per cent Veiled women in Iran, a society that has experienced
of Canadians attend religious services once a far less secularization than many Western countries.
month or more frequently (Statistics Canada, Do you think that some of the theories examined in
2013). Nonetheless, several indicators suggest Hay’s (2014) research can help to explain the high
that the proportion of Canadians who are firmly levels of religiosity found in Iran? For example, could
the acceptance of religious beliefs and practices
religious is considerably lower than it was sev-
there be related to cultural factors that favour so-
eral decades ago. What can account for this pro-
cietal cohesion over personal autonomy? Could low
found social change?
levels of religious pluralism be a factor?
Hay (2014) has taken a deductive approach
to this issue. As in all deductive research, his
goal was “theory testing.” He started by propos- indicators of factors that were relevant to each
ing some theories, derived testable hypotheses of the theories.
from them, and then set out to determine which, There was limited support for deprivation-
if any, of the hypotheses were supported by the compensation theory, but there was strong sup-
data. He reviewed seven well-known theories of port for three other theoretical positions. One
secularization, and deduced specific hypotheses theory that gained support involves the idea that
from each one. For instance, the deprivation- a growing acceptance of personal autonomy in
compensation theory holds that the existential spiritual and moral matters has led to increased
security that affluent people in developed soci- levels of secularization. A second position that
eties enjoy through health care and education, was consistent with the results holds that the re-
long life expectancy, and the absence of military ligious pluralism found in wealthy countries cre-
conflict shields many of us from the traumas that ates a situation in which it is very difficult for a
draw people toward religion. The hypothesis he particular faith group to make the claim that it is
derived from this theory was that people who the “one true religion,” which casts doubt on all re-
have experienced real tragedy in their lives are ligious doctrines and so increases levels of secu-
more likely to be religious than people who have larization. A third theory that was substantiated
no direct exposure to such things. Hay then ana- by the data maintains that increasing acceptance
lyzed contemporary survey data that included of human evolution and the belief that observable
measures of religious beliefs and practices, phenomena are purely material or physical in
as well as a number of variables that provided nature has reduced levels of religiosity in Canada.
1 | General Research Orientations 9
number of people in same-sex marriages in this knowledge. A fundamental debate in this context
country increased from 42,030 in 2011 to 48,740 in is whether the social sciences should follow the
2016 (Statistics Canada, 2017), which a researcher same principles and procedures as the natural sci-
could use in the context of a discussion of the ences. Three broad positions on these matters have
changing forms of the family in Canada. emerged: the positivist, interpretivist, and critical
The next section examines some epistemological approaches to social science.
issues that affect the conduct of social research.
Positivism
One epistemological position that affirms the im-
Epistemological portance of following the natural sciences is pos-
considerations itivism. Although definitions of the term vary,
Those who do social research base their work on a positivism is generally taken to entail the following:
number of epistemological assumptions—notions
of what can be known and how knowledge can be 1. Only phenomena confirmed by the senses
acquired. A related epistemological issue pertains to (sight, hearing, etc.) can be accepted as know-
the question of what should be regarded as acceptable ledge: this is the principle of empiricism.
10 Part I Fundamental Issues in Social Research
Ideas must be subjected to the rigours of em- philosophers of science and social science differ
pirical testing before they can be considered quite sharply over how best to characterize scientific
knowledge. practice, and since the early 1960s positivism has ac-
2. A key purpose of theory is to generate hypoth- quired some negative connotations. One reason for
eses that can be tested and thereby allow ex- dissatisfaction was the fact that certain researchers
planations of observed laws and principles to in the positivist tradition ignored some fundamen-
be assessed (deduction). tal differences between human beings and the often
3. Knowledge can also be arrived at through the inanimate or not fully conscious entities studied
gathering of facts that provide the basis for by natural scientists. Unlike subatomic particles or
generalizations or laws (induction). plants, for example, we humans have thoughts, feel-
4. Science must (and presumably can) be “value- ings, and values—perhaps even some capacity for
free.” That is, it must be conducted in such a volition. Those aspects of human behaviour were
way that different researchers, given the same often not addressed in the leading positivistic theor-
data, will always reach the same conclusions, ies of the day. The positivist idea that science can or
no matter how different their values might be. should be value-free was also a source of dissatisfac-
In the past this “value-free” quality was called tion. Critics pointed out that “neutral” social scien-
“objectivity”; today it is more likely to be tists often took moral positions on social issues, at
called intersubjectivity. least implicitly. For example, theories implying that
5. There is a clear distinction between scientific social equilibrium or harmony is normal seemed to
statements, which describe how and why cer- suggest that social change is not needed or not de-
tain social phenomena operate the way they sirable. Critical social science, discussed below, even
do, and normative statements, which outline went so far as to claim that it is the duty of the re-
whether certain acts or social conditions are searcher to help bring about social change in order
morally acceptable. Only scientific statements to create a better world.
have a place in the domain of science; norma-
tive statements belong in the realm of philoso- Interpretivism
phy or religion. This idea is implied by point Interpretivism to some extent grew out of the epis-
1 above, because the truth of moral claims temological critique of positivism, and provides an
cannot be confirmed by the senses. alternative to the sort of social science typically done
by positivists. Interpretive researchers maintain that
It is possible to see in these five points a link with it is the role of social scientists to grasp the subjective
some of the issues already raised about the relation- meanings of people’s actions. They make the point
ship between theory and research. Positivism as- that people act on the basis of the meanings that they
sumes a fairly sharp distinction between theory and attribute to their acts and to the acts of others. Indi-
research and includes elements of both deduction viduals use their own common-sense constructs to
and induction. One role of research is to test theor- interpret the reality of their daily lives, and it is those
ies and gather the information necessary for the de- interpretations that motivate their behaviour.
velopment of scientific laws. Positivism also implies Interpretivists claim that it is the job of the social
that it is possible to collect observations without any scientist to gain access to the “common-sense think-
reference to pre-existing theories, and to develop ing” of the people they study and hence to understand
new theories purely on the basis of those observa- people’s actions and their social world from the point
tions. Finally, theories and propositions not directly of view of the actors. Thus any thoughts constructed
testable through empirical observation are often not by the social scientist to grasp this social reality must
considered to be genuinely scientific. be founded on the common-sense interpretations
A common mistake is to treat positivism as syn- of those they study: people living their daily lives
onymous with science and the scientific. In fact, within their own social world (Schutz, 1962, p. 59).
1 | General Research Orientations 11
In order to gain access to other people’s perspectives, 1994). In research terms, according to Blumer (1962,
interpretivists commonly immerse themselves in the p. 188), “the position of symbolic interaction re-
social environments frequented by the people they quires the student to catch the process of interpreta-
study, or at least conduct lengthy interviews with tion through which [actors] construct their actions.”
them. The insights gained by the researcher com- Taking an interpretative stance can result in
monly follow from the information derived in this surprising findings, or at least findings that appear
way, so the process used tends to be inductive. surprising if the researcher’s position is outside
Many interpretive social scientists argue that the the particular social context being studied. Box 1.4
subject matter of the social sciences—people, groups, provides an interesting example.
and institutions—is fundamentally different from
that of the natural sciences. For them it follows that
the study of the social world requires a different logic
and research procedure, one that reflects what they
see as the distinctiveness of humans as against other BOX 1.4 Interpretivism in practice
living things or inanimate objects. This clash reflects
a division between the positivist emphasis on the ex- Foster (1995) conducted ethnographic re-
planation of human behaviour and society, and the search using participant observation and semi-
interpretivist preference for an empathetic under- structured interviews in a housing estate in East
standing and interpretation of human existence. This London referred to as Riverside, a residential
division, which precedes the emergence of modern complex experiencing a high level of crime ac-
social science, finds expression in Max Weber’s cording to official statistics. However, she found
(1864–1920) notion of Verstehen (which means “em- that residents did not perceive the estate to be
pathetic understanding”). Weber described soci- a high-crime area; nor were they overly anxious
ology as a “science which attempts the interpretive about becoming victims of crime. Those percep-
understanding of social action in order to arrive at tions could be attributed to a number of factors,
a causal explanation of its course and effects” (1947, but a particularly important one was “informal
p. 88). Weber’s definition seems to embrace both ex- social control,” which was used in conjunction
planation and understanding, but the crucial point with more formal methods such as policing.
is that the task of “causal explanation” is undertaken People expected a certain level of crime, but
with reference to the “interpretive understanding felt fairly secure because informal social control
of social action.” This is a different emphasis from a worked to keep the level of crime contained.
more Durkheimian view in which the external forces Informal social control has several aspects.
that affect behaviour may not be perceived by those One is that neighbours often look out for each
involved, or at least may have no meaning for them. other. In the words of one of Foster’s inter-
Symbolic interactionism is an example of a viewees: “If I hear a bang or shouting I go out. If
sociological perspective that falls under the heading there’s aggravation I come in and ring the police.
of interpretivism. The ideas of the founders of sym- I don’t stand for it” (Foster, 1995, p. 575). Another
bolic interactionism—in particular George Herbert aspect of informal social control is that people
Mead (1863–1931), who maintained that the individ- often feel secure because they know each other.
ual’s self-concept emerges through an appreciation of A second respondent said: “I don’t feel nervous
the perceptions of others—have been hotly debated. . . . because people do generally know each
Symbolic interactionists argue that interaction takes other. We keep an eye on each other’s proper-
place in such a way that individuals are continually ties. . . . I feel quite safe because you know your
interpreting the symbolic meaning of their environ- neighbours and you know they’re there . . . they
ment (including the actions of others) and acting look out for you” (Foster, 1995, p. 575).
on the basis of that imputed meaning (cf. Collins,
12 Part I Fundamental Issues in Social Research
Of course, as the example in Box 1.4 suggests, reality) into a Klasse für sich (a class for itself, one
when social scientists adopt an interpretive stance, with an awareness of its exploitation).
they are not simply revealing how members of a Critical social scientists also believe that research
social group interpret the world around them. The should be action oriented. It should involve praxis:
social scientist almost certainly aims to place those putting one’s theoretical and academic positions into
interpretations into a social scientific framework. practice. The idea of praxis is contained in Marx’s
Thus there is a double interpretation going on: the famous dictum that “philosophers have only inter-
researcher is interpreting others’ interpretations. preted the world in various ways; the point, how-
Indeed, there is even a third level of interpretation, ever, is to change it [emphasis in original]” (Marx,
because the researcher’s interpretations have to be 1845/1998, p. 574). This exhortation applies not only
further interpreted in terms of the concepts, theor- to positivism, but also to interpretive perspectives on
ies, and literature of the social sciences. Thus, in social life.
Box 1.4 the idea that Riverside was not perceived as A critical perspective has been adopted by a wide
a high-crime area by residents was Foster’s interpret- variety of scholars, including those engaged in fem-
ation of her subjects’ interpretations. She then had inist, post-colonial, anti-poverty, anti-racism, and
the additional job of placing her findings into a social queer studies. Another example is participatory
scientific framework, which she accomplished by re- action research, which is discussed in Box 1.5 and
lating them to existing concepts and discussions in examined again in later chapters. Institutional eth-
criminology: concepts such as informal social con- nography and critical discourse analysis, examined
trol, neighbourhood watch schemes, and the role of in Chapters 10 and 12, respectively, are further ex-
housing style as a possible cause of criminal activity. amples of critical social science.
The aim of this section has been to outline how
Critical approaches to social science epistemological considerations are related to re-
Like interpretivism, critical social science developed search practice. As noted, a key concern is whether
in part as a reaction to positivism. Social scientists a natural science or positivist approach can supply
who adopt a critical approach use a diversity of re- legitimate knowledge of the social world. We’ve seen
search methods, including those used by positiv- that positivists often use the deductive method, al-
ists and interpretivists, and may use a deductive though they concede that it is possible to generate
or inductive approach. But they disagree with the theories using induction. By contrast, we’ve seen that
positivist notion that researchers should take a interpretive researchers typically adopt an inductive
value-neutral stance regarding their subject matter. strategy, and that critical social scientists may use
In fact they maintain that research and knowledge either one. Another important issue involves the
should not be considered as ends in themselves, but purpose of social research, in particular whether
as means to be used to rid the world of suffering and social scientists, in both their academic work and
oppression (Neuman, 2003). their roles as citizens, should be actively engaged in
Marxists, for example, argue that those who own issues of social justice.
the means of production deceive, constrain, and It is important not to overstate the connections
exploit the weak. The masses could be free if social between epistemological issues and research prac-
scientists, by asking embarrassing questions and tice: they represent tendencies only. Particular epis-
making pointed arguments, would uncover exploit- temological principles and research practices do not
ation, expose hypocrisy, and reveal to the general necessarily go hand in hand in a neat, unambiguous
populace the nature and extent of their oppression. manner. Often hybrid approaches are taken, which
This would transform the masses from what Marx combine different positions and approaches. This
called a Klasse an sich (a class in itself, an objective point will be made again on several occasions.
1 | General Research Orientations 13
The origins of participatory action research (some- Participants first reflect on a social problem and
times referred to as “action research,” “emancipa- how it might be studied and rectified, and then use
tory research,” or “participatory research”) are the results of that process as the basis for social
found in the immediate post–Second World War action. Once action has been taken, participants
era in the work of Lewin (1946) and others who discuss the consequences of the action and how
sought to bring social science to bear directly on it might be used to generate more knowledge and
specific social problems. In the early 1970s, PAR inform further action. In this way a social situa-
began to take on a more explicitly activist orien- tion is improved by the people directly affected
tation when people in Latin America, Africa, Asia, by it, who work in collaboration with others. The
and elsewhere became disillusioned with the abil- process has the potential to transform the partici-
ity of conventional social science to improve social pants personally, intellectually, and even spirit-
conditions in their countries. These individuals ex- ually. Debbink and Ornelas (1997), for instance,
panded the methodology of PAR such that local describe how some Alberta dairy farmers worked
people affected by the issue at hand acted as part- together with Mexican campesinos (poor farm-
ners and co-equals with professional social sci- ers) and an activist intellectual to bring donated
entists and government officials. All three groups Alberta cattle to a poor, rural Mexican commun-
worked together to devise research questions, ity. Similarly, Frisby et al. (1997) outline how some
choose research methods, gather and analyze low-income women in British Columbia teamed
data, and undertake a course of action to amelior- up with social scientists and local recreation dir-
ate the problem (Smith, 1997; Frisby et al., 1997). ectors to improve poor women’s access to physical
Of all the approaches to social science, PAR is activity services. In both cases, people directly af-
the one most closely associated with social and fected by a problem and researchers from outside
political activism. In fact working for social jus- the community interacted as equals to produce
tice is built into the methodology of PAR itself. knowledge and address a social issue.
Lantz and Booth (1998) examined media treat- women’s lifestyles cause breast cancer was con-
ment of the apparent rise in the incidence of breast structed as a social fact by popular-magazine
cancer that began in the early 1980s and found writers.
that its depiction as epidemic can be treated as Similarly, Hallgrimsdottir et al. (2006, p. 266)
a social construction. They analyzed a variety of argue that the media “contributes to constructing,
popular magazines and noted that many of the reproducing and deepening the social stigmas
articles drew attention to the lifestyles of modern associated with working in the sex industry.” The
women, such as delaying first births and having authors compared media depictions of sex-trade
careers. The authors also argued that the articles workers in Victoria, British Columbia, from 1980 to
ascribe blame: “Women are portrayed as victims 2004 with accounts provided by street sex workers,
of an insidious disease, but also as victims of their escorts, and others in this sector. Whereas the media
own behaviours, many of which are related to the portrayed such people as wicked and blameworthy
control of their own fertility” (1998, p. 915). in the earlier years, and as exploited, trapped, and
This article concludes that, as a social category, innocent girls more recently, the sex workers them-
the breast cancer epidemic was represented in selves interpreted their work lives very differently.
popular magazines in a particular way—one that Although there was considerable heterogeneity in
blamed the victims and their lifestyles, particu- their experiences and attitudes toward their trade,
larly in the case of young women. Yet in fact fewer many saw sex work as a largely mundane matter
than 20 per cent of cases of breast cancer were of earning a living. Because they are marginalized
in women under the age of 50. Lantz and Booth’s and stigmatized, however, their voices are seldom
study is fairly representative of a constructionist heard; most people have their ideas about the sex
ontology in suggesting that the idea that young trade constructed for them by the media.
Box 1.6 provides two illustrations of the less strin- organization has rules and regulations and adopts
gent form of constructionism. standardized procedures for getting things done.
A second debate revolves around these questions: A division of labour assigns people to different jobs.
Is social reality akin to the physical world as most There is a hierarchy of authority, a mission state-
people see it—largely fixed and “out there,” some- ment, and so on. Objectivists see any organization
thing that individuals and groups have to confront as possessing a reality external to any of the specific
but over which they have little or no control, like individuals who inhabit it; they may leave, but it will
a snowstorm? Or is social reality not necessarily stay. Moreover, the organization represents a social
pre-existing and fixed, but rather created through order in that it exerts pressure on individuals to con-
our actions? A “yes” to the first question indicates form to organizational requirements. People learn
support for a variant of objectivism, while a “yes” to and apply the rules and regulations and follow the
the second affirms a kind of constructionism. In this standardized procedures. They do the jobs to which
chapter we’ll focus mainly on this second debate: they are appointed. If they do not do these things,
whether social reality can be created. they may be reprimanded or even fired. The organ-
Some social scientists suggest that social phe- ization is therefore a constraining force that acts on
nomena confront individuals as external facts and inhibits its members. To a large extent, this is
beyond their reach or influence. For example, an the “classic” way of conceptualizing an organization.
1 | General Research Orientations 15
An alternative ontological position challenges can negotiate their roles to the point where they are
the suggestion that things such as organizations are allowed to operate on patients. But not all writers
external realities confronting social actors who have adopting a constructionist position are similarly pre-
limited power to influence or change them. Strauss pared to acknowledge the existence or importance of
and colleagues (1973), for example, carried out an objective reality. It is precisely this split between
research in a psychiatric hospital and proposed that seeing the social world as an objective reality and
its organization was best conceptualized as one of seeing it as a subjective reality in a continuous state
“negotiated order.” Instead of viewing order as a pre- of flux that Giddens sought to straddle in formulat-
existing characteristic, they argued that it is worked ing his idea of structuration (recall Box 1.1).
at and created to some extent, and that the rules are The constructionist perspective that maintains
far less extensive and less rigorously imposed than that social reality can be negotiated also suggests
might be supposed from an objectivist account of that the concepts people employ to help them
organizations. understand the natural and social world are social
Indeed, Strauss and colleagues saw rules more as products whose meaning is constructed in and
general understandings than as commands (1973, through social interaction. For example, a concept
p. 308). Precisely because relatively little of the ac- such as “masculinity” is treated as a social construc-
tivity of doctors, nurses, and other personnel is spe- tion. This implies that masculinity is not a distinct,
cifically set down or prescribed, the social order of timeless, and universal entity, but something whose
a hospital is an outcome of agreed-upon patterns of meaning is built up through interaction. That mean-
action that are themselves the products of negotia- ing is likely to be ephemeral, in that it will vary over
tions among the different parties involved. For in- time and place. The construction of one’s masculin-
stance, the official rules may say that only a doctor ity may even have to be “recuperated” from time to
can increase medication; however, some nurses are time, such as when Canadian politicians endeavour
routinely given this power, even though this is never to restore their public gender identities (Maiolino,
actually stated in the regulations. The social order is 2015). This sort of social construction can be seen
in a constant state of change because the hospital is particularly well in discourse analysis, examined in
“a place where numerous agreements are continually Chapter 12. As Potter (1996, p. 98) observed: “The
being terminated or forgotten, but also are continu- world . . . is constituted [emphasis in original] in
ally being established, renewed, reviewed, revoked, one way or another as people talk it, write it, and
[and] revised. . . . In any pragmatic sense, this is the argue it”. This sense of constructionism frequently
hospital at the moment [emphasis added]: this is its sparks an interest in how social phenomena are
social order” (Strauss et al., 1973, pp. 316–317). The represented.
authors argued that a preoccupation with the formal
properties of organizations (rules, organizational Relationship to social research
charts, regulations, and roles) makes it hard to rec- Questions of social ontology cannot be divorced from
ognize the degree to which order in organizations issues concerning the conduct of social research.
has to be established in everyday interaction. This Ontological assumptions and commitments affect
informal organization arises because there cannot be both the way research questions are formulated and
rules for every possible contingency, and because the the way the research is carried out. For example, a
existing rules are sometimes problematic. However, researcher who sees organizations as objective social
this is not to say that the formal properties of organ- entities that act on individuals is likely to emphasize
izations have no effect on individual action. their formal properties. Alternatively, a researcher
Although Strauss and colleagues stressed the who is interested in the negotiated, changing nature
active role of individuals in the social construction of organizations is likely to focus on the active in-
of reality, they did not push the argument to an ex- volvement of people in reality construction. In each
treme. For example, they did not claim that nurses case, a different research design is required.
16 Part I Fundamental Issues in Social Research
TABLE 1.1 | F
undamental differences between quantitative and qualitative
research strategies
Quantitative Qualitative
Role of theory in research Mainly deductive; testing of theory Mainly inductive; generation of theory
Epistemological orientation Natural science model; positivism Interpretivism
Ontological orientation Objectivism Constructionism
all points in the process of social research, includ- Researchers are increasingly prepared to forewarn
ing the following: readers of their biases and assumptions and to ex-
plain how these may have influenced their findings.
• choice of research area Since the mid-1970s many researchers have pub-
• formulation of the research question lished “insider” accounts of what doing research is
• choice of method really like, as opposed to the generalities presented
• formulation of the research design and data in social research methods textbooks (like this one).
collection techniques These accounts frequently function as “confessions”
• data collection of personal biases and reveal the pride that research-
• analysis of data ers take in telling readers how open they are in re-
• interpretation of data vealing them.
• conclusions Still another approach is to argue for con-
sciously value-laden research. Some writers on social
There are, therefore, numerous points at which research celebrate what Mies (1993, p. 68) called
bias and the intrusion of values can occur during the a “conscious partiality.” For example, Tastsoglou
course of research. For example, researchers may de- and Miedema (2003) clearly adopted a feminist,
velop affection or sympathy for the people they are anti-racist a pproach in studying immigrant women
studying. It is quite common for researchers who in the Maritimes. A similar perspective allowed Pratt
spend a great deal of time with the people they study and Valverde (2002) to describe a large Canadian
to become so close to their subjects that they find it newspaper as a “notorious tabloid,” “obsessed” with
difficult to disentangle their role as social scientists what it called bogus refugees. It is also exemplified in
from their concern for the participants’ well-being. Hallgrimsdottir et al.’s (2006) condemnation of the
Equally, social scientists may be repelled by the media’s role in stigmatizing sex workers in British
people they are studying. In his research into an Columbia. In fact, some feminist researchers would
African society known as the Ik, a social anthropol- consider it inappropriate (as well as difficult) to do
ogist was appalled by what he saw as their cruelty research on women in an objective, value-neutral
toward each other (Turnbull, 1973). Although he way because that would be incompatible with their
was able to identify the desperate social and political values. Instead, many feminist researchers argue for
conditions that they faced, it was clear that he had research that exposes the conditions of women’s dis-
misgivings about what he witnessed, particularly advantage in a male-dominated society, as Demaiter
during his early time with them. and Adams (2009) did in their study of women in
One way of dealing with the problem of values and the IT sector, featured in the Methods in Motion box
bias is to recognize that research cannot be value- on page 19. Some feminist writers argue that only
free, and to try to ensure that values in the r esearch research on women intended for women is consistent
process are acknowledged and made explicit. This with women’s wider political needs.
is part of a larger process of reflexivity or self- The significance of feminism in relation to values
reflection that researchers are encouraged to carry goes further than this, however. Several feminist
out. As Turnbull (1973, p. 13) put it at the beginning social researchers in the early 1980s proposed that
of his book on the Ik: quantitative research is incompatible with feminist
ideals. For writers such as Oakley (1981), quantitative
The reader is entitled to know something of research is bound up with the male value of control,
the aims, expectations, hopes, and attitudes as seen in the researcher’s control both of the research
that the writer brought to the field [in his case, participant and of the research context and situation.
Western values about the family], for these will Moreover, the research process is seen as a one-way
surely influence not only how he sees things affair in which researchers extract information from
but even what he sees. their subjects and give little if anything in return.
1 | General Research Orientations 19
For many feminists, such a strategy borders on in society; some social scientists may favour
exploitation and is incompatible with the value
increased government intervention in eco-
that feminism attributes to sisterhood and non- nomic affairs, while others defend the free
hierarchical relationships. market; sociologists and political scientists in
The antipathy toward quantitative research re- Quebec may be split between sovereigntists
sulted in a preference for qualitative research among and federalists.
certain feminists. Qualitative research was seen not • A related issue involves research funding.
only as more consistent with the values of femin- Much social research is funded by organiz-
ism, but as more adaptable to those values. Feminist ations such as private firms or government
qualitative research came to be associated with re- departments that may have a vested interest
fusing to take a value-neutral approach and relating in the outcomes of the research. The very fact
to the people under study as human beings rather that these organizations fund some research
than research instruments. This stance demonstrates projects but not others opens the door to pol-
how strongly values can affect the process of social itical influence. Such organizations may seek
investigation. In more recent years, however, femin- to invest in studies that will be useful to them
ist attitudes toward quantitative research have been or supportive of their operations and world
softening, especially when it is employed in conjunc- views. They will often call for researchers to
tion with qualitative research (Oakley, 1998). This tender bids for an investigation in a certain
issue will be revisited in Chapters 5 and 14. area. When social researchers take part in
There are, then, different positions that can be such exercises, they enter a political arena,
taken in relation to values and value-free research. since their research may be designed to please
Few writers today believe it is possible to be truly the funding body. As a result, as Hughes
objective. There is a greater awareness of the limits (2000) observed in relation to research in the
of objectivity, and some of the more categorical pro- field of crime, an investigation of gun crimes
nouncements on the subject, such as Durkheim’s, among the “underclass” is more likely to re-
have fallen into disfavour. At the same time, giving ceive funding than one concerned with mis-
free rein to one’s political beliefs and value positions demeanours committed by agents of the state.
can be problematic. Researchers today still have to Morgan (2000) pointed out that research
fight the all-too-human propensity to demonize funded by government is typically empirical
those whose values are different from their own, and and quantitative; it tends to be concerned with
they still struggle with the temptation to summarily the short-term costs and benefits of a particu-
reject research findings when the researcher’s ideo- lar policy or innovation; and it is generally
logical or moral positions are not compatible with uncritical in the sense that the underlying
their own. government policies are not questioned: all
the government wants to know about is the ef-
Politics in social research fectiveness of their implementation. Political
At various places in this book we take the position issues often arise when the funding agency
that social research has political implications. Here itself is trying to secure a continuous stream
are some examples of the ways in which social re- of government funding.
search may be political: • Gaining access to research subjects can also be
a political process, especially in the case of or-
• Social researchers often take sides. There are ganizations. Access to organizations is usually
many ways this can happen. To consider just mediated by gatekeepers concerned not only
a few: feminist researchers may focus on the about the researcher’s motives but also about
disadvantages that women face and the possi- what the organization stands to gain from the
bilities for improving the position of women investigation, what it will lose by participating
1 | General Research Orientations 21
in terms of staff time and other costs, and the participate nevertheless reported annoyance
potential risks to its image. Often, gatekeepers with being asked about sexual orientation.
seek to influence how the investigation will • There may be pressure to restrict the publica-
take place: what kinds of questions can be tion of findings. Hughes (2000) cited a study
asked; who can and cannot be a focus of study; of plea bargaining in the British criminal jus-
the amount of time to be spent with each re- tice system as a case in point. The researchers
search participant; the interpretation of the had uncovered what were deemed at the time
findings; and the form the reports will take, to be disconcerting levels of informal plea bar-
even to the point of asking to approve drafts. gaining, and they concluded that the formal
• Public institutions, such as police departments, judicial process was being weakened. The
schools, and hospitals, as well as most com- English legal establishment sought to thwart
mercial firms, are concerned with how they the dissemination of the findings and was
are going to be represented in publications. persuaded to allow publication only when a
Consequently, gaining access is almost always panel of academics confirmed the validity of
a matter of negotiation and as such inevitably the findings. Similarly, the editors of academic
turns into a political process. The product of journals may refuse to publish pieces that do
such negotiations is often referred to as “the not conform to their own ideological or polit-
research bargain,” and in many cases there is ical preferences.
more than one bargain that has to be struck.
Once in the organization, researchers often This is just a small handful of the ways in which pol-
discover layers of gatekeepers. For example, itics intrudes in the research process.
let’s say a provincial government grants a re-
search team permission to talk to the boys in Issues related to the research
a group home. Before the research can begin, question
the head of the home has to be brought onside, Some important determinants of how research is
then the staff, and then the actual adolescents. conducted follow from the research question one is
Frequently, one of the staff is then given the trying to answer. Specifically, the choice of research
responsibility of dealing with the fieldwork- orientation, design, or method has to match the
ers. A suspicion that the researchers are really specific research question being investigated. For in-
working for management then has to be over- stance, if a researcher is interested in measuring the
come. And it is unwise to assume that simply impact of various possible causes of a social phenom-
because gatekeepers have granted access, enon, a quantitative strategy is probably appropriate.
a smooth passage will ensue in subsequent Alternatively, if the focus is on the world views of
dealings with the people to be studied. Per- members of a certain social group and how those
haps the most powerful of the boys will turn views develop, a qualitative research strategy—one
out to be the key gatekeeper. Researchers may sensitive to the way participants interpret their social
find themselves used as pawns if subgroups at- world—may be the way to go. If a researcher is in-
tempt to enlist them in advancing a particular terested in a topic on which little or no research has
goal. Research participants who doubt the util- been done, a quantitative strategy may be difficult to
ity of social research may even try to obstruct employ because there is little prior literature from
the research process. For example, in Beagan’s which to draw leads about possible causes. A more
(2001) study of “everyday inequalities” among qualitative, exploratory approach may be preferable
Canadian medical students, some of the
because that type of investigation is typically associ-
students refused to complete a survey when ated with the generation of theory rather than theory
they heard she was asking about the treatment testing (see Table 1.1) and with a relatively unstruc-
of gays and lesbians, and one student who did tured approach to the research process.
22 Part I Fundamental Issues in Social Research
A related dimension that has important im- takes may differ depending on whether the study is
plications for how the research will be conducted quantitative or qualitative in orientation. Quanti-
involves the nature of the topic and of the people tative questions usually ask whether a particular
being investigated. A researcher who wants to study variable or set of variables has an influence on the
individuals involved in illicit activities—for ex- phenomenon of interest. Duncan’s (2010) question,
ample, price fixing, shoplifting, or drug dealing— for example, reveals that her study was designed to
may find it difficult to develop the rapport with inquire into whether personal income affects the
them that is needed to conduct a social survey. It is likelihood that one will join a voluntary association,
not surprising, therefore, that researchers in these and if so, whether that relationship varies depending
areas tend to use a qualitative strategy. On the other on the neighbourhood in which one lives. Implied
hand, it’s unlikely that the hypotheses in Box 1.3 on in her questions is a certain causal model that may
secularization could have been tested using a quali- apply to some larger population, one that extends
tative approach. beyond the people studied. This is the deductive
method at work.
Formulating a research question Qualitative research questions tend to be less
Creating a research question is like picking a destin- specific, and are generally not designed to evaluate
ation for a hike: where you will start out, the route causal models of this kind. They tend to be “open-
you will take, and what you will experience along ended, evolving, and non-directional” (Creswell,
the way are largely determined by the endpoint you 2013, p. 138). The reason for this is that qualitative
are trying to reach. What you stand to accomplish researchers usually want to allow multiple interpret-
with a particular research study and how you will ations and perspectives to emerge from the people
accomplish it are profoundly affected by the goal you and settings they are examining. Also, since qualita-
have in mind, which is expressed in your research tive social scientists often use the inductive method,
question. they don’t have a fully preconceived notion of what
A research question states the purpose of the they expect to find, making it pointless to ask a
study in the form of a question. This is useful because specific question at the beginning of a study. Flex-
a question can often be more evocative and stimulat- ibility in exploring whatever one encounters in the
ing than a simple declarative statement. A question course of doing the research is essential if one is to
arouses curiosity and challenges the researcher to uncover the meanings and interpretations held by
find ways to answer it. Here are some examples of the people studied, and the processes by which they
research questions taken from a quantitative study of develop. For instance, Charpentier and Quéniart’s
voluntary associations in Canada: “Does individual (2017) research question, “What does it mean to
income level affect whether an individual is a vol- older women to grow old in a land of immigration?”
untary association member? Does this effect differ was purposely open-ended because their goal was to
between neighbourhoods?” (Duncan, 2010, p. 578). give the women they studied a voice, not to march
Here are two research questions taken from a quali- them through specific topics and issues that they as
tative study of older immigrant women in Quebec: researchers may have considered to be important.
“What does it mean to older women to grow old in a The data they collected allowed them to discern cer-
land of immigration? What are the effects of immi- tain themes and patterns of response offered up by
gration on their living conditions, especially when the participants in their study, which is the goal of
they have immigrated after 50?” (Charpentier & the inductive method.
Quéniart, 2017, p. 437). Quantitative research may start with the choice
The process of formulating and assessing research of a general area of interest: for example, male
questions is something of an art, but here are some homosexuality. At this stage, a very general research
general thoughts. As you may have noticed in the re- question might be “How do people in general feel
search questions quoted above, the form the question about gays?” This broad research area would have to
1 | General Research Orientations 23
be narrowed down: for example, to “How does the do the people under study deal with those who vio-
Canadian adult population react to the portrayal late group norms? In cases where people are pun-
of gays in television dramas?” But even that is too ished for their violations, how do the offenders make
broad, so the next level of specification might be sense of the punishment they receive, and how does
something like “Do young straight men react dif- it affect how they feel about the other members of
ferently than young straight women to the portrayal the group?”
of gay male romance on television program X? If No single study can answer all the research ques-
so, do those differences reflect a more fluid sexual tions that will occur to the researcher. Only a small
identity on the part of young women as compared number of them can be selected. The need to narrow
to young men?” Such questions could be linked to the topic is not only a matter of the time available
larger theories of sex roles, sexual orientation acqui- and the cost of doing research. It is also a reflection
sition, socialization, and the tolerance of difference of the need for a clear focus.
in society. As suggested above, the research questions may
Research questions in qualitative inquiries may change as the study progresses, for a number of rea-
become more specific over time too, but that usually sons. The discovery of a new data source may change
occurs primarily after the researcher has started the focus a bit, as might some of the initial findings.
to gather data. A general research question may be For instance, if in the study of male homosexual-
retained throughout the study, but more specific ity the researchers find that having a gay relative
sub-questions normally evolve as the research un- in the immediate family makes a large difference
folds. For example, ethnographic research may begin in people’s attitudes, the research question and the
with a general question such as “What is going on attendant methodology and theoretical orientation
here?” (Wolcott, 2008, p. 73), but as the researcher may be revised. The research question may also
learns more about the people to be analyzed, more change because of limitations in time and other
specific questions usually come to mind. These ques- resources available to the researcher. Box 1.7 offers
tions could be something like the following: “How some tips on developing research questions.
A good research question will: can make a contribution to the existing know-
• be as clear as possible so that it will be under- ledge on the topic;
standable to others; • be linked to the other research questions in the
• be researchable: it must allow for the develop- study, so that you can develop a single argument
ment of a research design and the collection of or at least a set of related arguments; this is hard
data; this means that extremely abstract terms to do with unrelated research questions; and
may not be suitable; • be neither too broad (no research project can
• relate in some way to existing studies that sug- do justice to all aspects of a topic) nor too
gest how your question may be approached. narrow (unable to make a meaningful contribu-
Even with a topic that has not been widely re- tion to an area of study).
searched, there will probably be some relevant If you are unsure about how to formulate re-
literature (for example, on related or parallel search questions (or about other aspects of research),
topics). Establishing connections with existing look at journal articles or research monographs to
studies will help to show how your research see how other researchers have handled them.
24 Part I Fundamental Issues in Social Research
Research questions set realistic boundaries for re- • the literature search;
search. A poorly formulated research question can • decisions about the kind of research design to
result in unfocused and substandard research. No employ;
matter how well designed a questionnaire is or how • decisions about what data to collect and from
skilled qualitative interviewers are, clear research whom;
questions are required to avoid going off in unneces- • the analysis of the data; and
sary directions and tangents. Research questions are • the writing up of the findings.
crucial because they guide:
Key Points
• Quantitative and qualitative research constitute • Theory can either precede research and data
different approaches to social investigation and gathering (the deductive method) or emerge out
carry with them important epistemological and of it (induction).
ontological assumptions. • Feminist researchers in the past have tended to
• Epistemological considerations loom large in the prefer a qualitative approach, although this situa-
choice of a research strategy. To a great extent, tion is changing now.
the issues revolve around the advantages and • Values can impinge on the research process in
disadvantages associated with the natural science various ways; as a consequence, research often
(in particular positivist), interpretivist, and critical has political dimensions.
approaches to science. • The political dimensions of research relate to the ex-
• Ontological considerations, such as objectivism ercise of power at different stages of an investigation.
versus constructionism, also constitute import- • Issues related to the research question can also
ant dimensions of the quantitative/qualitative affect decisions about research methods. Clear
contrast. research questions improve the chances of success.
ignored or rejected simply because they come have positive social change as its primary goal
from people who do not share the researcher’s is not worth doing.
political views.
b. The social researcher’s most important duty After all groups have made their arguments,
is to make the world a better place by ridding each one is allowed time for rebuttal. When that
society of things such as racism, economic is finished, the class votes to decide which group
inequality, and sexism. Research that does not made the strongest case.
Relevant Websites
A good way to become familiar with how research is Canadian Journal of Education
done is to read articles published in academic jour- https://cje-rce.ca/#
nals. You may not be able to fully understand the
American Journal of Sociology
methodologies and data analyses at this stage of
www.journals.uchicago.edu/toc/ajs/current
your academic career, but reading articles will help
you become familiar with formal research presenta- American Sociological Review
tions. Below are the Web addresses for some leading http://journals.sagepub.com/home/asr
Canadian and international social science journals. Social Forces
You may not be able to access articles in a particu- http://sf.oxfordjournals.org
lar journal directly; it may be necessary to sign on to
Qualitative Sociology
your institution’s library server first.
www.springer.com/social+sciences/journal/11133
https://zone.biblio.laurentian.ca/handle/10219/378
Chapter Overview
A research design is a framework for the collection and analysis of the data that will be used to answer
the research questions. It must satisfy certain criteria, and the form it takes depends on the research
questions being asked. This chapter will discuss four prominent research designs:
• experimental and related designs (such as the quasi-experiment)
• cross-sectional designs, including survey research, its most common form
• longitudinal designs, such as panel and cohort studies and various forms of qualitative research
• case study designs
“Would you tell me, please, which way I ought to go It’s not hard to feel a little like Alice when you start
from here?” out on the journey that is a research project, especially
“That depends a good deal on where you want to get if the voyage is one you are somewhat reluctant to
to,” said the Cat. undertake (this is not at all uncommon). Where should
“I don’t much care where—” said Alice. I be going with this study? How can I get there? How
“Then it doesn’t matter which way you go,” said can I tell if I’m on the right track? Such questions flood
the Cat. the mind at the beginning of the trip. As Alice learned,
“—so long as I get somewhere,” Alice added as an finding the right way starts with knowing where you
explanation. want to end up. In research, a lot depends on what you
“Oh, you’re sure to do that,” said the Cat, “if you only want to accomplish with the project, which in turn de-
walk long enough.” pends on the sort of research question you have posed.
Alice’s Adventures in Wonderland In this chapter we’ll explore what kinds of destination
(Carroll, 1865/2009, p. 56) are possible, some ways of getting to them, and some
techniques for determining whether you are on the
right path.
▲ Ditty_about_summer/Shutterstock
28 PART I Fundamental Issues in Social Research
people swimming: as one increases the other you’ve selected a case to investigate, how will you
also increases. But does purchasing ice-cream get data on it? Do you use participant observation?
cones cause people to go swimming? Of course Do you observe from the sidelines? Do you conduct
not. Those two things are influenced by a third interviews? Do you examine documents? Do you ad-
factor: warm weather. minister questionnaires? Any or all of these methods
can be used.
Quantitative researchers choose research designs
that help to satisfy these three criteria. Qualitative
researchers, on the other hand, aim for different Research designs
kinds of explanations. They usually seek to produce
a rich description of a person or group based on Experimental design
the perceptions and feelings of the people studied, When people hear the word “research,” the use of
rather than to discover general laws and principles. experiments often comes to mind. However, true ex-
They may also try to get a sense of how those phe- periments are fairly rare in sociology, though they
nomena develop over time and how they can result can be found in areas such as social psychology and
from interpersonal interactions or one’s position in studies of organizations. Researchers in social policy
a power hierarchy. These sorts of accounts are called may also use them to assess the impact of reforms
idiographic explanations, and they usually involve a or new policies. The main reason for discussing ex-
detailed “story” or description of the people studied periments here is that a true experiment involves a
that is based on empathetic understanding. Bowen systematic comparison of what happens when one
(2015), for instance, used this approach to determine set of participants has a particular experience or
how off-street sex workers in Vancouver viewed the treatment while another does not, a logic that can
experience of transitioning from the sex industry also be found in other research designs. The greatest
into “square” (non-sex) occupations and lifestyles. strength of experiments is that they can be effective
On the basis of the insights gained from the study, in establishing causation (sometimes referred to as
she revealed that people who are in the process of internal validity), especially when they are used to
leaving the industry or who are contemplating leav- isolate the effects of a small number of factors. This
ing are often stigmatized and humiliated for their does not mean that the experimental method is ne-
past behaviours, and that the voices of such people cessarily the best design to use. As we shall see below,
must be heard if government transition programs experiments, like other methodological approaches,
are to be effective. Qualitative researchers will have their strengths and weaknesses when they are
choose research designs that will produce these sorts used in social research, and are better suited to some
of idiographic explanations. research topics than others.
Once a design has been selected, a specific method
for collecting data has to be chosen. There are many Manipulation
different ways of gathering data. One way is to use a If experiments are so strong in establishing caus-
preset instrument, such as a self-completion question- ation, why do social researchers not make more
naire or a structured interview schedule. Another is use of them? The reason is simple: to conduct a
to utilize a less formalized method like participant true experiment, it is necessary to do something to
observation or ethnography in which the researcher people and observe the effects. To put the matter
takes part in the activities of a group of people, some- more formally, an experiment manipulates an
times even living among them for a time. independent variable to determine its influence
Consider one of the research designs to be covered on a dependent variable. Typically, some subjects
in this chapter: the case study. It entails a detailed are allocated to a “treatment” group in which the
exploration of a specific case, which could be a com- independent variable is changed or manipulated,
munity, an organization, a person, or an event. Once while others are placed in a “control” group where
30 PART I Fundamental Issues in Social Research
no manipulation takes place. The dependent vari- the rise of second-wave feminism in the 1960s?
able is then observed and measured. The problem is Unfortunately not: experiments are generally lim-
that many of the independent variables of concern ited to relatively simple, short-term manipulations
to social researchers cannot be manipulated. For of independent variables (see Brannigan, 2004).
example, some sociologists maintain that a coun- A further reason is that even where social scien-
try’s national character is affected by whether the tists are successful in using experiments to iden-
country came into existence through revolution tify causal variables, the perceptions and feelings
(for example, Lipset, 1990). In order to test this idea of the participants—which provide information
experimentally, revolution would have to be in- vital to a full understanding of the phenomenon
duced in some randomly selected areas to produce in question—are usually not examined in depth
new countries, while in others no revolution would (Dobash & Dobash, 2000).
be fomented. The researcher would then compare Before moving to a more complete discussion of
the national characters of the revolutionary coun- experimental designs, it is important to introduce
tries with the non-revolutionary ones. Needless a basic distinction between laboratory and field
to say, such experimental manipulations would experiments. The former take place in artificial set-
usually be impossible to carry out. tings, whereas the latter occur in real-life surround-
This example also illustrates a second reason ings such as classrooms and factories. The Rosenthal
why experiments are so rare in sociology: ethical and Jacobson (1968) study described below is a well-
concerns often preclude them. Suppose you are in- known example of a field experiment.
terested in the effect of poverty-induced poor nutri-
tion on children’s academic performance. In order Classic experimental design
to examine this issue experimentally, you could Rosenthal and Jacobson (1968) tried to determine
select a number of six-year-olds and randomly what effect teachers’ expectations had on their
assign them to one of two groups. Those in the first students’ academic performance. In addition to
group would be underfed for a month, while those illustrating the classical experimental design, their
in the second would be given ample, nutritious study is a good example of research that gets a lot
diets. Then you would test the two groups to see if of attention, is then subject to intense methodo-
they differed in academic ability. Although sound logical scrutiny (see Brannigan, 2004, pp. 80–89
from a methodological standpoint, such a study for critiques), and ultimately provides the impetus
would clearly raise serious ethical problems (and for further research designed to improve on the
did, when it was discovered that similar nutrition original.
studies were performed on Indigenous students at The research was conducted in a US school where
residential schools in British Columbia, Manitoba, many poor and minority-group children were en-
Alberta, Ontario, and Nova Scotia in the 1940s and rolled. In the spring all the students were given a test
1950s). In fact, as we will see in Chapter 3, ethical and told it was designed to identify “spurters”—that
issues may arise even when the experimental ma- is, students who were likely to experience a sudden
nipulation appears to be innocuous and is of very improvement in their academic performance. At the
short duration. beginning of the following academic year, the teach-
Another reason why experiments are uncom- ers were given the names of the spurters in their
mon in sociology is that many of the things of classes. But this was a ruse: the students identified as
interest to sociologists—gender roles, political spurters were simply selected at random; they were
preferences, the formation of social movements, not chosen on the basis of their test performance.
and so on—have complex, long-term causes that The test was re-administered eight months after the
cannot be easily simulated in experiments. Could original one, allowing the authors to compare the
the experimental method be used to explain so-called spurters with the other students on things
2 | Research Designs 31
such as IQ scores, reading ability, and intellectual To capture the essence of the classical experi-
curiosity. Since Rosenthal and Jacobson believed mental design, the following simple notation is used:
that there was no initial difference in ability between
the spurters and the others, any improvements were • Obs: an observation made of the dependent
attributed to the fact that the teachers had been led variable. There may be more than two obser-
to expect the spurters to perform better. The authors vations but, to simplify, the design shows the
report that the teachers’ expectations that the spurt- most common configuration: pre-test and
ers would show superior academic performance ac- post-test; here, IQ test scores before the experi-
tually caused those students to do better than the mental manipulation and after.
others, presumably because they received differen- • Exp: the experimental treatment (independ-
tial treatment from the teachers. ent variable); here, the creation of teacher
The Rosenthal and Jacobson study includes most expectancies. No Exp refers to the absence of
of the essential features of what is known as the an experimental treatment and represents the
classical experimental design. Subjects (in this case, experience of the control group.
students) are randomly assigned to two groups. The • T: the timing of the observations made in re-
experimental manipulation (different levels of the in- lation to the dependent variable; here, the time
dependent variable, here heightened teacher expect- when an IQ test is administered
ations) is carried out on the experimental group or
treatment group (the spurters). The other group is not What is the purpose of the control group? Surely
given the treatment and thus forms a control group. it is what happens to the spurters (the experimental
The dependent variable—academic performance—is group) that really matters. For this study to be a true
measured before the experimental manipulation to experiment, however, it must control for (in other
make sure that the two groups really are, on average, words, eliminate) rival explanations of its causal
equal at the start (see Figure 2.1). If they are equal, findings, leaving teacher expectations as the only
and because of random assignment they should be, factor that could have created any differences in per-
the researchers can feel confident that any differ- formance between the two groups. Of course, student
ences in student performance found between the two performance is a complex phenomenon, with many
groups after the manipulation were due to the treat- causes, but the present study wanted to examine only
ment. Everything else about the two groups is pre- one of those causes (teacher expectations). The pres-
sumed to be the same, leaving differences in teacher ence of a control group and the random assignment
expectations as the only possible explanation for any of subjects to the experimental and control groups
differences that are found between the spurters and help to eliminate rival explanations for differences
non-spurters. in academic performance, which in this case are any
explanations other than different teacher expecta-
tions. To see this, consider some threats that would
8 months
pose serious challenges to the study’s conclusions
Experimental T1 T2
group
if there were no control group or random assign-
Obs1 Exp Obs2
spurters IQ Teacher IQ
ment. The following list is based on a book written
expectancies by Cook and Campbell (1979). In each situation, the
Random
possibility of a rival interpretation of Rosenthal and
assignment
Jacobson’s findings is proposed, but the presence of
Control Obs3 No Exp Obs4 both a control group and random assignment greatly
group IQ No teacher IQ
non-spurters
reduces the threat. As a result, confidence in the
expectancies
finding that teacher expectations influenced student
FIGURE 2.1 Classical experimental design performance is enhanced.
32 PART I Fundamental Issues in Social Research
• History. This refers to events other than the they got older, regardless of the effect of teacher
manipulation of teacher expectations that expectancies. But the control group would also
might have caused the spurters’ scores to rise. mature, so maturation effects cannot explain
For example, suppose that the school’s princi- the differences between it and the control group.
pal had taken steps to raise standards in the • Selection. When subjects are not assigned ran-
school. Without a control group, we could not domly to the experimental and control groups,
be sure whether it was the teachers’ expecta- variations between them in the post-test may
tions or the principal’s action that produced be due to pre-existing differences between the
the increase in spurters’ grades. With a control two groups. For example, if all the best stu-
group, we can say that the principal’s action dents were given the spurter label, it might
should have had an effect on the control sub- have been their pre-existing academic ability
jects too, and therefore that the differences rather than teacher expectations that caused
between the experimental and control groups them to do better. However, since a random
can be attributed to the effect of teacher ex- selection process was employed here, the “se-
pectations alone. lection” risk is greatly reduced. With random
• Testing. This threat refers to the possibility assignment, the top students would make up
that subjects may become more experienced roughly the same proportion of the spurter
at taking a test or sensitized to the aims of group as the non-spurter group, cancelling
the experiment as a result of the pre-test. The out the effect of pre-existing academic ability.
presence of a control group, which presumably However, even with the use of random assign-
would also experience the same things, dimin- ment, if the number of people in each group
ishes this possibility. is relatively small, there is still a risk of pre-
• Instrumentation. This threat refers to the possi- existing differences between the experimental
bility that changes in the way a test is adminis- and control groups.
tered can account for an increase (or decrease)
in scores between a pre-test and post-test; for Even if all these threats have been overcome, fur-
example, perhaps the teachers know their stu- ther issues may arise. First there is the question of
dents better or are more friendly the second whether the variables used in the study have been
time they give the test. Again, if there is a con- adequately measured. This refers to the matter of
trol group, the people in that group should be measurement validity, a topic addressed in detail in
affected as well. Chapter 4. In the case of the Rosenthal and Jacob-
• Mortality. A particular problem for studies that son (1968) study, there are potentially two aspects
span a long period of time is the risk that some to this question. First, has academic performance
subjects will leave the experiment before it is been adequately measured? Reading scores seem to
over: for example, some students might move correspond to what they are supposed to be measur-
to a different school, or experience a long-term ing. However, given the controversy surrounding IQ
illness. Since this problem is likely to affect tests and what they measure, we may feel unsure that
the control group too, it may not make a dif- gains in IQ test scores can be regarded as strongly
ference to the results. However, experimenters indicative of academic performance. Similarly, to
should try to determine whether mortality has take another of the authors’ measures, is intellec-
affected the experimental and control groups tual curiosity a valid measure of academic perform-
differently. ance? Does it really measure what it is supposed to
• Maturation. Quite simply, people change over measure?
time and the ways in which they change may Another question is whether the experimental
have implications for the dependent variable. manipulation really worked. In other words, was
The spurters might have improved anyway as the identification of some schoolchildren as spurters
2 | Research Designs 33
enough to create the conditions needed for the hy- the result that their responses are affected as
pothesis about teacher expectations to be tested? The they become more test-wise. Consequently, the
study depended on the teachers’ remembering the findings may not be generalizable to groups
“information” they were given about the spurters that have not been pre-tested, and, of course,
for the duration of the experiment, but it is possible in the real world people are rarely pre-tested.
that as time went on some of them came to doubt or This may have occurred in the Rosenthal and
forget it; in that case, the manipulation would have Jacobson research, since all students were pre-
been contaminated. tested at the end of the previous academic year,
A second set of issues concerns the generaliz- so the students new to the school in the fall
ability of a study’s findings. In other words, do the (when the teachers were told about spurters)
study’s conclusions apply to other people, settings, presumably were not pre-tested.
or time periods? These matters pertain to the study’s • Reactive effects produced by the experimental
external validity. Cook and Campbell (1979) iden- arrangements. People are frequently, if not in-
tified five major threats to the external validity and variably, aware of the fact that they are par-
hence the generalizability of findings derived from ticipating in an experiment. Their awareness
an experiment: may influence how they respond to the experi-
mental treatment; for example, they may react
• The representativeness of the study participants. by behaving in a socially acceptable manner
To what social and psychological groups can a rather than sincerely and spontaneously,
finding be generalized? Can it be generalized to which could affect the generalizability of the
a wide variety of individuals who differ in eth- findings. Since Rosenthal and Jacobson’s sub-
nicity, social class, religion, gender, and type jects do not appear to have been aware that
of personality? In the Rosenthal and J acobson they were participating in an experiment,
study, the students were largely from poorer this problem is unlikely to have been signifi-
groups and a large proportion from ethnic cant. The issue of reactivity and its potentially
minorities. This may limit the generalizability damaging effects is a recurring theme in many
of the findings. types of social research.
• The effects of the setting. Can the results of a
study be applied to other settings (in Rosenthal Then there is the question of replicability. A study
and Jacobson’s case, to other schools)? There is is replicable if others are able to repeat it and get the
also the wider issue of whether expectation ef- same results. Sometimes replications are conducted
fects can also be discerned in non-educational to make sure that the original research was carried
settings. out properly; this is especially important if a study’s
• History effects. This threat raises the question results do not match prior findings on the topic. For
of whether the findings can be generalized to replication to be possible, the initial researcher must
the past and into the future. The Rosenthal and spell out all the research procedures in great detail.
Jacobson research was conducted more than Rosenthal and Jacobson laid out their procedures
50 years ago. Would the findings still apply and measures in detail, and anyone carrying out a
today? Also, their investigation was conducted replication could obtain further information from
at a particular juncture in the school year. them. Consequently, their research is replicable, al-
Would the same results have been obtained if though there has never been an exact replication.
the research had been conducted at different Clairborn (1969) conducted one of the earliest rep-
points in the year? lications and followed a procedure very similar to
• The effects of pre-testing. As a result of being pre- Rosenthal and Jacobson’s, although the study was
tested, subjects in an experiment may become carried out in three middle-class suburban schools,
sensitized to the experimental treatment, with and the timing of the creation of teacher expectancies
34 PART I Fundamental Issues in Social Research
was different from the original study. Clairborn also randomly assigned, were not told this. The data
failed to replicate Rosenthal and Jacobson’s findings, showed that women scored lower than men when
casting doubt on the external validity of the original informed of this “fact.” When told that the test was
research and suggesting that the first two threats to to compare Canadians and Americans, there was
external validity referred to earlier in this chapter no gender difference. Without the chance to ran-
may have limited the applicability of the findings. domly assign subjects, there would always be some
Virtually all experiments in the social sciences doubt that the treatment (being told that women
involve deception of some kind. In the Rosenthal score lower than men on the test) rather than a pre-
and Jacobson study, for example, the experimenters existing difference between the two sets of subjects
told the teachers that certain students were spurters, actually caused the women to do less well than the
which simply was not true. But could the experi- men. Also, laboratory experiments are more easily
ment have been carried out without any deception? replicated because they are less bound to a natural
For example, could the experimenters have told the milieu that could be difficult to reproduce.
teachers that they were testing the effects of teacher However, laboratory experiments also suffer from
expectations on student achievement? Could they a number of limitations. Many of those shortcomings
have given the teachers a list of students and said, arise from low external validity. For instance, the ex-
“Pretend that these students are gifted, although perimental setting may not mirror real-world experi-
they are not any more gifted, on average, than the ences and contexts, despite the fact that the subjects
students not on the list. We want to see whether you are very involved in most experiments and take them
would treat such children differently, and whether very seriously. In addition, the treatment effects may
differential treatment affects their academic per- be unique to the people in the study; others may not
formance.” Surely, that sort of approach would not react the same way. In the case of Fisher and Ma’s
have produced authentic behaviour on the part of the (2014) study described in Box 2.1, for example, the
teachers, and the teachers might well have refused to subjects were drawn from an online panel and hence
participate under those conditions for ethical and may not be representative of the general population,
practical reasons. Clearly, some form of deception if only because their willingness to go online and
was necessary for the experiment to work. their facility with the Internet may have made their
But deception raises ethical concerns—it is basic- responses to the experimental stimuli distinctive.
ally a form of lying. Chapter 3 will discuss the ethical They were also volunteers, who generally differ from
implications of using deception in social research, non-volunteers (Rosnow & Rosenthal, 1997, ch. 5). In
as well as other sorts of ethical dilemmas facing re- addition, the fact that they were given an incentive to
searchers in the social sciences. participate (they were paid a small fee) might have fur-
ther distinguished them from others since not every-
The laboratory experiment one would be equally motivated to earn the reward
One of the main advantages of laboratory over field given to those who participate. There was no effect
experiments is the researcher’s greater control over of pre-testing because, as in many experiments, the
the research environment. In particular, it is easier to participants described in Box 2.1 were not pre-tested.
randomly assign subjects to different experimental However, it is quite possible that reactive effects oc-
conditions in the laboratory than in a real-life curred: the subjects knew they were in an experiment,
situation, which enhances the researcher’s ability and that may have affected their behaviour.
to establish nomothetic causation. For example,
Walsh and colleagues (1999) were able to tell some Quasi-experiments
randomly assigned university students in Eastern Quasi-experiments have some characteristics of
Canada that previous results on the mathematics test the experimental model but lack some of the fea-
they were about to take showed that women perform tures that help the researcher establish causation.
less well than men on such tests. Other students, Several different types of quasi-experiments have
2 | Research Designs 35
It is commonly recognized, both in scientific circles from a village in Africa that had just been struck
and in everyday life, that there are tangible bene- by a mudslide, making her a homeless orphan.
fits associated with being physically attractive. It is People experiencing the “low severity” manipula-
also well known that attractive people are thought tion were told of the mudslide, but were informed
to possess more positive attitudes and behavioural that the young girl’s home and parents were not
traits than the less attractive; hence the “beauti- harmed by it. The results indicated that subjects
ful is good” stereotype. Yet two researchers at the in the low severity condition who were presented
University of Alberta (Fisher & Ma, 2014) found that with the “attractive” version of the girl’s photo-
under certain experimental conditions, partici- graph showed lower levels of empathy toward her
pants felt less empathy toward an “attractive” child. than those shown the less attractive image: the
The researchers wanted to know whether the opposite of the “beautiful is good” stereotype. In
attractiveness of children in need affects the em- the high severity condition, physical attractive-
pathy that non-relatives feel toward them. In one ness had no effect on the level of empathy. The
version of the study, the perceived attractiveness researchers reasoned that when severe harm is
of a young girl was manipulated by showing one not imminent, people perceive attractive chil-
group of participants a picture of her that had not dren to be more socially competent and hence
been altered. A second group was shown a version more capable of taking care of themselves. But
of the picture that had been digitally modified to when children are in grave danger or distress,
make the girl appear less attractive. A “high se- those superficial considerations are overridden by
verity” of need condition was created by telling strong feelings of compassion and a desire to help,
some participants that the child in the picture was regardless of how attractive the child may be.
been identified (Cook & Campbell, 1979). Not all groups may not have been equivalent on all relevant
of them can be covered here, but a particularly in- characteristics before the independent variable was
teresting type are “natural experiments,” in which introduced. For instance, in the earthquake example,
experiment-like conditions are produced by natur- there may have been some pre-existing differences
ally occurring phenomena or changes brought about between the two cities before the earthquake struck
by people not doing research. When that occurs, re- that contributed to post-earthquake differences. So
searchers can gather data in much the same way they if the stricken city showed an increase in civic pride,
do in experiments. For example, if an earthquake hit perhaps that was not because of the earthquake, but
a particular city, but a city of comparable size and because it had a more charismatic mayor than the
composition a short distance away was spared, the neighbouring city. However, the results of such stud-
conditions for a natural experiment would be present. ies are still compelling because they are real rather
The effects of the natural disaster on civic pride or at- than artificial interventions in social life, making
titudes toward local political leaders could be meas- them high in external validity.
ured by comparing the two cities on those variables. St Helena in the South Atlantic provided a fascin-
However, in natural experiments, it is usually im- ating natural laboratory for the examination of vari-
possible to randomly assign subjects to experimental ous claims regarding the effect of television violence
and control groups. The absence of random assign- on children when TV was introduced to the island for
ment casts doubt on any causal inferences, since the the first time in the mid-1990s. The findings—from
36 PART I Fundamental Issues in Social Research
video footage showing young children at play during allows researchers to estimate the strength of environ-
school breaks, from diaries kept by about 300 of the mental and genetic influences on a variety of attitudes
children, and from ratings by teachers—suggested and behaviours (Plomin et al., 2013). The presence
that the introduction of television was not followed of greater behavioural or attitudinal similarities
by an increase in the number of aggressive acts ob- among monozygotic than dizygotic twins suggests
served (Charlton et al., 1998, 1999). The research- a genetic influence. For example, recent behavioural
ers suggest that in environments such as St Helena, genetic studies indicate that social and political atti-
where children are closely watched by the commun- tudes are affected not only by societal and situational
ity and are expected to avoid violence or aggression, factors, but also by genes (Alford et al., 2005; Bell &
television may have little effect on their behaviour. Kandler, 2017; Bell et al., 2009, 2012). Additional ex-
Another type of natural experiment can be found in amples of quasi-experiments are presented in Box 2.2.
twin studies, which are commonly used in behavioural Quasi-experimental designs have been particu-
genetic research. Monozygotic (popularly known as larly prominent in evaluation research studies, which
“identical”) twins share 100 per cent of their genetic examine the effects of organizational innovations such
structure, whereas dizygotic (“fraternal”) twins have as a longer school day or greater worker autonomy in a
in common only about 50 per cent of the genes that plant (see Box 2.3). Sometimes the results are surpris-
vary between human beings. This natural difference ing. A quasi-experimental investigation on the effect of
A common type of quasi-experiment compares real social world, and as such have to meet goals
data collected before and after a policy shift by that are very different from those of experimental
government or industry. For example, the num- research. That makes it next to impossible for gov-
bers of car accidents before and after the lowering ernments to randomly assign people to different
of a speed limit can be compared. If the number conditions. Therefore, they usually have to make
goes down after the speed limit has been lowered, do with quasi-experimental evaluations of policy
the policy would appear to be a success, but one changes. A criminologist may want to randomly
cannot be sure because there is no control group assign criminals to jail or home custody and then
not experiencing the change. Perhaps the change compare the two forms of detention, but the po-
was caused by something other than the reduction tential for some of the at-homes to reoffend would
in the speed limit, such as increased media cover- be seen as too great a risk. Because minor crim-
age of car accidents. inals are more likely to get home custody than
Another example might involve the installa- are more serious criminals, a fair test is not pos-
tion of cameras to detect speeding. If the cameras sible without random assignment. Similarly, when
are placed only in randomly selected places and Canada abolished the death penalty, it was not for
not in others, the research changes from a quasi- the purpose of conducting an experiment: aboli-
to a real experiment, since the locations with- tion was a policy decision. Data could be examined
out cameras would constitute a control group. as if the change were part of a quasi-experiment,
Governments find it difficult, however, to subject for example, by comparing the murder rate before
some people to one condition and others to an- and after capital punishment was abolished; but
other. The prison system, schools, and other insti- the legislation was definitely not intended for that
tutions controlled by governments are part of the purpose.
2 | Research Designs 37
support for people who take care of the elderly (Demers, comparison: at the very least, an experiment allows
1996) showed that the extra support made caregivers the researcher to compare the results obtained from
feel less depressed, but more burdened. Demers was an experimental group with those obtained from a
not sure why, although she speculated that perhaps control group. In the case of the Fisher and Ma (2014)
the help was seen as something else to be coordinated experiment in Box 2.1, the research compared the ef-
and managed. The situation that Demers found herself fects of different levels of a child’s physical attract-
in—not being sure how to interpret her findings—is iveness and distress on the level of empathy felt by an
actually quite common. It’s one thing to come up with observer. The advantage of such comparisons is that
findings like hers, but another to explain them. The they permit a better understanding of the phenom-
solution usually involves doing more research. enon in question than would be possible if it were
Quasi-experiments are also used to evaluate the examined under one condition alone. The argument
effectiveness of institutional policies. For example, that in certain situations greater physical attractive-
Hanson et al. (2004) wanted to know whether treat- ness in children evokes less empathy is much more
ing sex offenders after they are released from prison persuasive when the empathy felt for attractive chil-
has any effect on their chances of reoffending. A total dren can be compared with that elicited by those per-
of 724 convicted male sex offenders were divided into ceived to be less attractive. While the experimental
two groups: those who underwent mandatory treat- design is typically associated with a quantitative re-
ment after release to community supervision, and search strategy, the specific logic of comparison pro-
those who were released prior to the implementation vides lessons of broad applicability and relevance.
of the mandatory treatment program. (Some of the
latter group had received treatment during or prior Cross-sectional design
to incarceration, although the extent of that treat- Many people associate cross-sectional designs with
ment was not known.) This division was not perfect, questionnaires and structured interviewing. How-
since the men in the first group may have differed ever, other data-gathering techniques may also be
in some relevant way from those in the second; the used in cross-sectional research, including struc-
fact that some in the latter category received treat- tured observation and analysis of official statistics
ment is also an issue. In any case, the results showed or diaries. These will be covered in later chapters,
that, after an average 12-year follow-up period, there but we will outline the basic structure of the cross-
were equal rates of reoffending in both groups: about sectional design here. In cross-sectional studies,
20 per cent for sex crimes, which was not very en- observations are taken at one point in time—there
couraging for the professionals who treat offenders. are no before-and-after comparisons. Also, cross-
sectional designs do not involve any manipulation
Significance of experimental design of the independent variable: they are like snapshots
As we noted at the outset of this section, true experi- taken of a group or phenomenon at one point in time.
ments are rare in sociology, but they are worth dis- A cross-sectional design entails the collection of
cussing because, where they are practicable, they allow data (usually quantitative) on more than one case.
researchers to isolate the causal influence of a particu- Researchers are interested in variation between dif-
lar variable—a goal to which researchers using other ferent people, families, nation-states, and so on, and
designs also aspire. Cross-sectional designs of the kind variation can be established only when more than
associated with survey research (discussed below) one case is examined. Researchers employing this
offer another way of evaluating causal hypotheses. design usually select many cases, for at least two
reasons. For one thing, a larger number makes it
Logic of comparison more likely that variation will be encountered in all
It is important to draw attention to a significant the variables of interest. A second reason is that cer-
general lesson that an examination of experiments tain statistical techniques are likely to require large
reveals. A central feature of any experiment is a sample sizes (see Chapter 7).
38 PART I Fundamental Issues in Social Research
A key question asked in evaluation research is at its completion. The researchers also did a quali-
whether a new policy initiative or organizational tative analysis of the project by conducting five
change achieved its goals. Ideally, to answer that focus groups at the conclusion of the program. The
question the design would have one group that is quantitative results indicated that the participants
exposed to the treatment—the new initiative—and had higher levels of perceived overall health and
a control group that is not. Since it is often not feas- sense of community, and lower levels of physical
ible or ethical to randomly assign research partici- pain, when the program was over. The themes
pants to the two groups, such studies are usually that emerged from the focus groups included the
quasi-experimental. For instance, data gathered conclusions that the program provided the seniors
from people before a change may be compared with structure and discipline, facilitated coping, re-
with data acquired after; the “before” people quired hard work and effort, brought out their art-
become the control group, the “after” people the istic side, promoted social involvement, and made
experimental group. This approach has the added a positive contribution to the community.
advantage that the two groups are basically the Quantitative quasi-experimental designs in
same, making random assignment unnecessary. evaluation research go back a long way, but as the
Such a design was used to evaluate the effect Phinney et al. (2014) study indicates, evaluations
of a community arts program on the well-being of based on qualitative research have also emerged.
older adults in the Vancouver area (Phinney et al., Although there are differences of opinion about
2014). Over three years, four groups of participants how qualitative evaluation should be carried
took part in the collective creation of a physical out, there is consensus on the importance of,
work of art or a performance that was presented first, understanding the context in which an inter-
to the public. Baseline quantitative measures of vention occurs and, second, hearing the diverse
well-being were taken in the first year of the pro- viewpoints of the stakeholders (Greene, 2000).
gram, with the same measures administered again For example, Pawson and Tilley (1997) advocate
In quantitative studies, data are collected on to difficulties in holding down a job and thus
two or more variables, which are then examined poverty? Or is it a bit of both? To take another
to detect patterns of association. This approach example, a study of 1000 men found that those
sometimes makes it difficult to show cause and who had two or more orgasms a week exhibited
effect because the independent and dependent a 50 per cent lower mortality risk compared with
variables are measured simultaneously, making men who had on average less than one orgasm
any demonstration of time order (showing that per week. It may be tempting to conclude that
the cause actually precedes the effect in time) hard male orgasm leads to longer life expectancy, but
to prove. For example, there is a well-supported it is also possible that the causal arrow points in
negative relationship between social class and the other direction: men who are ill (and thus at
serious forms of mental illness: more poor people greater mortality risk) are less likely to be sexually
are mentally ill than rich people. But there active in the first place (Houghton, 1998, p. 14).
is also a debate about the nature of that relation- This finding and the preceding one are similar,
ship: Does being poor lead to stress and therefore showing what Blaxter (1990) called “an ambiguity
to mental illness? Or does being mentally ill lead about the direction of causal influence.” There is
2 | Research Designs 39
only an association between the two variables—no administering research instruments (struc-
clear causal link. However, as will be shown below tured interview or self-completion question-
and in Chapter 8, there are a number of ways for naire, etc.), and the analysis of data.
researchers to draw cautious inferences about • Establishing causality can be problematic.
causality using cross-sectional designs. As was just suggested, it may be difficult to es-
tablish causal direction from the resulting data.
Replicability, causal inferences, Cross-sectional research designs can identify
and external validity associations that are indicative of causation,
How does cross-sectional research measure up in but other designs may have to be employed to
terms of replicability, the ability to establish causa- substantiate causal inferences.
tion, and external validity? • External validity is strong when the sample is
a random one. When non-random methods
• Replicability characterizes most cross-sectional of sampling are employed, external validity
research, so long as the researcher spells out becomes questionable, an issue addressed in
the procedures for selecting respondents, Chapter 7.
40 PART I Fundamental Issues in Social Research
Variables that cannot be manipulated socioeconomic status are “givens” and not really
amenable to the kind of manipulation necessary
As noted in the section on experimental design, for a true experimental design. Fictitious manipu-
in much (if not most) social research it is not lations are possible, such as when an experimenter
possible to manipulate the variables of interest. digitally alters photographs to produce different
This is a key reason why most quantitative social ages or ethnicities, perhaps to see the effects on job
research employs a cross-sectional design rather
offers, but the manipulation is limited to the exter-
than an experimental one. To more or less all in- nal signs of age and ethnicity, missing the more sub-
tents and purposes, things like ethnicity, age, and jective and experiential aspects.
2 | Research Designs 41
However, the very fact that certain variables are accounts of factors that affected their past and
givens provides a clue as to how to make causal in- present behaviour.
ferences in cross-sectional research. Many of the
variables of interest can be assumed to be tem- Structure of cross-sectional designs
porally prior to other variables. For example, in a A cross-sectional design collects data on a series of
relationship between ethnic status and alcohol con- variables (Obs1, Obs2, Obs3, Obs4, Obs5, . . . Obsn) for
sumption, the latter cannot be the independent vari- different cases (people, households, cities, nations,
able because it occurs after ethnicity. Ethnicity still etc.) at a single point in time. The effect is to create
cannot be said with certainty to be the cause, how- a data set that comprises variables Obs1 to Obsn and
ever: it is just a possible cause. In other words, even cases case1 to casen, as in Figure 2.2. Each cell in the
though researchers are unable to manipulate things matrix has data in it.
like ethnic status or gender, causal inferences can
still be cautiously drawn from cross-sectional data. Longitudinal design(s)
The current discussion of the cross-sectional With a longitudinal design, cases are examined at a
design places it firmly in the context of quantitative particular time (T1) and again at a later time or times
research. But qualitative research can also use a form (T2, T3, and so on), but without the manipulation
of cross-sectional design. For example, Beardsworth of an independent variable that characterizes ex-
and Keil (1992) carried out a study of the dietary be- periments (see Box 2.4). When used in quantitative
liefs and practices of vegetarians. They administered research, it allows insight into the time order of vari-
“relatively unstructured interviews,” which were ables and is better able to deal with the problem of
“guided by an inventory of issues,” with 76 vegetarians “ambiguity about the direction of causal influence”
and vegans (1992, p. 261). The interviews were taped that plagues cross-sectional designs. Because poten-
and transcribed, yielding a large body of qualitative tial independent variables can be identified at T1, the
data. The research was not preoccupied with quanti- researcher is in a better, if not perfect, position to
tative criteria such as establishing causation, external infer that the effects identified at T2 or later occurred
validity, replicability, measurement validity, and so after changes occurred in the independent variables.
on. Nonetheless, the conversational interview style In all other respects, the points made above about
made the study more externally valid than research cross-sectional designs are the same as those for
using more formal instruments of data collection. longitudinal designs. In spite of its heightened abil-
The study was concerned with the factors that in- ity to show cause and effect, the longitudinal design
fluence food selection, like vegetarianism. The very is not frequently used in quantitative social research
notion of an “influence” carries a strong connotation because of the additional time and cost involved.
of causality, suggesting that qualitative researchers
can also be interested in the investigation of causes
and effects, although they do not use the language Obs1 Obs2 Obs3 ... Obsn
of quantitative research with its talk of independ-
Case1
ent and dependent variables. As well, the emphasis
was much more on understanding the experience of Case2
something like vegetarianism than is often the case Case3
with quantitative research.
Case4
This qualitative research bears many similarities
to the cross-sectional design in quantitative re- Case5
search. It entailed interviewing quite a large number ...
of people at a single point in time. And, as with Casen
many quantitative studies using a cross-sectional
design, the examination of people’s past and current FIGURE 2.2 The data rectangle in cross-sectional
eating habits was based on the subjects’ retrospective research
42 PART I Fundamental Issues in Social Research
Case study research frequently includes a longi- A longitudinal element also occurs when a
tudinal element. The researcher can be a partici- case that has been studied is returned to at a
pant observer in an organization for an extended later time. A particularly interesting example of
length of time, or may do ethnographic research this occurred in “Middletown,” a pseudonym for
with a community for many months or years, or an American Midwest town first studied by Lynd
may conduct structured or qualitative interviews and Lynd (1929) in 1924–5 and restudied in 1935
with individuals over a prolonged period. More- during the Depression to see what changes had
over, the researcher may be able to inject an addi- occurred (Lynd & Lynd, 1937). In 1977, the com-
tional longitudinal element by analyzing archival munity was again restudied, this time in a post–
information and asking respondents to recall Vietnam War setting (Bahr et al., 1983), using
events that occurred before the study began, thus the same research instruments but with minor
discovering some history. changes.
Longitudinal studies can take different forms. There are two basic types of longitudinal design:
For example, Goyder et al. (2003) examined how the panel study and the cohort study. With the
gender-influenced evaluations of occupational pres- former, the same people, households, or other groups
tige had changed over 25 years. Some of the earlier are studied on at least two different occasions. One
male advantage had disappeared; indeed, some oc- example is the National Longitudinal Survey of
cupations showing a female incumbent were rated Children and Youth (NLSCY), described in Box 2.5,
more highly than the same occupation with a male in which the same children were studied in succes-
incumbent. Baer et al. (2001) did a 15-nation study sive years. Another is the three-wave study of family
examining whether membership in clubs and associ- structure and children’s socioeconomic attainment
ations had changed over time. Kerr (2004) and Kerr (Seabrook & Avison, 2015) discussed in the Research
and Michalski (2007) investigated hyperactivity in in the News box (p. 43). A panel study, especially
Canadian children as they grew older and its sources one at the household level, needs rules for handling
in poverty and family structure. new entrants (e.g., as a result of marriage or elderly
This panel study is a long-term effort to monitor Resources Development Canada. The study hopes
Canadian children’s development and well-being to follow the subjects until they are 25 years old,
as they mature from infancy to adulthood. It began and to contribute to the development of policies
with a representative sample of children 11 years that help children live healthy, active, and reward-
of age or younger in 1994–5 being interviewed, ing lives (see Michaud, 2001). The data for Cycle 8,
with follow-ups every two years. Statistics Canada which began in September 2008, were released in
collects the data, with direction provided by Human November 2010 (Statistics Canada, 2010).
2 | Research Designs 43
relatives moving in) and exits (e.g., as a result of some subjects may die, some may move away, and
marital break-up or children leaving home). some may simply choose to withdraw at later stages
In a cohort study people sharing the same experi- of the research. For instance, in the study by Din-
ence, such as being born in the same year or gradu- ovitzer et al. (2003) on the educational attainment of
ating from a particular school at the same time, are immigrant youth, researchers were able to talk to only
studied over time, but the same people may not be 65 per cent of those originally surveyed 19 years ear-
studied each time. For example, Walters (2004) used lier. Comparing those who participated in the second
the 1982, 1986, 1992, and 1995 National Graduates round with those who were lost from the study, the
Surveys to examine trends in the economic fortunes authors found no significant differences. That sort of
of Canadians graduating from post-secondary insti- comparison is a common practice; when no difference
tutions. The same people were not selected for the is found, the losses are treated as random and thus ac-
sample each year, but the information was still useful ceptable to ignore. The main problem with attrition
in understanding each graduating cohort, as well as is that those who leave the study may differ in some
the similarities and differences between cohorts. important respects from those who remain, so that
Panel and cohort studies share similar features. the latter do not form a representative group. How-
In social sciences such as sociology, social policy, and ever, there is some evidence from panel studies that
human geography, quantitative research using these attrition declines with time (Berthoud, 2000); in other
designs usually takes the form of repeated survey words, those who do not drop out after the first wave
research using a self-completion questionnaire or or two of data collection tend to stay on the panel.
structured interview. Used in this way, panel and Second, there are few guidelines for deter-
cohort studies can both illuminate social change and mining the best timing for further waves of data
improve the understanding of causal influences. collection. Finally, there is evidence of a panel con-
Panel and cohort studies also share similar prob- ditioning effect, whereby continued participation in
lems. First, there is the problem of sample attrition: a longitudinal study affects respondents’ behaviour.
Menard (1991) cited a study of family caregiving in • a sector of the economy, such as the seafood
which 52 per cent of respondents attributed a change processing industry in rural New Brunswick
in the way they cared for relatives to their participa- examined in Knott’s (2016) research on differ-
tion in the research. ent types of mobile labour.
As suggested above, qualitative research may also
incorporate elements of a longitudinal design. For There is a tendency to associate case studies with
example, it is used in ethnographic research when qualitative research, such as Pratt and Valverde’s
the ethnographer is in a location for a lengthy period (2002) study of the Toronto Somali community.
of time or when interviews are carried out on more Exponents of the case study design often favour
than one occasion in order to examine change. In an methods like participant observation and unstruc-
example of the latter, Smith et al. (2004) described a tured interviewing, which are viewed as particularly
study of the experiences of citizenship for 110 young helpful in generating an intensive, detailed examin-
people. They were interviewed in depth in 1999 and ation of a case. However, case studies can also use
then re-interviewed at two-year intervals to exam- quantitative methods, or some combination of quan-
ine changes in their lifestyles, feelings, opinions, titative and qualitative.
and ambitions in relation to citizenship issues. Only With a case study, the case is an object of interest
64 people participated in all three waves of data col- in its own right and the researcher aims to provide
lection, illustrating the high rate of sample attrition an in-depth elucidation of it. Sometimes it is diffi-
in this style of research. cult to distinguish case studies from other research
designs because almost any kind of research can be
Case study design construed as a case study. Even research based on
The basic case study design entails a detailed and in- a national, random sample of Canadians could be
tensive analysis of a single case. A case may be: considered a case study of Canada.
What distinguishes a true case study is the goal
• a single community, as in Hughes’s (1943) clas- of finding and revealing the features of the case.
sic study of Drummondville, a textile town in Collecting in-depth, often qualitative data that may
Quebec, or in Pratt and Valverde’s (2002) re- be unique to time and place is characteristic of this
search on Somalis in Toronto; sort of research. Case studies are often idiographic in
• a single family, as in Lewis’s (1961) study of the nature, seeking to provide a rich description of the
Sánchez family in Mexico; subject matter, like Shalla’s (2002) study of how Air
• a single organization, such as the automobile Canada’s customer sales and service agents became
factory studied by Rinehart (1996), or a group victims of airline restructuring.
within an organization, like the nurses in a When the predominant research orientation is
Hamilton hospital researched by White (1990); qualitative, a case study tends to take an inductive
• a person, as in Nemni and Nemni’s (2006) approach to the relationship between theory and re-
study of Pierre Trudeau or Foran’s book on search. If a mainly quantitative strategy is adopted,
Mordecai Richler (2010); such research is the research is often deductive, guided by specific
characterized by use of the life history or bio- research questions derived from social theories.
graphical approach (see Chapter 11);
• a single event, like the fight against locating a Measurement validity, causal inference,
home for recovering addicts in a Richmond, external validity, and replicability of
British Columbia, neighbourhood (Huey, case studies
2003), or the Alberta election analyzed by Bell The question of how the case study fares on the re-
et al. (2007); search design criteria of measurement validity, estab-
• a state or province, as in Laplante’s (2006) lishing causation, external validity, and replicability
study of the rise of cohabitation in Quebec; or depends in large part on whether the researcher feels
2 | Research Designs 45
these criteria are appropriate for their work. Writers up in Samoa seems to have been motivated
of qualitative case study research tend to play down by her belief that it represented a unique case
or ignore the salience of these factors (cf. Stake, and thus could challenge the then popular
1995). Those influenced by the quantitative research nature-over-nurture hypothesis. She reported
tradition see them as more significant, and usually that, unlike adolescents in most other societies,
try to develop case studies that meet the criteria. Samoan youth did not suffer a period of anxiety
One question that has generated a great deal of and stress in their teenage years. She explained
discussion concerns the external validity or gener- this by their culture’s strong, consistently en-
alizability of case study research. How can a single forced standards of conduct and morality. These
case possibly be representative of other cases? For factors were of interest because many readers
example, would the findings from a study of the To- thought they might contain lessons for the West.
ronto police department be generalizable to all large • The revelatory case. The basis for the revelatory
urban police departments in Canada? The answer, case exists “when an investigator has an oppor-
of course, is that the findings probably cannot be tunity to observe and analyse a phenomenon
applied to other police departments. Case study previously inaccessible to scientific investiga-
researchers do not usually delude themselves into tion” (Yin, 1984, p. 44). This can happen when
thinking that it is possible to identify typical cases previously unavailable evidence becomes ac-
that can represent a class of objects, whether factor- cessible, as was the case with certain KGB files
ies, mass media reporting, police services, or com- after the fall of the Soviet Union.
munities. In other words, they typically do not think
that a case study is a sample of one. Cases may also be chosen for mundane reasons,
such as convenience, and still provide an adequate
Types of case context for answering certain research questions or
Yin (1984) distinguished three types of case, each of examining key social processes. To take a concrete
which relates to the issue of external validity: example, Russell and Tyler (2002) studied a Girl
Heaven store in the UK (which caters to 3- to 13-year-
• The critical case. Here the researcher has a old girls) not because it was a critical or unique case,
clearly specified hypothesis, and a case is chosen or because it offered a context never before studied,
on the grounds that it will allow a better under- but because of its capacity to illuminate the links be-
standing of the circumstances under which the tween gender and consumption and the commodifi-
hypothesis does or does not hold. The classic cation of childhood in modern society. Indeed, often
study by Festinger et al. (1956) of a dooms- it is only at a very late stage in the research that the
day/UFO cult is an example. The fact that the singularity and significance of the case becomes ap-
world did not end allowed the researchers to parent (Radley & Chamberlain, 2001).
test propositions about how people respond to As we have mentioned, one of the standard criti-
thwarted expectations. What did cult members cisms of the case study is that its findings cannot be
do when, after quitting their jobs, leaving their generalized. Case study researchers argue strenu-
homes, and waiting on a mountaintop, nothing ously that this is not the purpose of their craft.
happened? Did they sneak down and move to A valid picture of one case is more valuable than a
another town? No: they decided that their faith potentially less valid picture of many. Their aim is to
had saved humankind and that their new role generate an intensive examination of a single case,
was to tell others of that miracle so more people which may or may not be used for theoretical analy-
could be converted to their religion. sis. Pratt and Valverde (2002) studied only Somalis
• The extreme, even unique, case; a common focus and expressed a hope that others would study other
in clinical studies. Margaret Mead’s (1928) well- immigrant groups in other places. Their central con-
known (albeit controversial) study of growing cern was the richness of the data and the quality of
46 PART I Fundamental Issues in Social Research
the theoretical reasoning that the case allowed. As approach is associated with a qualitative research
we have noted, sometimes case studies are primarily strategy. As illustrated in Box 2.6, comparative case
inductive, used as information to generate theories. study work is often quantitative in orientation.
Other times they may be deductive in nature, pro- The strength of comparative designs is that
viding the data required to assess theories. they highlight the similarities and differences be-
Problems can arise when the research involves tween cases, which can be used to assess or gener-
comparison of two or more cases. Dyer and Wilkins ate t heories. They exhibit certain features similar to
(1991), for example, argued that a m ultiple–case experiments and quasi-experiments, which also rely
study approach means that less attention is paid to on the capacity to establish comparisons.
the specific details of a particular case, and more to
the ways in which multiple cases can be contrasted.
Furthermore, the need for comparison often leads
Bringing research orientation
the researcher to choose an explicit focus at the and design together
outset, when it might be advantageous to adopt Finally, we can bring together the two general re-
a more open-ended approach. A preference for search orientations covered in Chapter 1 with the
contextual insight and a less structured research research designs outlined in this chapter. Table 2.1
Phenomena such as voting behaviour or crime cities may be very different in terms of ridership,
victimization in two or more countries can be safety, cleanliness, and so on.
compared using the same research instruments, A strength of cross-cultural research is that it
seeking similarities and differences and a deeper helps to illustrate how social scientific findings may
understanding of social reality in different na- be culturally specific. For example, Wilson’s (2002)
tional contexts. At the very least such research examination of Ontario raves made frequent com-
supplies a replication. parisons to the earlier rave scene in Britain. How-
Cross-cultural research is more expensive than ever, the UK scene was primarily an outgrowth of
other approaches. It also presents other problems. working-class struggles, whereas Canadian raves
When using existing data such as official statistics appealed more to middle-class, culturally alien-
or survey evidence, the researcher must ensure ated youths. Similarly, Baer et al. (2001) found
that the variable categories and data-collection that joining clubs and voluntary organizations in-
methods are comparable. When new data are creased toward the end of the last century in the
being collected, the researcher must ensure that US, West Germany, and the Netherlands, was stable
data-collection instruments (for example, ques- in Canada and 10 other countries, but decreased in
tionnaires and interview schedules) are translated Spain. Finally, Young and Dugas (2012) compared
properly. Even when translation is carried out Canadian print media coverage of climate change
competently, there may still be a problem with issues in English-language publications with those
insensitivity to specific national and cultural mi- in written in French, and found that the different
lieus. For example, the London “tube,” the Toronto environmental and media cultures in the two lan-
“subway,” and the Montreal “métro” differ in more guage communities contributed to important differ-
than name: public transit experiences in the three ences in how climate change topics were narrated.
2 | Research Designs 47
shows the typical form associated with each com- studies, research that concentrates on a specific
bination of research orientation and research design, issue over time, and ethnography, in which the
along with a number of examples that either have researcher charts change in a single case, contain
been encountered so far or will be covered in later elements of both designs. Such studies are per-
chapters. Table 2.1 also refers to research methods to haps better conceptualized as longitudinal case
be introduced in later chapters, but not referred to studies rather than as belonging to one category
so far. The Glossary provides a quick reference for or another. A further point is that there is no typ-
unfamiliar terms. ical form in the qualitative research orientation/
The distinctions are not always perfect. In experimental research design cell. Qualitative re-
particular, in some qualitative research it is not search in the context of true experiments is very
obvious whether a study is an example of a longi- unusual; a quasi-experimental design is a more
tudinal design or a case study design. Life history realistic alternative.
Key Points
• There is an important distinction between a gen- • Replicability, validity (measurement and external),
eral research orientation (quantitative versus and the ability to establish causation are import-
qualitative) and a research design. ant criteria for evaluating the quality of quantita-
• The nomothetic approach to explanation involves tive social research.
discovering general laws and principles. • Four key research designs are experimental,
• Nomothetic explanations must satisfy three cri- cross-sectional, longitudinal, and case study.
teria of causation: correlation, time order, and • Threats to the establishment of causation are of
non-spuriousness. particular importance in non-experimental, quan-
• Qualitative researchers usually take the idio- titative research.
graphic approach to explanation, which entails • External validity is a concern with case studies
creating a rich description of a person or group (generalizability) and laboratory experiments
based on the perceptions and feelings of the (findings may not be applicable outside the re-
people studied. search environment).
Relevant Websites
The National Longitudinal Survey of Children and The Human Genome Project site offers some basic
Youth site provides information on an ongoing, longi- information on this 13-year project and its implica-
tudinal study in Canada. tions for understanding human behaviour.
www12.statcan.gc.ca/census-recensement/2011/ https://web.ornl.gov/sci/techresources/Human_
ref/92-135/surveys-enquetes/nationalchildren- Genome/index.shtml
nationaleenfants-eng.cfm
The Behavior Genetics Association, an international
The Online Dictionary of the Social Sciences site, organization, provides a forum for the dissemination
maintained by Athabasca University, “has 1000 entries of behavioural genetic research and offers methodo-
covering the disciplines of sociology, criminology, pol- logical workshops.
itical science and women’s study with a commitment www.bga.org
to Canadian examples and events and names.”
(Websites accessed 16 October 2018)
http://bitbucket.icaap.org/dict.pl
Chapter Overview
Before research on human subjects can begin, researchers must demonstrate that what they are about
to do is ethical. This is no small task, given that ethical considerations pertain to all research meth-
ods and come into play at every stage of research, from the recruitment of subjects to the publishing
of the results. This chapter examines some general ethical principles and their applications to real re-
search situations. The backdrop for the discussion is the Tri-Council Policy Statement: Ethical Conduct for
Research Involving Humans (TCPS2) (2014), which governs most research practices in Canada. The TCPS2
was prepared by the Canadian Institutes of Health Research (CIHR), the Natural Sciences and Engineer-
ing Council of Canada (NSERC), and the Social Sciences and Humanities Research Council of Canada
(SSHRCC). It identifies three core principles, which are complementary and mutually reinforcing:
• Respect for persons stipulates that people’s autonomy and their freedom to choose what will
happen to them must be ensured. Research participants must therefore be made aware of what
the study will entail, so that their consent is not only freely given but fully informed. Consent must
also be ongoing; thus participants should be allowed to withdraw from the study at any time.
• Concern for welfare covers all aspects of a person’s life as well as the welfare of groups and com-
munities. To protect the welfare of individuals and groups, a favourable balance between the risks
and benefits of research must be sought.
• Justice requires that people be treated fairly and equitably, so that no segment of the population
bears an undue burden of the risk associated with research, and no part of society is excluded from
the benefits that research may bestow.
Each of these principles provides valuable guidance and can be used to identify research activities and
practices that should be avoided for moral reasons.
You may be surprised to learn that there are strin- who take pleasure in abusing the volunteers who partici-
gent formal policies in place to ensure that research in pate in their studies. While it is true that those who do re-
Canada is conducted in an ethical manner. Why, you ask? search are generally honourable people, every research
Surely researchers are decent people, not mad scientists situation has an ethical dimension to it, and sometimes
▲ AP Photo/Jae C. Hong
3 | Research Ethics 51
the ethical implications of a procedure are not obvious. quite acceptable when they were done, and it was only
In fact the history of the social sciences is riddled with with the passage of time that the consequences for par-
studies in which well-meaning researchers harmed their ticipants were recognized. How is that possible? And why
subjects in some way. In many cases the studies seemed are ethical dilemmas still endemic to social research?
(which in the case of social research is usually a uni- have contact with the drug users, a move that would
versity). If the institution with which the research- have made it all but impossible to conduct the study.
ers are affiliated administers research funds issued Indeed, some researchers claim that the advent of
by any of the three agencies that created the TCPS2 REBs has had serious, deleterious consequences for
(CIHR , NSERC , & SSHRCC), then its REB must abide by the social sciences as a whole (e.g., van den Hoonaard
the TCPS2 in making its judgments. What this means & Tolich, 2014). The fact that TCPS2 acknowledges
in practice is that REBs at virtually every university the right of researchers to appeal REB decisions is a
in Canada use the TCPS2 in making their decisions tacit admission that at times they can be overly re-
and recommendations. strictive. In any event, a strong case can be made that
REBs may approve a research project, request REBs perform a valuable function in protecting the
modifications, or reject it outright. If the research is rights of research participants, as the discussion to
to be conducted over an extended period of time, the follow will show.
REB may also hold periodic reviews, or at least require
that the researchers file an annual report. In addi-
tion, REB members themselves are required to follow
General ethical principles
procedures outlined in the TCPS2. For example, they The first thing to note regarding general principles
must avoid any conflict of interest, such as having a is that researchers themselves disagree on the ethical
financial stake in a project that is up for approval. acceptability of certain research practices. There are
Van den Hoonaard (2001) has suggested that disputes from the 1960s that are still being rehashed
it may be more difficult to get ethics approval for today. Also, although the TCPS2’s codes and guidelines
qualitative research than for quantitative studies. provide some direction, they often allow for a degree of
Qualitative research is sometimes considered less flexibility. Nonetheless, there is little disagreement that
scientific than the quantitative kind, and ethics com- recognizing the ethical implications of doing social re-
mittees may prefer the epistemology of the latter, search is crucial if informed decisions are to be made.
with its derived hypotheses and specific plans. This The TCPS2 outlines three core principles for the
can be a fatal problem, because if the REB rejects conduct of research involving human subjects: re-
the research proposal, the work can neither receive spect for persons, concern for welfare, and justice.
funding nor be carried out. Also, whereas in quanti- Although they are listed separately, there is consider-
tative research data are usually collected one-on-one, able overlap between them.
in qualitative research a whole group may be under
observation, including some people who might not Respect for persons
want the attention. One REB allowed a researcher to The principle of respect for persons maintains that
go ahead with her ethnographic work but required humans should not be treated as mere “objects” or
that she turn her head when someone who had not means to an end. Even if the end is a worthy one,
explicitly agreed to be part of the study entered her it cannot justify the use of unethical means. As we
field of vision. This directive is an example of what noted above, even if the knowledge a researcher is
some researchers would consider an overzealous at- seeking will be of benefit to all humankind, the wel-
tempt to prevent harm. fare of the individuals and groups under study is of
Another example of possible REB overreach in- paramount importance. Researchers must recog-
volving ethnographic research occurred in Van- nize that people have rights, including the right to
couver, where researchers sought permission to do be treated with dignity by social researchers. Respect
observational fieldwork and in-depth interviews for persons is arguably the most fundamental of all
with intravenous drug users (Small et al., 2014; see the principles considered here.
Box 16.2). The ethnographers were told by their REB, The starting point of respect for persons in re-
which cited safety concerns, that they had to provide search is the stipulation that subjects must give free,
police with the names of the observers who would informed, and ongoing consent to participate in the
3 | Research Ethics 53
study. Most researchers now adhere to the principle It also typically contains assurances of confidential-
that potential research subjects should not only be ity along with a description of how the data will be
asked for their permission before they are placed in stored and the form in which the results are to be
a study, but should also be informed of all the risks published. A sample from a research study in which
and benefits to which they would be exposed. To one of the authors of this book participated is shown
assess those risks and benefits, prospective partici- in Figure 3.1.
pants need a basic idea of what the study will entail. Prospective subjects should be encouraged to
In most quantitative studies and in some forms think about the project; they should also be given a
of qualitative research, potential participants are chance to ask questions and seek clarification if ne-
given an information sheet or letter of information cessary. If they have all the information required and
outlining what the research is about, how it will be are willing to participate, a consent form is sometimes
conducted, and what the risks and benefits of par- provided for them to sign. This form may include a
ticipation are. The document normally includes the brief synopsis of the study and state explicitly that re-
names of the people in charge of the research, their search subjects are free to leave the study at any time
institutional affiliation, and contact information. without penalty. Consent must be freely given and
cannot be coerced. Sometimes the consent form and want the participants to behave naturally and au-
the information document are merged into one, as thentically. If participants were to be fully informed
illustrated in Figure 3.1. A consent form taken from a regarding the purpose of the experiment, it would
Canadian study involving qualitative interviews con- be difficult to avoid the reactive effects that occur
ducted by university students is shown in Figure 3.2. when people know they are being observed. For this
In some forms of qualitative research, such as reason, experiments usually involve some deception.
ethnography and participant observation, provid- For example, Milgram (1963) designed a series of
ing information sheets and consent forms may be studies in an effort to understand the brutality that
impractical; in some cases it might even defeat the occurred in Nazi concentration camps during the
purposes of the study. In cases where a formal agree- Second World War. In particular, he wanted to know
ment between the researchers and the informants how a person could be led to cause extreme harm to
might signal a lack of trust, less formal approaches an innocent human being, and whether being or-
such as verbal assurances and ad hoc explanations dered to do so by an authority figure had anything to
may be employed. Whichever method they choose, do with it. Milgram devised a laboratory experiment
researchers must provide REBs with documentation in which a supposedly reputable researcher (the au-
outlining how informed consent will be obtained. thority figure) asked participants to act as “teachers”
The principle of free and informed consent seems who would punish “learners” by giving them increas-
straightforward, but putting it into practice can be ingly severe electric shocks whenever they gave in-
difficult and usually involves some ethical trade- correct answers to questions. Neither the shocks nor
offs. Homan (1991, p. 73) observed that abiding by the howls of pain coming from the “learners” were
this principle “is easier said than done.” At least two real, but the “teachers”/subjects did not know this.
major points stand out here: Some of them were further disadvantaged in that
they could not see the “learners”: they could only hear
• It’s extremely difficult to give prospective par- the shrieks of pain supposedly coming from them.
ticipants absolutely all the information re- In all instances, the “teachers” were told that the
quired to make an informed decision about shocks were part of the study and would not cause
their involvement. Relatively minor transgres- any permanent harm, despite the increasingly shrill
sions of the principle of informed consent are cries of pain. The experiment continued until the
pervasive in social research: for example, not “teacher”/subject refused to administer more shocks.
giving all the details about the study for fear of In the version of the study in which the “teachers”
contaminating people’s answers to questions. could not see the “learners,” a substantial majority
On the other hand, Hessler et al.’s qualitative (62 per cent) of the former administered the strong-
study of adolescent risk behaviour (2003) in- est shocks possible. Further, in one variant of the
cluded a consent form that was quite elaborate. experiment, the subjects were given the option of
• In ethnographic research, the researcher is ordering another person (actually an accomplice of
likely to come into contact with a wide spec- Milgram’s) to administer the shocks; in this situ-
trum of people, and ensuring that absolutely ation, more than 90 per cent of them ordered the
everyone has the opportunity to give informed greatest shocks possible. Many of the participants
consent is not practicable. Even when all par- were visibly upset as they carried out the experi-
ticipants in a certain setting are aware that they menter’s orders, but they still did what the authority
are being observed, it’s unlikely that all of them figure told them to do.
will have been similarly (let alone identically) Milgram saw these results as shedding light on
informed about the nature of the research. the circumstances that led to the horrors of the
Nazi concentration camps. It appeared that ordin-
Experiments are particularly problematic when ary, everyday people were willing to inflict serious
it comes to informed consent because researchers harm on innocent others if ordered to do so by an
3 | Research Ethics 55
authority figure—a finding that may have important shown that empirical, theoretical, or technological
implications for the understanding of human atroci- expertise is no guarantee of moral wisdom. But do
ties and ultimately their prevention. Nonetheless, listen carefully to the arguments of experienced re-
the study involved a number of ethically question- searchers. It is incumbent on you to consider what
able elements. For one thing, the “teacher”/subjects is at stake in any research situation, especially the
themselves suffered at least short-term harm; most probable harms and benefits to the research sub-
felt very uncomfortable administering what they jects, and to think about the effects—both good and
believed to be powerful shocks, and while the ex- ill—that the knowledge produced by the study would
periment was in progress many participants came generate for the world at large.
to doubt the experimenter’s claim that the “learners” Deception can be a controversial issue in quali-
were not suffering any serious harm. tative research as well. Punch (1994, p. 91), for
Another issue was the deception that the subjects example, observed that “some dissimulation is in-
experienced: they were told that the experiment was trinsic to social life and, therefore, to fieldwork.”
investigating how people learn, not how people come He quoted Gans (1962, p. 44) in support: “If the re-
to commit cruel acts, and they were led to believe that searcher is completely honest with people about his
the shocks were real. Thus to the extent that the par- activities, they will try to hide actions and attitudes
ticipants were not fully informed about the purpose they consider undesirable, and so will be dishonest.
of the experiment, they were lied to, although once it Consequently, the researcher must be dishonest to
was over they were debriefed (informed of the decep- get honest data.”
tion and the reason for it). Most people would agree Goode (1996) argued that judgments about de-
that in everyday life, lying and deception should be ception should be considered on a case-by-case
avoided. On the other hand, the experiment would basis. Some writers argue that if all ruses were dis-
not have worked had the subjects known the truth allowed, a great deal of useful information—about
from the outset. Imagine the outcome if subjects had terrorists, cults, drug gangs, white-collar criminals,
been told, “We’re going to see whether you are willing and so on—would be inaccessible, because finding
to inflict severe pain on an innocent person.” But does out about their activities often requires some kind
that justify the deception? Was the knowledge gained of disguised observation. Some social reformers
worth the psychological discomfort the subjects ex- might add that exposing the sins of the powerful
perienced when they believed they were causing in- would be another acceptable use of deception, but
nocent persons to suffer? This is the sort of question that argument opens the door to its use in other
that researchers and REB members must ponder. situations. Would it be acceptable to use deception
It must be emphasized that Milgram is not alone to expose the transgressions of the underclass and
in his use of deception: in fact, most social research the less powerful? Is lying to a priest to uncover
experiments involve some kind of pretense. How- pedophilia more acceptable than lying to an animal
ever, it is unlikely that an REB in North America rights activist to reveal a plan to destroy a mink
would allow a similar study to proceed today. For coat factory?
that reason Milgram’s experiment has rarely been Sometimes it may be beyond the power of a quali-
replicated, except in other parts of the world where tative researcher to obtain informed consent. For ex-
research ethics codes are less stringent. ample, Van Maanen (1991) used his experience as a
As you think about the ethics of Milgram’s study, ride operator in Disneyland as the basis for a study
keep in mind that REBs are not infallible and that he produced many years after he had been employed
REB members sometimes disagree among themselves there during school vacations. He never went back
about the ethics of a particular study. As a student, and asked permission from the people he described
you need to make up your own mind about ethical in his work; to do so would have been close to impos-
issues. Don’t feel obliged to accept the ethical claims sible. Is a lack of subject consent acceptable in such a
of more experienced social researchers; history has situation? Is it a form of deception?
3 | Research Ethics 57
Although the use of deception can be controver- especially if the study involves children who are very
sial in both quantitative and qualitative research, young. Exceptions include situations in which older
there is consensus on a number of criteria regarding children fill out questionnaires that the parents or
its use. First, deception should be used only as a last guardians have seen and approved.
resort. Second, if deception is used, that use should be
as sparing as possible. Finally, where feasible, anyone Concern for welfare
deceived should be debriefed as quickly as possible; “Welfare” in this context covers all aspects of the
specifically, the researcher must reveal the deception well-being of the person, group, or community af-
and explain why it was used. What might have hap- fected by the research, including physical, mental,
pened had Milgram’s subjects not been debriefed? emotional, spiritual, social, and economic well-being.
Another issue regarding informed consent is A very basic “welfare” concern in social research in-
whether the researcher is obliged to reveal the names volves respect for privacy and confidentiality. REBs
of all those who have sponsored the project, whether normally require that the identities and records of
individuals or organizations. At one Ontario univer- subjects remain confidential to prevent any embar-
sity, ethical rules prohibit any reference to the names rassment or harm, especially harm to their reputations
of research sponsors on the information sheet. This or personal relationships. This injunction means that
rule reflects the possibility that some people might feel the research subjects should not be identifiable when
pressured or coerced if they knew that the research the findings are published, although exceptions may
was sponsored by (for example) the Canadian Cancer occasionally be made in the case of historical work or
Society. But another university considers it unethical of studies dealing with public figures. In quantitative
not to supply that information, on the grounds that it research, subjects are usually identified by code num-
is necessary to enable the participant to give informed bers rather than names when the data are processed;
consent. For instance, some people might not want their actual identities are stored in a secure location.
the government to know anything about them, so if In such studies research participants are normally
a department such as Fisheries and Oceans Canada told on the information sheet that the data they pro-
were funding the study, they would want to be in- vide will form part of a statistical aggregate (as in
formed so they could decide not to participate. Yet “50 per cent of respondents expressed some dissatis-
even if the sponsor’s name is revealed, participants’ faction with their romantic partner’s listening skills”)
ability to give informed consent may not be guaran- and therefore they will not be personally identifiable.
teed, since the name may be virtually unknown to the Researchers have to be especially resourceful in
general public. Ethical issues are never simple. situations where participants prefer to remain an-
Only people competent to make a decision about onymous. One method that is sometimes employed
participating in a research study should be asked to is the random response technique, which was origin-
give their consent. That category normally includes ally designed for interview situations in which the
most adults, although for reasons of illness, other respondent is asked about controversial or illegal
incapacitation, or language difficulties some adults activities. One reason for developing this technique
may not be capable of making an informed decision was to protect researchers as well as respondents
and must have a competent person make the decision from the risk that police would examine the data and
on their behalf. Similarly, children are not ordinar- then bring the researchers into court as witnesses.
ily considered capable of giving informed consent. (Can some harm be anticipated here, as when the
Parents or legal guardians are usually given the still-free criminals repeat their crimes?)
information sheet as well as an opportunity to ask The random response process begins with
questions and seek clarifications about the research respondents flipping a coin but not revealing the
before signing the consent form for their child. results. About half of them should get heads, half
Quite often a parent or guardian is required to be tails. All respondents are then instructed: “For the
present when the child interacts with the researcher, following question, if you have heads you must say
58 PART I Fundamental Issues in Social Research
Michael Simons/123RF
cases the researcher is not required to obtain in-
formed consent, although he or she must obey the
relevant laws and regulations. Another exception
applies to online information that is publicly ac-
cessible and does not involve any expectation of
As a researcher, consider how you could conduct
privacy. Such material would include films, news-
online research in an ethical way. What kinds of steps
paper materials, official publications, third-party
would you need to take to secure the informed con-
interviews, and any other online research that does sent of your participants and to protect your data?
not entail direct interaction with research subjects.
Any situation that does involve direct contact
with research participants, including online sur- be quite easy, for example, for a researcher to pose
veys and interviews, is subject to the same ethics as a regular participant in a self-help chat room
protocols as in-person research. The same goes for sex addicts, and use the material as data for a
for gaining access to online materials in situa- qualitative study. But doing so would be equiva-
tions where there is a reasonable expectation of lent to joining an in-person discussion group and
privacy, as in the case of social media that may not revealing one’s identity as a researcher. (The
limit access to materials unless the creator’s per- ethical implications of covert research will be dis-
mission is granted, such as Facebook, Instagram, cussed later in this chapter.)
and LinkedIn. Internet chat room discussions, Sending research data over the Internet is an-
email, and the postings of organizations having other concern. If data containing identifiable in-
restricted memberships, such as self-help groups, formation are to be transferred via the Internet,
fall into this category as well. In these sorts of situ- the data should be encrypted to ensure the sec-
ations, both researchers and research participants urity of the information. Even when sensitive data
have to be especially vigilant, since it is relatively are merely stored on a computer that is connected
easy to conduct covert research online. It would to the Internet, encryption should be considered.
yes, regardless of whether it is true or not. With tails used cocaine. Here the excess is 10 per cent, presum-
you should answer the question truthfully.” Then ably representing people who got tails and have used
comes the question: “Have you ever used cocaine?” the drug. Then, since one would expect roughly the
Even if no one had in fact used it, roughly 50 per cent same number of cocaine users in the heads group,
should say yes because they got heads. that figure must be doubled to get the percentage of
Assume that 60 per cent of respondents say yes. the total sample who had used cocaine.
The excess over the 50 per cent (heads who had to One of the authors of this book did this as a
say yes) can be used to calculate how many people class exercise and had students raise their hands to
3 | Research Ethics 59
say yes or no. When the behaviour in question was that the benefits of the research outweigh any harm
shoplifting after the age of 16, 80 per cent answered that may come to the subjects. Virtually all codes of
“yes,” meaning a 60 per cent (30 per cent plus an- ethics allow covert research as a last resort, but most
other 30 per cent) shoplifting rate. When the topic state that it should be avoided “as far as possible.”
was post-puberty homosexual behaviour, however, Those who use this method argue that it is neces-
“yes” responses came out at only 50 per cent. Here sary to prevent research participants from changing
there are several possibilities: no one had engaged in their behaviour because they know they are being
the activity; a few had, but because heads and tails studied. Still, where informed consent has not been
rarely come out exactly 50–50, they were concealed obtained prior to the commencement of the re-
in the difference; some “heads” students were afraid search, permission to use the data gathered should
their classmates would think a raised hand was an be obtained post hoc whenever possible. That is a
admission; or perhaps some “tails” students who had general rule, but it is often broken in dealing with
engaged in the activity did not want to give any indi- dangerous subjects such as pimps and other crimin-
cation that they had, fearing that a “yes” response als. Where such permission has not been obtained,
would be interpreted as an admission. it is absolutely crucial to maintain the anonymity of
In qualitative research, dealing with privacy and the people observed.
confidentiality issues is quite different, especially One difficulty with the arguments justifying the
if the study involves in-depth analysis of a small use of covert methods is the assumption that it is im-
number of people. Pseudonyms are typically used possible to obtain the data using other means. Covert
to protect people’s identities, but quite often the observers sometimes base their judgments on the
detailed description of the physical and social set- anticipated difficulty of gaining access to a setting
ting, which qualitative researchers usually include rather than the actual experience of being denied
to provide the needed context for the study, can entry. For example, Homan justified his use of covert
offer enough clues to identify some of the partici- participant observation of a religious sect on the
pants. The study of an American town that Vidich grounds that sociologists are viewed very negatively
and Bensman (1968) called “Springdale” (not its real by group members; hence “It seemed probable that
name) is instructive in this regard. Their book was the prevalence of such a perception would prejudice
uncomplimentary about the town and many of its the effectiveness of a fieldworker declaring an iden-
leaders, and many people felt that its tone was pa- tity as sociologist” (Homan & Bulmer 1982, p. 107).
tronizing. To make matters worse, it was possible to One of the most famous and controversial cases of
identify individuals in the published account. The covert research can be found in Humphreys’s Tearoom
town’s inhabitants responded with a Fourth of July Trade (1970), a study that Desroches (1990) partially
parade in which many of them wore badges citing replicated in Canada using less contentious methods.
their book pseudonyms, and an effigy of Vidich was The researchers (Humphreys himself and police acting
set up so that it was peering into manure. The towns- for Desroches) observed homosexual encounters in
people also announced their refusal to cooperate in public toilets (“tearooms”), taking the roles of “watch
any more social research; they were clearly upset by queens”—people who watch out for possible intruders
the publication, and to that extent were harmed by it. while other men meet and engage in sexual activity.
Another area of controversy surrounding privacy Such “voyeurism” was offensive to some critics.
and confidentiality in qualitative studies pertains to As part of his research, Humphreys recorded some
covert research: investigations in which the people of the participants’ car licence numbers. He was then
being observed are not informed that they are part able to track down their names and addresses, thus
of a study. In spite of the widespread insistence on further compromising their privacy, and ended
informed consent and privacy rights, and the rec- up with a sample of 100 active tearoom-trade par-
ognition that covert observation is especially prone ticipants. To reduce the risk of being recognized,
to violations of those principles, covert studies still Humphreys waited a year before contacting the re-
appear from time to time. The defence is usually spondents whose licence numbers he had written
60 PART I Fundamental Issues in Social Research
down, and also changed his hairstyle. After this de- The ethical quandaries surrounding privacy
ception, he then conducted an interview (Desroches and confidentiality inherent in qualitative research
did not do this) with a sample of the men about their are also illustrated in the work of O’Connell
health issues (a ruse), including some questions Davidson and Layder (1994). These researchers
about marital sex. He did not tell them he knew about conducted a small-scale ethnographic project on a
their activity, nor did he debrief them when the study sex worker, whom they called Desiree, and her cli-
was finished. Humphreys learned that many of the ents and assistants. While Desiree and the women
men he observed were actually married, respected who served as her receptionists were fully aware of
members of the community—a finding that allowed O’Connell Davidson’s status, the clients were not.
him to challenge a popular stereotype of the day. Both researchers admitted their invasion of client
Was his research ethical? Those who say “yes” privacy and failure to obtain consent from them,
argue that he had no choice but to do covert re- but were “untroubled” by the intrusion because the
search because very few people engaging in gay sex clients were anonymous to O’Connell D avidson
in those days were willing to talk about it; in fact, and she was not “in a position to secure, store
sexual acts between men were against the law and or disclose information that could harm them”
the people committing them were subject to prison (O’Connell Davidson & Layder, 1994, p. 214). The
sentences. Humphreys’ supporters also maintain fact that there was no harm to participants was re-
that the knowledge his studies produced led to a garded as the litmus test of the ethical status of the
better understanding of human sexuality and the research. They offered a further defence by saying
treatment to which stigmatized individuals are that ethical transgression is pervasive: “Virtually
subjected. His detractors argue that the knowledge all social research is intrusive and exploitative
gained did not justify the deception and invasion to some degree” (O’Connell Davidson & Layder,
of privacy that his subjects experienced. The debate 1994, p. 215). There was an acknowledged commit-
continues to this day. ment to attend to the concerns of Desiree and her
Regardless of one’s position on the ethics of Hum- receptionists, as well as considerable sympathy for
phreys’s work, the public reaction when it was first them, but these sentiments did not extend to her
published illustrates what can happen to the reputa- clients:
tion of social researchers as a whole if there is a per-
I have . . . no personal liking and no real
ception that one of their kind has committed serious
sympathy for them. I have a professional
ethical violations. The following is an excerpt from
obligation to preserve their anonymity and
an article written in the Washington Post in 1970,
to ensure that they are not harmed by my re-
shortly after the controversy erupted:
search, but I feel no qualms about being less
than frank with them, and no obligation to
We’re so preoccupied with defending our pri-
allow them to choose whether or not their
vacy against insurance investigators, dope
actions are recorded. (O’Connell Davidson &
sleuths, counterespionage men, divorce de-
Layder, 1994, p. 215)
tectives and credit checkers, that we overlook
the social scientists behind the hunting blinds Although an effort was made to follow ethics
who’re also peeping into what we thought were protocols, it appears that not all the participants re-
our most private and secret lives. But they are ceived the same level of ethical protection, and that
there, studying us, taking notes, getting to the researchers believed that not all research sub-
know us, as indifferent as everybody else to the jects deserved the complete range of ethics cover-
feeling that to be a complete human involves age. This sort of arrangement is more common than
having an aspect of ourselves that’s unknown. one might think and has far-reaching implications.
(Von Hoffman, 1970) It also illustrates how difficult it can be to resolve
3 | Research Ethics 61
ethical dilemmas in social research. Do sex-trade research. Ethical issues arise if the sponsor places
clients have a right to choose to be excluded from conditions on the methodology to be used or the
research studies? What if they are abusing the publication of the results, or has certain expecta-
sex workers? What if they are not? Does it matter tions regarding the findings. In the United States,
whether such studies contribute to a better under- a social researcher was asked by the owner of an
standing of the sex industry—for instance, by re- apartment building to do a study of the behaviour
vealing the physical risks that people in the sex trade of tenants in “singles” apartments after neighbours
take? Such questions must be weighed carefully and complained about bad behaviour (Bailey, 1994).
are not easily answered. The sponsor in this case made it clear that the re-
Sometimes a researcher may feel obliged to dis- search was expected to show that the tenants were
regard concerns about privacy and confidentiality in fact good people and desirable neighbours. Per-
if the research uncovers evidence of serious harm haps the most famous Canadian case of a sponsor
or criminal activity. In his ethnographic research, trying to influence the research process involved
Totten (2001) wanted to know how youth gang Dr Nancy Olivieri, who took part in research on a
members interpreted the violence they perpetrated drug designed to treat thalassaemia, a blood dis-
against their girlfriends, and how they accounted for ease. She claimed that the drug was not effective
their racist and homophobic activities. One under- and had harmful effects, and wanted to inform the
lying factor he observed was their mental construc- research participants. A controversy arose when
tion of masculinity, which included the belief that the pharmaceutical company that sponsored the
such activities are acceptable for men. Most had lived research denied her claims and cited a confiden-
on the street, were themselves victims of severe child tiality agreement that she had signed that would
abuse and neglect, and had witnessed their mothers be violated if she were to contact the participants.
being beaten. They saw their own violence as af- Another famous case involved three Canadian
firming their masculinity. scientists who claimed that their supervisors at
That was the sociology, but what about the wel- Health Canada had tried to coerce them into ap-
fare of the people concerned? Here the researcher proving certain drugs for use on cows and pigs
was in some instances compelled to act. By law, he even though the scientists objected that the drugs
had to report any abuse suffered by children under could have harmful effects on humans; the scien-
the age of 16, regardless of whether they were gang tists were subsequently fired, although one was
members or their victims. Second, those deemed at later reinstated. In 2010, the head of Statistics
risk of suicide had to be referred to mental health– Canada resigned over the federal government’s
care providers. Finally, some of the gang members decision to discontinue the mandatory long-form
were on probation or living in treatment facilities, census and replace it with a voluntary version,
so the relevant authorities also had to be informed. which in his view would not be adequate to fulfill
Totten handled the confidentiality issue by studying Statistics Canada’s mandate.
only the boys who agreed to those limitations on People conducting social research should be
confidentiality. mindful of the perils of sponsored research. The
Another aspect of the “concern for welfare” cri- TCPS2 states that financial considerations do not
terion, one that can affect the larger community reduce the need to abide by the principles that
or even the society as a whole, involves sponsored underlie its policies, and that the scientific integrity
research. Most social research in Canada is funded of the research process must be upheld. Any conflict
either by SSHRCC or by the university with which of interest, including a financial conflict of interest
the researcher is affiliated, but other government involving a relationship with a sponsor, must be re-
agencies, private individuals, corporations, unions, ported to the appropriate REB. Most fundamentally,
and non-profit organizations can also sponsor the public good, in particular the welfare of the
62 PART I Fundamental Issues in Social Research
research participants, “takes precedence over the in- or group should be exploited in the research pro-
terests of researchers and sponsors” (CIHR , NSERC , & cess or systematically excluded from its benefits.
SSHRCC , 2014, p. 151). This is a major concern in health research, in which
care has to be taken that poor or vulnerable people
Justice are not treated as guinea pigs for the benefit of
Justice in the context of research ethics includes others, whether some interest group or the larger
the idea that the burdens and benefits of research society. Similarly, medical researchers should
should be spread evenly across society. No person not arbitrarily exclude from their studies certain
3 | Research Ethics 63
categories of people (such as members of specific line. For example, in the Rosenthal and Jacobson
ethnic, gender, or age groups), especially if such (1968) study of teacher expectations discussed in
people stand to benefit from experimental drugs or Chapter 2, it is at least possible that the pupils not
other treatments. identified as “spurters” (students expected to excel
With regard to social research, inclusiveness is quickly) were adversely affected in their intellectual
usually not a pressing issue, partly because social development by the increased attention given to the
scientific studies lack the immediate life-and-death “spurters.” After all, teachers have limited time and
consequences that are common in medical research, energy, and time spent with the “spurters” meant
and also because vulnerable and exploited groups time away from everyone else. Was that a violation
are often sought out for study, particularly by soci- of research ethics? Was the harm suffered by the
ologists and anthropologists. However, there is “non-spurters” outweighed by the knowledge gained
always a risk that members of disadvantaged groups by the researchers? Similarly, in the Festinger et al.
will be abused, mistreated, or disrespected by the (1956) study of a doomsday/UFO cult, the research-
researchers who study them. For example, some ers joined the group at a crucial time (close to the
street people might resent an ethnographer’s im- projected end of the world), thereby deluding group
plication, however subtle, that they need “help” of members into thinking that they had succeeded in
some kind, or the assumption that the researcher recruiting new converts. Was that ethical? Was the
has a superior sort of wisdom or political insight. harm to the cult members less important than the
Some Western anthropologists have been barred understanding of cults gained through this research
from re-entering the countries where they did their project?
fieldwork because the locals found them to be intru- One of the problems with trying to balance harms
sive and disrespectful. and benefits is that it is not possible to identify all
Justice in a research context also includes the the circumstances in which harm is likely—although
principle of minimizing harm to participants. As that fact should not be taken to mean that there’s
we suggested above, it is a rule of thumb when harm no point in seeking to protect research subjects.
is expected that it must not outweigh the potential Kimmel (1988) noted this in connection with the
benefits of the research. But as the examples dis- 1939 C ambridge–Summerville Youth Study: an
cussed earlier have shown, harms and benefits can experiment conducted on 506 boys aged 5 to 13 who
be difficult to judge correctly, and assessments of were identified either as likely to become d elinquent
them ultimately involve value judgments as well or as having an average likelihood in that regard. The
as scientific decisions. Most social scientists are boys were randomly assigned either to an experi-
not curing cancer or ending racism. How much mental group in which they would receive preventive
suffering or harm, if any, can be justified in social counselling, or to a no-treatment control group. In
scientific research? That is a key ethical question the mid-1970s the boys (by then men) were exam-
facing both researchers and the REBs that oversee ined and the results were quite shocking:
their work.
Treated subjects were more likely than controls
Research likely to harm participants is regarded
to evidence signs of alcoholism and serious
by almost everyone as unacceptable. But what is
mental illness, died at a younger age, suffered
“harm”? Harm can mean a number of things, includ-
from more stress-related diseases, tended to be
ing physical harm, loss of self-esteem, stress, or even
employed in lower-prestige occupations, and
a mild sense of shame. And what does “likely” mean:
were more likely to commit second crimes.
a 1 per cent chance? 10 per cent? 50 per cent?
(Kimmel, 1988, p. 19)
Researchers should do their best to anticipate and
guard against consequences for research partici- In other words, the treatment itself may have
pants that are expected to be harmful or disturbing. had unintended negative consequences, although
But it’s not always easy to know where to draw the given the complex and long-term nature of the
64 PART I Fundamental Issues in Social Research
forces that cause delinquency it seems likely that humiliating smocks and headgear to wear, and had
non-experimental factors contributed to the results. to address the “guards” in a respectful manner re-
Nonetheless, the findings do illustrate the difficulty gardless of how they were being treated. The aim of
of anticipating harm to participants. Might a ques- the study was to see whether taking on the role of
tionnaire on marital happiness lead a respondent “guard” would create feelings of omnipotence and
to question and eventually leave a marriage? Could lead to belligerent behaviour, and whether being
asking a grade eight boy in gym class about steroid treated like a prisoner would produce feelings of sub-
use encourage him to experiment with a drug that servience, self-alienation, resentment, and hostility.
until then was unknown to him? Some interviewees In short, the aim of the study was to see whether
may be uncomfortable answering any question; placing ordinary people in these roles would lead to
some may find the cut and thrust of a focus group stereotypical prison behaviour.
discussion stressful, especially if they inadvertently The study was designed to last two weeks. Zim-
reveal more than they had intended. Does it follow bardo describes what happened:
that such studies should not be conducted? Is the
At the end of only six days we had to close
risk of harm, even mild harm, worth the potential
down our mock prison because what we saw
increase in knowledge that this kind of research
was frightening. It was no longer apparent to
may make possible? That is the crux of the issue
us or most of the subjects where they ended
when a researcher seeks a proper balance between
and their roles began. The majority had indeed
harms and benefits. Most ethical codes maintain
become “prisoners” or “guards” no longer able
that if there is any prospect of harm to participants,
to clearly differentiate between role-playing
and if the risks of the research are greater than the
and self. There were dramatic changes in vir-
risks of everyday life, informed consent is a min-
tually every aspect of their behavior, thinking
imum requirement.
and feeling; human values were suspended,
Clearly, the principle of justice in social research
self-concepts were challenged, and the ugli-
is closely tied in with the precepts discussed ear-
est, most base, pathological side of human
lier, in particular the notions of informed consent,
nature surfaced. We were horrified because we
privacy, and confidentiality. People are certainly
saw some boys (“guards”) treat other boys as
less likely to be harmed if they know what to expect
if they were despicable animals, taking pleas-
in a research situation, and if they know that their
ure in cruelty, while other boys (“prisoners”)
participation is voluntary. Thus it is particularly
became servile, dehumanized robots who
important to inform subjects of their right to with-
thought only of escape, of their own individ-
draw from the study at any time—and to honour
ual survival, and their mounting hatred of the
that right.
guards. (Zimbardo, 1971; quoted in Aronson,
A good illustration of what can go wrong when
1992, pp. 10–11; emphasis added)
subjects are pressured into remaining in a study
can be found in the Stanford mock prison A whole host of ethical issues come into play in
study (Haney, Banks, & Zimbardo, 1973). In this ex- this experiment, but let’s focus on the right to leave a
periment, psychologically healthy male university study at any time. Many of the “prisoners” wanted to
students were randomly selected to be either “guards” leave long before the study was terminated. Yet they
or “prisoners” in a simulated prison set up in the were given every indication that they had no choice
basement of a university building. “Guards” were but to stay in the study, and were convinced that for
given uniforms, wooden batons, and reflective sun- all practical purposes they were in a prison. A few
glasses (to avoid eye contact with the “prisoners” and were allowed to leave early, but only after showing
thus to dehumanize them), and were told to instill signs of extreme stress and even hysteria. It’s hard to
fear in the “prisoners.” The “prisoners” were given see how this was not a violation of research ethics.
3 | Research Ethics 65
Looking back on what he had created, Zimbardo medical research, so those sorts of questions rarely
himself was appalled. have to be considered.
Another point this experiment illustrates is Benefit in the social sciences is typically in-
the need for REB s. Like the Milgram (1963) study, direct; in most cases it follows from an increase in
Zimbardo’s research was conducted before the era knowledge. However, direct benefits can accrue, for
of compulsory ethics reviews. REB s are absolutely instance in cases where research findings inspire
necessary because even well-meaning researchers positive social reform or contribute to beneficial so-
like Zimbardo and Milgram may be completely cietal conservation. One way to maximize the bene-
unaware that their work raises ethical problems— fit derived from social research is to conduct it in a
even if those problems seem quite obvious to methodologically sound manner. It is considered un-
others. For example, it simply did not occur to ethical to put research subjects through the rigours
Zimbardo that his experiment was causing his re- of participation if the study is not designed or con-
search participants harm. It was not until a student ducted properly and therefore is of little or no value.
research assistant told him she thought the treat- As you will learn as you make your way through this
ment of the “prisoners” was awful that he relented book, researchers can disagree over what constitutes
and brought the study to an end. Presumably, an a sound methodology. Nonetheless, each methodo-
REB review would have spared the “prisoners” logical perspective assumes that there are optimal
considerable agony. ways of doing the research, and if these are ignored
Another issue related to harm to participants it’s doubtful that anyone, including the researcher,
concerns payment for participation. In general, it is will get the maximal benefit from it. Maximizing
considered unethical to offer payment if that would benefit is also related to the inclusiveness principle
lead subjects to take risks they would not otherwise discussed above: no group or individual should be
accept. Similarly, payment should not put undue arbitrarily excluded from research.
pressure on people to take part or remain in a study, Finally, justice in the conduct of social research
and for that reason should not exceed the sorts of includes fairness in the adjudication of research pro-
benefits to which subjects are accustomed. posals by REBs. The procedures and protocols that
The participants in the Dinovitzer et al. (2003) REBs use have to be reasonable and equitable, and
study of educational attainment among immigrants their decisions unbiased. In order to achieve those
(discussed in Chapters 1 and 2) were offered $5 to ends, REBs must be independent of the institutions
participate. It seems unlikely that anyone answered in which they function. This means, for example,
the questionnaire simply for the small amount of that universities cannot overturn the decisions of
money involved. Still, some REBs do not allow sub- their REBs.
jects to be paid for their participation, except to As we noted above, sometimes REBs can be over-
cover the costs of taking part (parking, etc.). The cautious in trying to protect the interests of research
purpose of this rule is to prevent poor people from participants. This is often a concern in survey re-
“selling themselves” to researchers. Few would be search, which normally involves minimal risk to
critical of Beagan’s (2001) offer to medical students participants. J. Paul Grayson, a sociologist at York
for a chance at a $200 draw (which technically got University, conducted a study in which he found
around the payment rule) for taking part in her evidence that the excessively formal and legalistic
study of discrimination. But would offering $10,000 letters of introduction that some universities re-
for testing a promising drug with potentially danger- quire researchers to send to prospective participants
ous side effects be acceptable? What about selling a lowers participation rates (Charbonneau, 2005). Low
kidney for research purposes? Fortunately, the risk response rates may result in sample sizes that are
of harm to participants in social scientific research too small for analysis and can lead to additional ex-
is usually not as severe as the risks associated with penses for follow-up.
66 PART I Fundamental Issues in Social Research
Key Points
The TCPS2 outlines three core principles for the eth- Most social research in Canada has to abide by
ical conduct of research: those principles. In most cases, before social re-
search can be conducted it has to be approved by a
• Respect for persons demands that research par-
Research Ethics Board (REB) that uses the TCPS2 to
ticipants be able to give free, informed, and on-
guide its decisions. It may be difficult to measure
going consent.
harms and benefits in social research, so the imple-
• Concern for welfare includes all aspects of the
well-being of individuals, groups, and commun- mentation of these principles is rarely simple and
ities affected by the research. straightforward.
• Justice covers all dimensions of fairness in the Researchers may disagree on how the three core
treatment of research subjects and researchers principles are to be implemented, but they still must
themselves. do everything they can to comply with them.
and Group 3 can choose either of the first two pos- 2. Using the classroom screen, the instructor can walk
itions. All groups are to justify their positions as the class through the TCPS2 tutorial available at
best they can, and are encouraged to allude to the www.pre.ethics.gc.ca/eng/education/tutorial-
material discussed in this chapter. Once all three didacticiel/ (accessed 18 October 2018). It
groups have argued their case, a vote is held to de- includes multidisciplinary examples, some inter-
termine how the class as a whole views the ethics of active features, audio and video downloads, and
the study, i.e. whether, on balance, they find it to be a quiz.
ethical or not.
Relevant Websites
The TCPS2 (CIHR , NSERC , & SSHRCC , 2014) is avail- This link leads to the Canadian Charter of Rights and
able online. This site includes a TCPS2 tutorial that Freedoms.
provides an excellent introduction to the document https://laws-lois.justice.gc.ca/eng/Const/page-15.
and gives users the opportunity to apply ethical prin- html
ciples to case studies involving research with human
Quiet Rage is a documentary film about the Stanford
subjects.
mock prison study.
www.pre.ethics.gc.ca/eng/policy-politique/
www.youtube.com/watch?v=yUZpB57PfHs
initiatives/tcps2-eptc2/Default/
(Websites accessed 18 October 2018)
Visit the Canadian Sociological Association website
to read its Code of Ethics.
www.csa-scs.ca/code-of-ethics
APPENDIX TO PART I
An Overview of the Research Process
You now have some notion of how and why social projects. In other cases, the process of inquiry may
research is done. Before moving on to the rest of the begin with simple curiosity: as one wag put it, if you
book, which expands on many of the points raised in walk by a house with its curtains closed and wonder
the previous three chapters, let’s step back and take a what goes on inside, you’d probably make a good
look at the big picture. social scientist.
What follows could be described as the “ideal” Not all research begins with lofty motivations.
stages of research. They are ideal not because they In fact, the reasons for conducting a study are often
represent the best way to do things, but rather to quite mundane. For example, research may be done
indicate that they are an abstraction, a simplification to enhance the career of the researcher, especially if
of a more complex reality. There is no single best way it is supported by substantial funding. Sometimes
to do research, any more than there is a single best researchers compete with one another to see who
way to build a house or paint a portrait. Sometimes can produce the most publications or get the most
the order of the tasks presented in the overview of attention for their work. More generally, research
the research process on p. 70 can be rearranged. may be motivated by “sheer egoism,” a reason that
Also, it’s not uncommon to go from one step to the topped the list of motivations for writing drawn up
next and then backtrack to adjust something in a by George Orwell (1946/2004). Graduate students
previous step. Nonetheless, some general points and (and some undergraduates) may do research simply
observations can be made that outline how research to fulfill degree requirements. In most instances, the
may progress from its initial stages to its completion. motivations for doing research are some combina-
tion of the exalted and the mundane.
Motivation
Research Question(s)
Literature Review
Quantitative Qualititative
(choose design) (choose design)
Experiments Ethnography
Surveys Participant observation
– Structured interviews Unstructured interviews
– Questionnaires Semi-structured interviews
Structured observation Life history
Case studies Oral history
Narrative analysis
Ethics review Focus groups
Content analysis
Gather data – Ethnographic
– Semiotics
– Hermeneutics
Analyze data
– Conversation analysis
– Discourse analysis
Findings
Ethics review
Conclusions
– Do the findings support the hypotheses?
– Implications for existing theories? Gather data
– New hypotheses, concepts, or theories
generated? Analyze data
– Implications for further research?
Findings
Conclusions
– Do the findings provide a new interpretation
of a person, group, setting, event, or text?
– Implications for existing theories?
– New hypotheses, concepts, or theories
generated?
– Implications for further research?
– Do the findings illustrate group processes
or social conflict in a new way?
APPENDIX TO PART I | An Overview of the Research Process 71
PART II
Quantitative Research
© kendo_OK/Shutterstock.com
4
The Nature of Quantitative Research
Chapter Overview
This chapter is concerned with quantitative research, one of the two principal investigative orientations
taken in the social sciences. The emphasis here is on what quantitative research typically entails, though
departures from this ideal type will be outlined later on. This chapter explores:
• the main steps of quantitative research, presented as a linear succession of stages;
• the importance of measurement in quantitative research, and the ways in which measures are
devised for concepts;
• procedures for checking the reliability and validity of those measures;
• the primary goals of quantitative researchers: measurement, establishing causality (sometimes
referred to as internal validity), generalization, and replication; and
• some criticisms of quantitative research.
For his class project, Mario wants to ask some fellow be offended by the question? And when they do answer,
students questions about a topic that’s always fascin- how will he know if they’re telling the truth? Is there a
ated him: drug and alcohol use, and how it might affect second method he could use, such as observing people
a person’s self-image. But how will he get the informa- in a local pub, to see if their behaviour is consistent with
tion he needs? He quickly jots down a few notes, but the survey data? And once he’s asked the basic ques-
soon realizes that coming up with effective questions tions about consumption, how will he approach the
is more difficult than he imagined. His first question self-image part of his research? What exactly should he
reads: “Do you take drugs?”; but he quickly realizes that be asking about? Should the information be recorded
“drugs” is too vague, since for some people that would in numeric form? If so, how do you do that? When he
include things like Aspirin and cough medicine. And has finished analyzing the data, will he be able to say
what about alcohol? Isn’t alcohol a kind of drug too? that his findings apply to some larger population, such
Now he begins to think about other potential problems. as all university students? In this chapter we will ex-
What if some people refuse to say what their level of plore how researchers deal with the kinds of issues that
drug and alcohol consumption is? Would some people Mario is facing.
▲ pidjoe/iStockphoto
4 | The Nature of Quantitative Research 75
restrict their study to boys old enough to give their the research? Do the findings support the theory? If
own permission. Finally, because of a less-than- not, should the theory be revised? Should it be aban-
eager response, all boys who were willing to partici- doned entirely?
pate became part of the project. In bigger studies, Then the research must be written up in step 11.
especially those in which the findings are expected Until it enters the public domain in some way, as a
to be applicable to some larger population, an elab- paper to be presented at a conference, a report to
orate sampling process (discussed in Chapter 7) may the agency that funded the research, or as a book
be used to select research participants, although this or journal article to be read by other researchers or
approach is rare in experimental research. students, it cannot have an impact beyond satisfy-
Step 7 is the administration of the research instru- ing the researcher’s personal curiosity. In writing up
ments. In experimental research, this usually means the findings and conclusions, researchers must do
pre-testing subjects, manipulating the independent more than simply relay their results to others: they
variable for the experimental group only, and must convince readers that the research conclusions
post-testing. In cross-sectional research using survey are important and the findings robust. A significant
research instruments, it involves interviewing the part of doing research is convincing others of the rel-
members of the sample with a structured interview evance and validity of one’s findings.
schedule (a list of questions or statements to which the Once the findings have been published they
person being interviewed responds) or distributing become part of the stock of knowledge in their
a self-completion questionnaire. In research using domain. Thus there is a feedback loop from step 11
structured observation, this step involves watch back to step 1. The presence of both an element of
ing the setting of interest and the people in it, and deduction (steps 1 and 2) and induction (the feed-
then recording the types of behaviours observed. back loop) is indicative of the positivist foundations
Step 8 simply refers to the fact that the informa- of quantitative research. The emphasis on the trans-
tion collected must be systematically recorded so lation of concepts into measures (step 4) is also a fea-
that it can be analyzed. With some information this ture of positivism, and it is to that important phase
can be done in a relatively straightforward way. For of the research that we now turn. As we will see, cer-
example, for information relating to matters such as tain considerations follow from the need for meas-
people’s ages, incomes, and the number of years they urement in quantitative research—in particular,
spent at school, the information does not have to be finding valid and reliable ways to measure concepts.
transformed: the scores can simply be recorded in
a computer file. For other variables, processing the
data entails coding the information—that is, trans-
Concepts and their
forming it into numbers to facilitate quantitative measurement
data analysis. This consideration leads into step 9:
analysis of the data. In this step, the researcher What is a concept?
chooses among various statistical techniques to test Concepts are ideas or mental representations of
for relationships between variables, tries to deter- things. They are the building blocks of theory and
mine whether the measures are reliable, and so on. represent the points around which social research is
Next, in step 10, the researcher must interpret the conducted. Just think of the numerous concepts that
results of the analysis; it is at this stage that the “find- have already been mentioned so far in this book:
ings” emerge. Here the researcher must consider the
reasons why the research was done in the first place, emotional labour, secularization, hyperactiv-
in particular how the results may be used to answer ity, academic achievement, teacher expecta-
the research question(s). If there is a hypothesis, is it tions, charismatic leadership, crime, research
supported? What are the implications of the findings ethics, gatekeepers, drug and alcohol use, and
for the theoretical ideas that were the background to self-image
4 | The Nature of Quantitative Research 77
Each represents a label given to elements of the consistency relates both to time and to the
social world that seem to be significant and to have people using the measure. A measure’s results
common features. As Bulmer succinctly puts it, con- should not be affected by either the time when
cepts “are categories for the organization of ideas it is administered or the person who admin-
and observations” (1984, p. 43). One item mentioned isters it. With regard to time, this rule means
in Chapter 2 but missing from the list above is IQ. It that the measure should generate consistent
has been omitted because it is not, strictly speaking, results unless the phenomenon or characteris-
a concept. It is a measure of the concept of intelli- tic being measured has changed. For example,
gence. This is a rare case for social science, a measure measures of worker morale administered at
so well known that the measure and the concept are different times should indicate the same level
almost synonymous. of morale unless the morale itself has changed.
A concept can be either an independent or a Similarly, those measures and the data they
dependent variable. In other words, a concept can produce should be the same regardless of who
be presented as a possible cause of a certain aspect is using them. Whether a measure actually
of the social world (independent), or it can repre- possesses this consistency pertains to the issue
sent something needing an explanation (depend- of reliability, which will be examined later in
ent). Often the same concept will be an independent this chapter.
variable in one context and a dependent variable in • Measurement provides the basis for estimates
another. The concept “social mobility,” for example, of the nature and strength of the relation-
can be used in either capacity: as a possible explan- ship between variables (for example, through
ation of certain attitudes (are there differences be- correlation analysis, which will be examined
tween the downwardly mobile and others in their in Chapter 8).
political attitudes?) or as something to be explained
(why are some people upwardly mobile and others Indicators
not?). Equally, concepts may be used for descriptive Concepts used in quantitative research are given
or comparative purposes. For instance, a researcher two types of definition. One is a nominal definition,
might be interested in changes in the amount of which describes in words what the concept means,
social mobility in Canada over time, or in variations much as a dictionary definition does. For example,
among nations in their levels of social mobility. a nominal definition of “crime” might be: “any vio-
lation of the Criminal Code of Canada.” By contrast,
Why measure? a concept’s operational definition (from the idea of
There are three main reasons for the attention given operationalization, discussed above) spells out the
to measurement in quantitative research: operations the researcher will perform to measure
the concept. For example, one way to measure the
• Measurement allows for delineation of fine dif- incidence of crime is to use statistics provided by
ferences between people in terms of the char- police forces; another is to ask a sample of people
acteristic in question. This is very useful since, whether they have been victims of certain crimes.
although it is often easy to distinguish between To measure a concept it is necessary to have an
people in terms of extreme categories, finer dis- indicator or indicators that stand for or represent the
tinctions are much more difficult to recognize. concept. Sometimes the term indicator simply refers
Clear variations in levels of job satisfaction— to a measure in the ordinary sense: for example, the
people who love their jobs and people who hate total income reported on a tax return is an indicator
their jobs—are easy to see, but small differences of a person’s wealth. However, sometimes it refers
are much more difficult to detect. to an indirect measure of a concept that cannot be
• Measurement provides a consistent device or tapped easily or directly. For example, absenteeism
yardstick for gauging such distinctions. This may be used as an indirect measure of employee
78 Part II Quantitative Research
morale. Income declared for tax purposes may be • A single broad, general indicator may not cap-
a direct measure of personal income, but if used as ture all the meaning in the underlying con-
an indicator of social class, it becomes an indirect cept. If you were to ask people how satisfied
measure. they are with their work, for instance, you
There are a number of ways to devise indicators, might miss the complexity of the situation.
for example: People who answer “not satisfied” may like
some parts of their work, just as those who
• through questions that are part of a structured claim to be “satisfied” may dislike specific as-
interview schedule or self-completion ques- pects of their work.
tionnaire. The questions can be concerned • Alternatively, a more specific question may
with the respondents’ attitudes (e.g., toward a cover only one dimension of a particular con-
social issue like immigration), their personal cept. For example, to measure job satisfac-
experiences (e.g., stress), their behaviours (e.g., tion, is it sufficient to ask people how satisfied
leisure pursuits), and so on; they are with their pay? Almost certainly the
• by developing criteria for classifying observed answer is “no,” because for most people there
behaviour (e.g., pupil behaviour in a classroom); is more to job satisfaction than just satisfac-
• through the use of official statistics (e.g., those tion with pay. In this case, a single indicator
from Statistics Canada); and will miss things like satisfaction with benefits,
• by developing classification schemes to ana- with the work itself, and with other aspects of
lyze the content of written material (e.g., the work environment. By asking a number of
Sobel’s [2015] assessment of Canadian citizen- questions, the researcher can get access to a
ship guides issued between 1947 and 2012 as wider range of issues covered by the concept.
indicators of how the federal government de- • Multiple indicators of a concept allow for
fined what it means to be a Canadian and how sophisticated data analyses, such as factor
naturalized citizens were expected to behave). analysis and cluster analysis. Factor analysis
provides an indication of the extent to which
Indicators can be derived from a wide variety of (a) multiple measures really are measuring a
sources and methods. Very often a researcher has single concept as opposed to more than one
to consider whether one indicator of a concept suf- concept; and (b) individual items are measur-
fices. Instead of relying on a single indicator, the re- ing the concept in question. Bell et al. (2009),
searcher may use several to tap a concept. One way for example, used factor analysis to develop
to do that is to combine a number of Likert items measures of political attitudes. Cluster analy-
(see Box 4.1). sis allows researchers to determine whether
people or other units of study can be grouped
Using multiple-item measures in together on the basis of shared characteristics.
survey research Kennedy et al. (2013), for instance, identified
The main advantage in using a multiple-indicator four clusters of Edmonton residents based on
measure of a concept is that there are potential prob- the sustainable consumption behaviours in
lems in relying on just a single indicator: which they engaged (see Box 4.1).
• Multiple indicators can help to weed out
• A single indicator may misclassify some indi- response sets (see Chapter 5), which can
viduals if the wording of the question leads to occur if, for a variety of reasons, a person’s
a misunderstanding of its meaning. If there are response does not reflect his or her real pos-
several indicators, however, it is possible that ition on the matter examined. For example,
the other items will allow proper classification someone may “agree” with an item or set of
of the person. items just to appear cooperative or friendly.
4 | The Nature of Quantitative Research 79
The investigation of attitudes is a prominent area There are several points to remember in com-
in survey research. One of the most common tech- bining Likert items. The following are particularly
niques for investigating attitudes is the Likert scale. important:
Named after Rensis Likert, who developed the • The items must be statements, not questions.
method, it is typically a multiple-indicator meas- • The items must all relate to the same object (in
ure of the intensity of feelings about a particular the example above, sustainable consumption).
topic. In its most common format, it comprises a • The items that make up the scale should be
series of statements (known as “items”) on an interrelated.
issue or theme. Usually, each respondent is asked • It is useful to vary the phrasing so that some
to indicate his or her level of agreement with the items imply a positive view of the phenom-
statement. The format for indicating level of agree- enon of interest and others a negative one.
ment is normally a five-point scale from “strongly For example, in the 2007 Alberta Survey
agree” to “strongly disagree,” but seven-point and conducted by the Population Research Lab-
other formats are used as well. There is usually a oratory at the University of Alberta, attitudes
middle position of “neither agree nor disagree” or toward body checking in minor hockey were
“undecided” indicating neutrality on the issue. A measured using five-point Likert items. One
respondent’s reply on each item is scored and then item read: “Body checking is the main cause
the scores are aggregated to form an overall score. of injuries for children in organized minor
Variations on the typical format of agreement hockey.” A second was: “Body checking is an
are scales referring to evaluation (for example, essential part of the game of organized minor
“very poor” to “very good”) and frequency. Ken- hockey.” The first implies a negative view of
nedy et al. (2013) used the latter in a study of body checking, the second a more positive
sustainable consumption that was conducted in one. Under these conditions, it would be logic-
Edmonton. Respondents were presented with sev- ally inconsistent to answer every question
eral items such as “Leave vehicle at home to avoid the same way, so if someone has done that it
driving,” “Buy local food,” and “Actively encourage indicates that response set may be a problem.
others to reduce their consumption,” 12 of which A respondent who either agreed or disagreed
were selected for analysis. Participants in the with all the items could not be answering
study chose either “Never,” “Rarely,” “Sometimes,” truthfully, and might have to be excluded from
“Often,” or “Always” as their response. the study.
In Hay’s (2014) study of secularization (see Box 1.3), frequency of religious attendance, was measured
religious pluralism was measured using a single, with the question, “Do you currently attend church
five-point Likert item that formed part of an exten- temple or mosque?” Respondents who answered
sive survey of Canadians’ value systems. However, “yes” were then asked: “How often?” The response
secularization (the dependent variable in Hay’s choices were: “once a week or more” (given a code
analysis) was measured with several different indi- of 5 after reverse-coding), “monthly” (4), “every
cators in order to tap into different dimensions of few months” (3), “once or twice a year” (2), and
the concept. One dimension, religiosity, was meas- “never” (1). A third dimension, concerned with the
ured by averaging the responses to three 10-point participants’ belief in the religion of their par-
items indicating the importance respondents ents, had the response categories “believe all of
placed on: (a) “believing in God;” (b) “obeying God, it” (4), “believe most of it but not all” (3), “believe
doing what he wishes;” and (c) “relating to God in some parts but disbelieve others” (2), and “don’t
a personal way.” A second dimension, relating to believe any of it” (1).
dimension of the concept (e.g., respecting confi- simply tags to allow the material to be stored quanti-
dentiality) may not necessarily score high on other tatively. Then it is necessary to go through the infor-
dimensions (e.g., fiscal honesty or continuing edu- mation again to look for incidences of the theme or
cation), so that for each respondent one can have a category, and to record the appropriate numbers on
multidimensional “profile.” a computer spreadsheet. This approach is sometimes
However, in much quantitative research, there called post-coding. Post-coding can be an unreliable
is a tendency to rely on a single indicator for each procedure because there may be inconsistencies in
concept. This is quite adequate for some purposes, in the judgments of different coders, which leads to both
particular when one is measuring an uncomplicated measurement error and lack of validity. Box 4.3 pro-
variable such as age. Some studies, like Hay’s (2014) vides an example of this kind of coding.
research on secularization in Canada (see Box 4.2), When Schuman and Presser (1981; see Box 5.1)
employ both single- and multiple-indicator meas- asked an open question about the features of a job
ures of concepts. What is crucial is that the measures that people like, the answers were to be grouped into
be reliable and valid representations of the concepts 11 codes: pay; feeling of accomplishment; control of
they are supposed to be gauging. work; pleasant work; security; opportunity for pro-
motion; short hours; working conditions; benefits;
Coding unstructured data satisfaction; other responses. Each of these 11 cat-
Many forms of social science data are essentially un- egories was assigned a number: 1 for pay; 2 for feel-
structured and unorganized, including answers to ing of accomplishment; 3 for control of work; 4 for
open questions in interviews and questionnaires, and pleasant work, and so on.
the content of newspaper articles. To make sense of Murphy and Fedoroff (2013) used a combination
the information, researchers must go through it all, of fixed-response and open-ended questions to ex-
deriving themes or categories of behaviour to form the plore how 30 registered sex offenders viewed their
basis for codes (the labels or titles given to the themes experiences with either the Ontario or the National
or categories): for example, “hostile to outsiders,” or Sex Offender Registry. People on these registries have
“not hostile to outsiders.” Next, the researcher usually to report annually to their local police and receive
assigns numbers to the codes. This may be a largely a yearly visit from the police at their residence, but
arbitrary process in the sense that the numbers are unlike in the United States, the registries themselves
4 | The Nature of Quantitative Research 81
Foddy (1993) asked a small sample of his students to do with the difficulty of coding an open question;
to state their fathers’ occupations; then he re- it is also due to a lack of specificity in the question.
quested three details: nature of business, size of If Foddy had asked, “How many employees are
business, and whether their father was the owner (were) there in your father’s place of employment?”
or an employee. The replies to the “size of business” a more meaningful set of answers would have been
question were particularly varied, including “big,” forthcoming. Whether his students would have
“small,” “very large,” “3000 acres,” “family,” “multi- known this information is, of course, yet another
national,” “200 people,” and “Philips.” The problem issue. However, his experience illustrates some
here is obvious: these categories do not provide a potential problems with asking an open question,
useful measure of size. The problem has only partly particularly one that lacks a clear reference point.
are not available for public viewing. The responses to Coding is also used in qualitative research, but its
the open-ended questions were analyzed to discern role and significance there are somewhat different
whether there was consensus or disagreement among (see Chapter 13).
participants and to identify recurrent themes. About
two-thirds of the offenders who commented on the
matter stated that being on a sex offender registry
Reliability and measurement
was not a great burden or even an intrusive experi- validity
ence, and about half said that having to register was
“fair,” given the crimes they had committed. How- Reliability
ever, some offenders objected to the manner in which Reliability is concerned with the consistency of meas-
the police conducted their annual visits. ures. The term has at least three different meanings.
When coding, you should observe three basic
principles (Bryman & Cramer, 2001): Stability over time
“Stability” refers to whether the results of a measure
• The categories must not overlap. If they do, the fluctuate as time progresses, assuming that what is
numbers assigned to them will not denote dis- being measured is not changing. This means that
tinct behaviours or types of responses. if one administers a measure to a group and then
• The list of categories must be exhaustive, cover- re-administers it perhaps an hour later, there should
ing all possibilities. If it does not, some material be little variation in the results. Most thermometers
cannot be coded. That is why many classifica- have this kind of reliability.
tion schemes, such as those used to code open The most obvious way of testing for the stability
questions, include the category “other.” of a measure is the test–retest method. This involves
• There should be clear rules about how codes administering a test or measure on one occasion and
are to be applied, with examples of the kinds then re-administering it to the same sample on an-
of answers that may be subsumed under a par- other occasion, that is:
ticular category. Such rules are meant to pro- T1 T2
mote a coder’s consistency over time in how
Obs1 Obs2
the material is assigned to categories and, if
more than one person is coding, that the vari- One would expect to find a high correlation be-
ous coders are consistent with each other. tween Obs1 and Obs2. For example, those who score
82 Part II Quantitative Research
high on the first observation should also score high Its value varies from 1 (denoting perfect internal reli-
on the second, and those who score low on the first ability) to 0 (denoting none). The value .80 is typically
should score low on the second. Imagine a m ultiple- employed to mark the minimum acceptable level of
indicator measure that is supposed to tap a con- internal reliability, though many writers work with
cept called “designerism” (a preference for buying a lower value. Berthoud (2000, p. 169), for instance,
goods and especially clothing with “designer” labels). wrote that a minimum level of .60 is “good.” In the
The measure would be administered to a sample of case of George and Chaze’s (2014) battery of items
respondents and then later re-administered. If the to measure discrimination against foreign trained
correlation between Obs1 and Obs2 is low, the measure engineers in Canada, alpha was .92 for both the
is unstable, implying that it cannot be relied upon. international and locally trained subsamples. In a
However, there are a number of problems with study of environmental values among Chinese- and
this approach to evaluating reliability. Respondents’ Anglo-Canadians by Deng et al. (2006), the alphas
answers at T1 may influence how they reply at T2: for ranged from .50 to .89 for various variables. The
instance, perhaps answering a question on designer study of domestic labour by Gazso-Windlej and
goods increases some respondents’ interest in them. McMullin (2003) could only achieve alphas aver-
Yet giving an answer at T1 is not supposed to affect later aging .50 for its Likert-style items measuring gender
measurements. Second, events may intervene between ideology, although the study was only exploratory.
T1 and T2 that influence the degree of consistency. For The use of Cronbach’s alpha has grown as a result of
example, if a long span of time is involved, changes in its incorporation into computer software for quanti-
the economy or in respondents’ personal financial cir- tative data analysis.
cumstances can influence their views about designer Another way of testing for internal reliability is
goods and their predilection for them. the split-half method. Take the Likert items used
There are no clear solutions to these problems, by Kennedy et al. (2013) in their study of sustain-
other than introducing a complex research design able consumption (see Box 4.1). The 12 indicators
and so turning the investigation of reliability into would be divided into two halves of 6, allocated on
a major project in its own right. Perhaps for these a random or an odd-numbered item/even-numbered
reasons, many, if not most, research projects do not item basis. The degree of correlation between the
include tests of stability over time. scores on the two halves for all respondents would
When such tests are done, they may indicate that then be calculated. If the 12 items are consistent, a
there is no problem, even when a substantial amount respondent’s score on the two groups of indicators
of time has elapsed. Berthoud (2000), for example, should be similar, perhaps high on both, or low on
was pleased that an index of illness achieved high both. A perfect positive correlation and therefore
test–retest reliability even though the tests were con- complete internal consistency would yield a correla-
ducted a year apart. tion coefficient of 1; no correlation and therefore no
internal consistency would produce a coefficient of
Internal reliability (or internal consistency) 0. The meaning of correlation and correlation coeffi-
The key issue here is whether multiple measures cients will be explored in much greater detail later
that are administered in one sitting are consistent— on. The chief point at this stage is that the correla-
in other words, whether respondents’ scores on any tion coefficient establishes how closely respondents’
one indicator tend to be related to their scores on scores on the two groups of indicators are related.
the other indicators. For example, on a scale cre-
ated to measure attitudes toward liberal democracy, Inter-observer consistency
people who agree with the statement that voting is When more than one researcher is taking part in
an important right should also agree that freedom of an activity involving subjective judgment, such as
speech is an important right. Cronbach’s alpha coeffi- the recording of observations or the translation of
cient is a commonly used test of internal reliability. data into categories, their judgments may differ. This
4 | The Nature of Quantitative Research 83
problem can arise in a number of contexts: for ex- (for example, people) are known to differ. Assuming
ample, if answers to open-ended questions have to that as job satisfaction goes down, absenteeism goes
be categorized, or in structured observation when up, a researcher seeking to establish the concurrent
observers have to decide how to classify subjects’ validity of a new measure of job satisfaction might
behaviour. Is the person under observation “afraid,” look to see if people who are satisfied with their jobs
“concerned,” or “just thinking” when reading about are less likely to be absent from work than those who
the spread of yet another deadly new virus? Prob- are not satisfied. If there is no difference in absentee-
lems arise if all observers do not classify a particu- ism between those who are satisfied and those who
lar behaviour in the same way. Cramer (1998, ch. 14) are not, doubt is cast on whether the new measure is
provides a very detailed treatment of inter-observer really gauging job satisfaction.
consistency and the techniques that can be used to
maximize it. Construct validity
Some writers advocate that researchers estimate the
Measurement validity construct validity of a measure, which involves seeing
As we noted in Chapter 2, measurement validity whether the concepts used in the research relate to
refers to whether an indicator (or set of indicators) each other in a way that is consistent with what their
devised to gauge a concept really does so. When theories would predict. For example, suppose that a
people argue about whether an IQ score really meas- researcher wants to establish the c onstruct validity
ures or reflects someone’s level of intelligence, they of the new measure of job satisfaction. Drawing on a
are raising questions about the measurement valid- theory maintaining that job satisfaction is influenced
ity of the IQ test in relation to the concept of intelli- by the stimulation that comes from performing sev-
gence. Whenever students and their teachers debate eral different activities, he or she may anticipate that
whether multiple-choice examinations provide an people who do routine jobs are less satisfied with
accurate measure of academic ability, they too are their jobs than those who have greater opportunities
raising questions about measurement validity. Estab- for variety, complexity, and creativity. If these vari-
lishing face validity, concurrent validity, construct ables correlate in the expected way, then the meas-
validity, and convergent validity shows that you have ure in question has construct validity. On the other
in fact measured what you wanted to measure. hand, some caution is required if the relationship is
weak or non-existent, for example, if those who do
Face validity routine jobs are just as satisfied as those with jobs
At the very minimum, a researcher who develops a new involving a lot of variety. If so, the measures used
measure should establish that it has face validity—that may be invalid, or the deduction that is made from
is, that it appears to reflect the content of the concept the theory may be misguided, or the theory itself
in question. Face validity can be established by asking may be in need of revision. Whatever the problem
those with expertise in a field to act as judges to deter- may be, it’s probably best to seek another measure
mine whether, on the face of it, the measure seems to and try again.
reflect the concept concerned. Alternatively, research-
ers can assess their own measures for face validity, as Convergent validity
Kennedy et al. (2013) did in their study of sustainable In the view of some methodologists, the validity of
consumption (see Box 4.1). Establishing this sort of a measure ought to be gauged by comparing it to
validity is essentially an intuitive process. measures of the same concept developed through
other methods. For example, if a questionnaire asks
Concurrent validity managers how much time they spend on various
The researcher can also seek to gauge the concurrent activities (attending meetings, touring their organ-
validity of the measure by employing a criterion rel- ization, informal discussions, and so on), its valid-
evant to the concept in question but on which cases ity may be determined by directly observing the
84 Part II Quantitative Research
managers to see how much time they actually spend surveys measure the crimes that people experi-
on those activities. ence, as well as some things that may not be crimes;
An interesting instance of convergent invalid- they also omit certain crimes (for example, when
ity is described in Box 4.4. Crime surveys were a stolen item is presumed to be “lost”) and rely
consciously devised to serve as a check on official on respondent memory, which can be faulty (see
police statistics. The two sets of data are collected Chapter 5). Police statistics, on the other hand,
in quite different ways: official crime statistics are measure the crimes that people are willing to call
collected as part of the bureaucratic processing of the police about, plus the crimes the police them-
offenders in the criminal justice system, whereas selves discover; therefore, not all crimes end up in
crime victimization surveys provide data from official police statistics. Figure 4.2 illustrates some
interviews with members of the general public. In of the problems associated with crime rates that
the case reported in Box 4.4, a lack of convergent are based on police statistics. At any step along
validity was found. This illustrates a problem with Figure 4.2, some potential crimes drop away. For
the convergent approach to testing validity: it’s example, if a criminal act goes unnoticed, or is no-
not easy to establish which of the two measures ticed but not recognized as criminal, or is noticed
provides the more accurate picture. In the case of and recognized as criminal but is not reported to
crime statistics, the two methods are really meas- the police, it does not enter the official statistics. As
uring somewhat different things. Victimization you can see, a substantial amount of crime—often
11. A criminal offence validity of many measures are never tested. When
a new multiple-indicator measure is devised, there
may be an examination of face validity and a test
22. Member of the public for internal reliability, but in many cases no further
is witness to or victim
of a crime
testing takes place.
It should also be remembered that, although reli-
ability and validity are analytically distinguishable,
they are related: if a measure is not reliable, it cannot
33. Recognition of offence be valid. For example, a multiple-indicator measure
as criminal by a member
of the public lacking internal reliability is probably measuring
two or more different things and therefore is not a
55. Police decisions valid indicator of the single concept it is supposed to
about patterns of
surveillance be measuring.
44. Reporting of offence
by a member of the
public
6 Police discretion
The main goals of quantitative
researchers
Measurement
7 Recording of offence To understand the social order one must be able to
by police
go into the world and measure things like preju-
dice, the incidence of homelessness, attitudes
toward the federal government, or the distribu-
8 Notifiable crimes tion of wealth in society. It would be awfully hard
recorded by the to make sense of such phenomena or to evaluate
police—”the crime
rate” theories and interpretations of them without ac-
FIGURE 4.2 The social construction of crime quiring data on them: that is, without measuring
s tatistics: Eight steps them. It’s not surprising, then, that the reliabil-
Source: Adapted from a figure in Beardsworth et al. (n.d.) ity and validity of such measurement are matters
of concern.
Establishment of causality
referred to as “the dark figure”—is not included in (internal validity)
crime statistics (Silverman et al., 2000). The “true” Most quantitative research involves a search for
volume of crime at any one time is almost always a causal explanations. Quantitative researchers are
contested notion (Reiner, 2000). rarely satisfied with merely describing how things
are: they are keen to find out why things are the way
they are—an emphasis also found in the natural sci-
Reflections on reliability ences. Researchers examining prejudice, for instance,
and validity may want not only to describe it but also to explain
There are, then, a number of ways to investigate the it, which means finding its causes. They may seek to
merit of measures devised to represent social scien- explain prejudice in terms of personal characteristics
tific concepts. However, the discussion of reliability (such as a low level of education) or social character-
and validity is potentially misleading because not istics (such as low social mobility). In the resulting
all new measures of concepts are submitted to the reports, prejudice is the dependent variable, the one
rigours just described. In fact, the reliability and to be explained, and levels of education and social
86 Part II Quantitative Research
mobility are independent variables, the ones tested sense or prior theory, as in the example concern-
as possible causal influences on prejudice. ing level of education and prejudice in the previous
When an experimental design is employed, the paragraph. However, there is always the risk that the
independent variable is the one that is manipulated, inference is wrong, that the variable purported to be
and there is little ambiguity about the direction of the cause did not precede the dependent variable in
causal influence. However, with cross-sectional time (see Box 4.5).
designs of the kind used in most social survey re- Good quantitative research inspires confidence
search, there is often ambiguity about the direction in the researcher’s causal inferences. Research of the
of causal influence because the data on all variables more experimental type is often better able to estab-
are collected simultaneously: this means that in lish causality than cross-sectional research because
many instances we cannot say with full confidence with the former it is easier to demonstrate the direc-
that a particular independent variable preceded the tion of causal influence and to control for other in-
dependent one in time, one of the three criteria of dependent variables. However, as we saw in Chapter 2,
nomothetic causation mentioned in Chapter 2. To experiments may not be appropriate for many of the
refer to independent and dependent variables in the topics of interest to social scientists. Instead, quanti-
context of cross-sectional designs, we must infer the tative researchers who employ cross-sectional designs
temporal sequence of variables based on common often use techniques that allow them to make at least
Following a review of the literature, Sutton and between staff and customers; semi-structured
Rafaeli (1988) hypothesized a positive relation- interviews with store managers; casual conversa-
ship between the display of positive emotions by tions with store managers, supervisors, executives,
staff to retail shoppers (smiling, friendly greet- and others; and data gathered through posing
ing, eye contact) and the level of retail sales. In as a customer. The qualitative investigation sug-
other words, when retail staff are friendly and give gested that the relationship between the display
time to shoppers, higher sales follow. Sutton and of positive emotions and sales was indeed nega-
Rafaeli had data from 576 convenience stores in tive, but that sales were likely to be a cause rather
a US national retail chain. Structured observation than a consequence of the display of emotions. In
of the retail workers provided data on the display stores with high levels of sales, staff were under
of positive emotions, and quantitative sales data greater pressure and encountered longer lineups
provided information for the other variable. at checkouts. Therefore they had less time and in-
The hypothesis was not supported; indeed, clination for the pleasantries associated with the
stores in which retail workers were less inclined display of positive emotions.
to smile, be friendly, and so on, had better sales. Thus the causal sequence was not
Sutton and Rafaeli (1992, p. 124) considered restat-
More positive emotions → More retail sales
ing their hypothesis to make it seem that they had
found what they had expected (which would have but
raised ethical concerns) but fortunately resisted
More retail sales → Fewer positive emotions
the temptation. Instead, they conducted a qualita-
tive investigation of four stores in order to under- This study highlights the main difficulty a
ssocia-
stand what was happening. They used several ted with inferring causal direction from a c ross-
methods: unstructured observation of interactions sectional research design.
4 | The Nature of Quantitative Research 87
tentative causal inferences, such as statistical controls applied to the population from which the sample was
(see Chapter 8). The rise of longitudinal research drawn. The goal of arriving at generalizable findings
almost certainly reflects a desire on the part of quan- in this way can be seen as a parallel to the goal of
titative researchers to improve their ability to gener- developing scientific “laws” in the natural sciences.
ate findings that permit a causal interpretation. Probability sampling, explored in Chapter 7, is
normally the first choice among researchers seek-
Generalization of findings ing a representative sample. This procedure largely
In quantitative research, researchers usually want eliminates bias by using a random selection process.
to generalize their findings beyond the people who Random selection does not guarantee a represent-
participate in the study and the particular context ative sample, but it does improve the chances of
in which the research is conducted. These concerns getting one. Here it is important to emphasize that
pertain to external validity, an issue introduced in even a perfectly representative sample would be rep-
Chapter 2. resentative only of the population from which it was
A particularly important matter in this regard is selected: strictly speaking, therefore, results obtained
whether a study’s findings are applicable in settings from that sample cannot be generalized beyond that
outside the research environment—in particular, population. This means that, if the population from
everyday or natural social settings. Sometimes the which a sample is taken is all inhabitants of a particu-
research setting is so artificial or so different from lar town, city, or province, generalizations should be
real life that we may question whether the results ac- made only to the inhabitants of that town, city, or
tually tell us anything about the sorts of things that province. It may be very tempting to see the find-
people normally experience. The more the social ings as having a wider applicability, so that results
scientist intervenes in natural settings or creates un- from a sample representing a city such as Vancouver
natural ones, such as a laboratory (or even just a spe- or T oronto could be considered relevant to similar
cial room for interviews), the greater the chance that cities. Even so, no inferences should be made beyond
the findings will be externally invalid. Measurement the population from which the sample is selected.
validity and plausible evidence of causation cannot The concern with the generalizability of research
guarantee this form of external validity. findings is particularly strong among quantitative
A second concern with external validity is researchers using cross-sectional and longitudinal
whether the results of a study can be generalized designs. Experimental researchers are concerned
beyond the people or cases analyzed by the re- about generalizability too, as the discussion of exter-
searcher. The researchers conducting the study in nal validity in Chapter 2 suggested, but people doing
Box 1.2 asked 57 adolescent male volunteers about experiments usually give greater attention to estab-
their delinquency and examined whether any of lishing causation than to external validity.
the currently popular sociological theories of delin-
quency could explain their behaviour. How far can Replication
the findings of this study be generalized? To other The natural sciences are often depicted as seeking to
male adolescents who attended the same school? To reduce to a bare minimum the contaminating influ-
all Canadian male adolescents? If the findings are ence of the scientist’s biases, values, characteristics,
fully externally valid, they can be applied to any and expectations. Were those influences allowed to
population of interest. If they possess no external affect research, the claims of the natural sciences to
validity, the findings apply only to the 57 boys stud- provide a definitive picture of the world would be
ied and to no one else. Usually, external validity falls seriously undermined. Scientists also try to mini-
somewhere between these extremes, and it is in this mize if not eliminate routine errors in the conduct
context that how people are selected to participate in of their research. To guard against such problems,
research becomes crucial. If a representative sample scientists believe that they should be able to replicate
of people is selected (see Chapter 7), the researcher one another’s research. If a certain finding cannot be
can be confident that the results of the study may be reproduced after repeated tries, serious questions are
88 Part II Quantitative Research
raised about its validity. Likewise, researchers in the affect his or her findings would appear to be all the
social sciences often regard the ability to replicate as greater in the case of the social world. Consequently,
an important criterion for judging research. It’s easy researchers usually take care to spell out all research
to see why: the risk that the researcher’s values will procedures so that others can replicate them.
Unfortunately, replication tends to be considered from exponents of qualitative research. The criticisms
pedestrian and uninspiring. It is not a high-status ac- pertain to quantitative research as a general research
tivity in either the natural or social sciences, and stan- strategy, and to specific methods and research de-
dard replications are rarely accepted for publication signs with which quantitative research is associated.
in many academic journals. A further reason for the
low incidence of published replications in the social Criticisms of quantitative research
sciences is the difficulty of ensuring that the condi- To give a flavour of the critique of quantitative re-
tions are precisely the same as those in the original search, six criticisms are covered briefly here:
study. If there is room for doubt on that score, any dif-
ferences in findings may be attributed to the design of • Quantitative researchers fail to distinguish
the replication rather than to a deficiency in the ori- people and social institutions from “the world
ginal study. Nonetheless, it is crucial that the methods of nature.” Some critics object to the idea of
used to generate findings are made explicit, so that treating the social world as if it were no dif-
replication is possible. Providing all the information ferent from the natural order. They argue that
needed to conduct a replication is often regarded as an because people interpret the world around
important quality of quantitative research. them, their actions and experiences cannot
An example of the benefits of replication can be be studied using the methods employed in the
found in Goyder et al.’s (2003) study of occupational natural sciences. They claim that science is
prestige conducted in the Kitchener–Waterloo area applicable only to entities and processes lack-
of Ontario. The authors replicated research done ing this sort of self-reflection, such as chemical
25 years earlier (Guppy & Siltanen, 1977), taking elements, photosynthesis, and the circulation
great pains to ensure that the methodologies of the of blood. But many quantitative researchers
two studies were as similar as possible. The replica- maintain that humans (and other animals)
tion not only provided a check for biases and routine really are part of the world of nature, and that
errors; given that a quarter-century had passed since it is simply wrong to think that our existence
the original study, it also provided a measure of how cannot be usefully analyzed using science. For
much attitudes had changed with regard to occupa- instance, they would claim that consciousness
tional prestige and gender. As reported in Chapter 2, itself is amenable to the scientific method, as
the authors concluded that the earlier male advan- are emotions, decision-making processes, and
tage in occupational prestige had disappeared by so on. The debate continues.
the time the replication was done; in fact, women in • The measurement process produces an artifi-
people-oriented jobs enjoyed higher prestige than cial and false sense of precision and accuracy.
men did in the more recent study. If the authors For example, quantitative research presumes
had not taken care to ensure accurate replication of that different individuals responding to the
the original study, it would be impossible to know same question on a survey are interpreting
whether the differences in the findings represented its key terms in the same way. Yet a question
actual social change or were simply a by-product of such as “What is your social class?” can be in-
using different methodologies. terpreted in at least two ways: as a reference to
current wealth, or as a reference to ancestry
going back generations. Many methodologists
Critiques of quantitative find significant variations in respondents’ in-
research terpretations of such terms. Researchers often
Every approach to research has its strengths and attempt to solve this problem by using ques-
weaknesses. Over the years quantitative research, tions with fixed-choice answers—“Are you
along with its epistemological and ontological foun- upper class, middle class, working class?” and
dations, has been the focus of criticism, particularly so on—but this approach provides “a solution
90 Part II Quantitative Research
to the problem of meaning by simply ignoring in their everyday lives. But is this the case?
it” (Cicourel, 1964, p. 108). Quantitative social How can researchers who pose a set of ques-
researchers counter that there are ways to test tions designed to measure attitudes toward
for shared meanings, although they would federal–provincial relations, for example, be
concede that this issue deserves more attention sure that respondents have sufficient interest
than it usually gets. in the topic to produce meaningful responses?
• The reliance on instruments and procedures Similarly, there may be an important d ifference
produces a disjuncture between research and between what people say to researchers and
everyday life. This issue relates to the question what they actually do in real life (see Box 4.6).
of external validity. For example, many meth- Another criticism, introduced in Chapter 2, is
ods of quantitative research rely heavily on directed at experimental research that relies
research instruments such as structured inter- on controlling situations to determine causal
views and self-completion questionnaires. connections. According to Brannigan (2004),
As Cicourel (1982) pointed out, researchers such research can produce only small, short-
tend to assume that survey respondents have term manipulations of independent variables;
both the knowledge they need to answer the yet much of everyday life is affected by long-
questions and a sense that the topic matters term, ongoing social processes.
A study of racial prejudice conducted many years though it is unlikely that this point had a signifi-
ago by LaPiere (1934) illustrates that there may be cant bearing on the findings and their implications
a difference between what people say and what for survey research. On the other hand, the results
they actually do. LaPiere spent two years travel- might have been just another example of the wide-
ling with a young Chinese student and his wife, ob- spread difference between holding a prejudiced
serving from a distance to see if they were refused attitude and engaging in a discriminatory act. An
entry at hotels and restaurants. Of 66 hotels ap- experimental study of prejudice among college
proached, only one turned the couple away; of 184 students (Frazer & Wiersma, 2001) showed that
restaurants and diners, none refused entry. in hypothetical situations the students hire black
LaPiere then allowed six months to elapse and white applicants of varying abilities equally,
before sending questionnaires to the hotels and but a week later recall the black applicants as less
restaurants visited. One question asked: “Will you intelligent than the whites even though the two
accept members of the Chinese race as guests in groups were equal. In the real world, peer pres-
your establishment?” Of the establishments that sure can make an unprejudiced person discrimin-
replied, 92 per cent of restaurants and 91 per cent ate, to “go along,” while a prejudiced person may
of hotels said no. LaPiere’s simple though striking not discriminate for fear of a lawsuit.
study clearly illustrated a gap between reports The gap is usually worst when predictions
of behaviour and actual behaviour. The fact that of future behaviour are involved. In 2004 the
the question asked was somewhat unclear is not Canadian Blood Services reported that 28 per cent
usually noted in connection with this widely cited of Canadians said they intended to give blood in
study. “Will you . . .?” can be interpreted in two the next year, but only 3.7 per cent actually did.
ways: as a question about the future or as a ques- And in any election many people say that they
tion about current policy. Why the more obvious intend to vote—and even reveal their preferences
formulation “Do you . . .?” was not used is not clear, to pollsters—but then do not follow through.
4 | The Nature of Quantitative Research 91
• The analysis of relationships between variables perceptions of the people to whom the findings
promotes a view of social life that is remote purportedly pertain. A related issue arises
from everyday experience. Blumer argues that when valid quantitative findings are achieved,
studies designed to bring out the relationships but the explanation for them is not informed
between variables omit “the process of inter- by an empathetic understanding of the people
pretation or definition that goes on in human involved. For example, it’s empirically true
groups” (1956, p. 685). This symbolic interac- that in many developed countries today much
tionist assessment incorporates the first and larger proportions of children are born to
third criticisms above, that the meaning of unwed parents than was the case in previous
events to individuals is ignored and that the decades. Many researchers conclude from
connection to everyday contexts is missing. this that marriage is losing its popularity in
Quantitative researchers recognize those risks, certain sectors of society, such as among the
but claim that their research does not pre- poor in inner-city areas. However, in-depth
clude a search for how people interpret their field research conducted by Edin and Kefalas
everyday existence. For example, quantitative (2005) suggests that among the poor single
sociologists sometimes ask respondents what mothers they studied in Philadelphia, mar-
they think about social inequality, and try to riage is in fact highly valued, but is expected
link the responses with variables such as age, to happen later in life, after the children have
gender, class, and ethnicity. Nonetheless, more been born. “What is striking about the body
thorough and research participant–centred in- of social science evidence,” they write, “is
quiries into questions of meaning are usually how little of it is based on the perspectives
undertaken by qualitative sociologists. and life experiences of the women who are its
• Explanations for findings in quantitative stud- subjects. . . . [S]urveys, though they have me-
ies may be provided without examining the ticulously tabulated the trend, have led us to a
dead end when it comes to fully understand- or given—at least at a particular point in
ing the forces behind it” (Edin & Kefalas, history—rather than created by individuals
2005, p. 5). through negotiation, although the latter view
• Quantitative researchers tend to assume an is not universal among quantitative research-
objectivist ontology. As we saw in Chapter 1, ers. People on the quantitative side reply that
quantitative researchers often assume that it is not a mistake to assume that some things
a reality exists that may be independent of may exist and have certain characteristics re-
the observer or of individual consciousness. gardless of how we perceive them. Again, the
They may also see the social order as fixed debate is ongoing.
Key Points
• Quantitative research can be characterized • Quantitative research has the following key goals:
as a linear series of steps moving from theory measurement, the establishment of causality, gen-
to conclusions, but the process described in eralization, and replication.
Figure 4.1 is an ideal from which there are • Quantitative research has been criticized by quali-
many departures. tative researchers. These criticisms tend to revolve
• Examining the reliability and validity of measures around rejecting the view that a natural science
is important for assessing their quality. model is appropriate for studying the social world.
Relevant Websites
The Statistics Canada site provides access to a wide helps economically developing nations enhance
variety of quantitative information about Canada. their research output on child development.
It also offers up-to-date figures on things such as www.unb.ca/research/institutes/crisp/
the inflation rate, the level of unemployment, and
The Web Center for Social Research Methods pro-
Canada’s population.
vides examples and discussion of Likert scaling.
www.statcan.gc.ca/start-debut-eng.html
www.socialresearchmethods.net/kb/
The Canadian Research Institute for Social Policy scallik.php
(CRISP), at the University of New Brunswick, conducts (Websites accessed 20 October 2018)
Chapter Overview
Survey research is one of the most commonly used data-gathering techniques in the social sciences.
Surveys can take the form of either interviews, in which participants are asked questions orally, or
questionnaires, which respondents read and fill out themselves. One of the strengths of survey research
is that it allows for standardization in the asking of questions and the categorization of the answers given.
This chapter explores:
• the reasons for the widespread use of surveys, including a consideration of the importance of stan-
dardization to the process of measurement;
• issues associated with asking questions in surveys:
»» whether to use open or closed questions
»» different kinds of questions that can be asked in structured interviews and questionnaires
»» rules to bear in mind when designing questions
»» optimal question order
»» how questionnaires can be designed to minimize error and make answering easier for respondents
»» projection or vignette questions in which respondents are asked to reflect on a hypothetical
scenario presented to them
»» the importance of pre-testing questions
»» the use of questions taken from previous research
• the different ways to do survey research, such as using more than one interviewer, conducting
interviews by telephone, or using the Internet or email to administer questionnaires;
• various prerequisites of structured interviewing, including establishing rapport with the inter-
viewee; asking questions exactly as they appear on the interview schedule; keeping to the question
order as it appears on the schedule; and recording exactly what is said by interviewees;
• problems with survey research, including the influence of the interviewer on respondents and the
possibility of systematic bias in answers (known as response sets);
▲ AndreyPopov/iStockphoto
96 PART II Quantitative Research
• the advantages and disadvantages of the questionnaire compared to the structured interview;
• researcher-driven diaries as a form of survey research;
• secondary analysis of survey data; and
• a feminist critique of structured interviews and questionnaires.
I never do surveys,” said Anna-Lise as she hung up “Well, sometimes they give you a statement to respond
the phone. to rather than a question. And I guess they do the ‘On a
“Why?” I asked. scale from 1 to 5’ thing because they want to compare re-
“For one thing, they always call at dinnertime, when sponses between individuals in a way that is meaningful, or
I’m tired and hungry.” they may want to put a number on what people in general
“Maybe that’s because most people are home at think of the issue, by calculating a mean or something.”
dinnertime,” I said. “They turn phone conversations into numbers? Very
“That doesn’t make it any more comfortable. And weird.”
they always ask questions in the weirdest ways, like: ‘On “Ever step on a scale to weigh yourself? Hey, there’s
a scale from 1 to 5, with 1 being strongly dislike and 5 another kind of scale! Or how about the numbers that
being strongly like, how do you feel about blah, blah, appear on the gas pump when you’re filling up, or num-
blah.’ Why can’t they just ask you your opinion on some- bers showing the time in the bottom right-hand corner of
thing and let you say what you want? That would be way your computer? Same idea—turning things into numbers.”
more interesting.” “I still say the whole thing is very weird.”
that students vary in the amount of alcohol they con- the word “soon” in a question can vary widely
sume. Most measurement will contain an element of from person to person.
error, so it is helpful to think of the measured varia- • In interviews, a large number of closed ques-
tion in alcohol consumption as made up of two com- tions reduces conversation and gives the
ponents: true variation and error. In other words: interview an impersonal feeling, reducing
rapport.
measured variation = true variation + v ariation
due to error
The aim is to keep the error portion to a min-
Types of questions
imum (see Figure 5.1), since error reduces the valid- Structured interviews and questionnaires generally
ity of a measure. Standardization in interviews and contain several different types of question. There are
questionnaires means that two sources of variation various ways of classifying them, but here are the
due to error—items 2 and 5 in the preceding list—are main forms:
likely to be less pronounced.
Variation due to error can come from several • Personal, factual questions. These are ques-
other sources. tions that ask respondents to provide personal
information, such as their age, occupation,
Disadvantages marital status, and income. This also includes
Closed questions also present certain disadvantages, questions about their behaviour, such as the
however: frequency of their attendance at religious ser-
vices or how often they go out to a movie.
• Respondents’ answers may lack spontaneity • Factual questions about others. It’s usually
and authenticity. It is also possible that the an- better to interview the person directly than to
swers provided do not cover all the potential rely on the second-hand accounts of others,
replies. One solution is to use an open question because respondents’ knowledge of others’
to generate the categories (see Box 5.1). Another behaviour may be distorted or inaccurate
is to include “other” as a response option and (Beardsworth & Keil, 1997). However, if the re-
category and then invite respondents who searcher is interested in the respondent’s own
choose it to elaborate on their choice. Closed perceptions of how someone else acts or feels,
questions may irritate respondents who are questions on that topic are appropriate.
unable to find an answer category that they feel • Factual questions about an entity or event.
applies to them. In theory, all possible answers Sometimes respondents will act as informants
should be provided. about some phenomenon or event with which
• Respondents may differ in their interpretation they are familiar, particularly when informa-
of answers; for example, the understanding of tion is not readily available from other sources.
Asking people about what they witnessed at
a riot—how many people were there, the ap-
True variation proximate ages of people engaged in looting—
is an example. This sort of question can lead to
Variation due to error
problems because in everyday life most people
are not careful, systematic observers, and they
may have biases that affect their perceptions
True variation Variation due and recollections.
to error • Questions about attitudes. A five-point Likert
FIGURE 5.1 A variable with little error and one with scale is one of the most common formats for
considerable error measuring attitudes. Often the responses to a
5 | Survey Research: I nterviews and Questionnaires 99
Schuman and Presser (1981) conducted an experi- the answers they had received to the open-ended
ment to determine whether responses to closed question, and re-administered the questions to
questions could be improved by asking the ques- two large samples. The question (with revised re-
tions first in open-ended format and then develop- sponse categories) and the answers they received
ing categories of reply from respondents’ answers. were as follows:
They asked a question in both open and closed “This next question is on the subject of work.
formats about what people look for in work, using People look for different things in a job. Which one
different samples for each. They found consider- of the following five things do you most prefer in a
able disparities between the two sets of answers. job? [closed question] What would you most prefer
They then revised the closed categories to reflect in a job? [open question]”
With the revised form of the closed question, open question matched the categories offered in
Schuman and Presser found that a much higher the closed one. Also, the distributions were some-
proportion of the answers to the open question what different: for example, twice as many re-
corresponded to the closed categories. They spondents cited a feeling of accomplishment with
argued that the new closed question was superior the closed format as with the open one. Nonethe-
to its predecessor and also superior to the open less, the experiment demonstrated the desirabil-
question. However, it was still disconcerting to find ity of generating forced-choice answers from open
that only 58 per cent of the answers given to the questions.
100 PART II Quantitative Research
number of Likert items are aggregated to form What is this question seeking to tap? Car ownership?
a composite measure of an attitude. Access to a household car? Use of a company car? The
• Questions about beliefs. For example, in Hay’s ambiguity of the word “have” means that a “yes” answer
(2014) study of secularization in Canada (see could signify any of the three. If you want to find out
Box 1.3), one item used to measure moral au- whether a respondent owns a car, you need to ask spe-
tonomy pertained to the belief that “it is up cifically about ownership. Be as specific as possible.
to the individual to decide what is right and
wrong.” The response categories were: “Agree How would you answer it?
strongly” (5), “Agree somewhat” (4), “Neither Rule of thumb number three is to put yourself in the
agree nor disagree” (3), “Disagree somewhat” position of the respondent. Ask yourself the question
(2), and “Disagree strongly” (1). and try to work out a reply. If you do this, there is at
• Questions about knowledge. Sometimes ques- least a chance that the ambiguities inherent in a ques-
tions are designed to “test” respondents’ know- tion like “Do you have a car?” will become evident.
ledge in a certain area. For example, a poll For example, you may remember that you have access
sponsored by Historica Canada in 2015 deter- to a car that you don’t actually own. Putting yourself
mined that about one-quarter of the people in the position of the respondent can reveal the diffi-
surveyed did not know what year Confedera- culty of answering poorly constructed questions.
tion took place, and a similar number could not
identify Canada’s first prime minister (Ipsos Specific rules when designing
Reid, 2015). questions
More than 10 times per day _____ While there may well be symmetry between
5–9 times per day _____ the two ideas in this answer—pleasant work and
1–4 times per day _____ nice people—there is no necessary correspondence
A few times a week _____ between them. Pleasant work may be important
A few times a month _____ for people who are indifferent to their co-workers.
Almost never _____ For an example of a triple-barrelled question, see
Box 5.2.
Alternatively, you could simply ask respondents to
estimate the number of times per day they use their Avoid very general questions
cellphone to make a call. It may be tempting to ask a very general question
It’s also important to bear in mind that certain when what is needed is a response to a specific issue.
common words, such as “dinner,” mean different A question such as
things to different people. For some, dinner is a light
How satisfied are you with your job?
midday meal, whereas for others it is a substantial
evening meal. In such cases it is necessary to clearly seems harmless, but it lacks specificity. Does it
define the terms that are being used. refer to pay, working conditions, the nature of
the work, or all of these? Respondents are likely
Avoid long questions to vary in their interpretation of the question,
Most methodologists agree that long questions are and this will be a source of error. A favourite gen-
undesirable. Interviewees can lose track of long ques- eral question comes from Karl Marx’s Enquête
tions, and respondents completing a questionnaire ouvrière, a questionnaire sent to 25,000 French so-
may be tempted to skim or even skip over them. How- cialists and others (whether any were returned is
ever, Sudman and Bradburn (1982) suggested that this unknown). The final question (the one-hundredth!)
advice is more applicable to attitude questions than to reads:
those asking about behaviour. They argued that when
What is the general, physical, intellectual, and
the focus is on behaviour, longer questions can have
moral condition of men and women employed
certain positive features in interviews: for example,
in your trade? (Bottomore & Rubel, 1963,
they are more likely to provide memory cues and fa-
p. 218)
cilitate recall because of the time it takes to ask them.
By and large, however, it’s best to keep questions short. If you find it difficult to come up with precise
questions to replace very broad, general ones, you
Avoid double-barrelled questions might take inspiration from Blauner (1964), who
How satisfied are you with your pay and work- provides a very detailed and nuanced description
ing conditions? of job satisfaction among American factory workers
based on some 80 specific questions.
The problem here is obvious: people may be satis-
fied with one but not the other. Not only will the Avoid leading questions
respondent be unclear about how to reply, but the Leading or loaded questions push respondents in a
answer provided may pertain to pay, to conditions, or particular direction, for example:
to both.
Would you agree to cutting taxes, even though
The same rule applies to fixed-choice answers.
welfare payments for the neediest sections of
In Box 5.1, one of Schuman and Presser’s (1981) an-
the population might be reduced?
swers is:
Phrasing the question in this way is likely to make it
Work that is pleasant and people who are nice difficult for some people to agree, even if they do feel
to work with. that taxes are too high.
102 PART II Quantitative Research
Was the writing elegant, seamless, imaginative? This approach would also solve the additional
problem of the “triple-barrelled” question that asks
None of the response options provided by the about three distinct aspects of the writing, leaving the
publisher could be considered logical answers reader unable to express differing opinions of each.
to the questions that were posed. The questions It could be argued that readers would understand
would have been better presented as statements, that they were being asked to rate the book in terms
for example: of each attribute. The problem is that the impact of
a disjunction between question and answer is simply
Please indicate the quality of the writing in not known—which means that any conclusions
each of the following categories: drawn by the publisher might well be wrong.
Avoid questions that actually ask two they intended. There may be situations when it’s im-
(or more) questions possible to avoid using a negative, but questions like
the following should be avoided as far as possible:
The double- and triple-barrelled questions cited ear-
lier clearly transgress this rule, but there are also more Do you agree with the view that university
subtle versions of the problem. A question such as students should not have to take out loans to
finance their education?
Which party did you vote for in the 2015 fed-
eral election? Instead, the question should be asked in a positive
form; for example: “Do you find it acceptable that
provides an illustration. What if the respondent did
some university students have to take out loans to fi-
not vote in that election? The question is really two
nance their education?” Questions with double nega-
questions, which should be asked separately:
tives are never appropriate, because it’s impossible to
Did you vote in the 2015 federal election? be sure that respondents have interpreted them cor-
Yes rectly. The following provides an example:
No
Would you rather not drink non-alcoholic beer?
If YES, for which political party did you vote?
Minimize technical terms
Avoid questions that include negatives
Use simple, plain language and avoid jargon. Don’t
The problem with questions that include negative
ask a question like this:
terms such as “not” is that some respondents may
miss the negative and give answers opposite to what Do you sometimes feel alienated from your work?
5 | Survey Research: I nterviews and Questionnaires 103
The problem here is that many respondents may Don’t overstretch people’s memories
not be familiar with the term “alienated,” and even
if they are, they may not understand it in the same It might be useful to know exactly how many times
way. The same goes for acronyms and abbreviations: respondents have used their cellphones in the previ-
avoid them whenever possible. ous month, but few people will be able to give you an
accurate answer for such an extended period of time.
Ensure that respondents have For this reason it’s important to choose a suitable time
the requisite knowledge frame, such as per day in the case of cellphone use.
There is little point in asking about matters unfamil-
iar to respondents. For example, it’s doubtful that Carefully consider using “Don’t know”
meaningful data regarding opinions on foreign af- as a closed answer option
fairs can be gained from people who do not follow One area of controversy regarding closed questions
international political issues. is whether the answer set should include a “don’t
know” or “no opinion” option when asking ques-
Ensure that the answers provided tions concerning attitudes. The main argument for
for a closed question are balanced including a “don’t know” option is to ensure that
It’s important that the range of answers provided respondents will not be forced to express views that
for a closed question be properly balanced between they don’t actually hold. Many recommend offering
positive and negative options. A set of responses the “don’t know” option in the form of a filter question
like this to remove those respondents who have no opinion
on the topic. Filter questions are used to determine
Excellent _____
whether it is appropriate to ask certain questions of
Good _____
a particular respondent. The interviewer could ask
Borderline _____
the filter question, “Do you have an opinion on this
Poor _____
issue?” of everyone and then ask the second question
is weighted toward a favourable response. “Excel- only of those who say that they do have an opinion.
lent” and “Good” are both positive; “Borderline” is a The alternative argument in connection with the
neutral or middle position; and “Poor” is a negative “don’t know” option is that it offers an easy way out
response. A second negative response choice, such as for respondents. A series of experiments conducted
“Very poor” is needed. in the US found that many respondents who claim to
have no opinion on a topic do in fact hold one (Kros-
Ensure answers provided for a closed nick et al., 2002). Respondents with lower levels of
question do not overlap education are especially likely to select the “don’t
Care must be taken to ensure that answers to closed know” option; and later questions in a questionnaire
questions are mutually exclusive. For example, a (see the next section for question-order effects) are
question on the respondent’s age with a set of re- the most likely to elicit a “don’t know” response. The
sponses like this researchers concluded that data quality is not en-
hanced by the inclusion of a “don’t know” option and
20–30 _____ that it may prevent some respondents from express-
30–40 _____ ing an opinion that they do actually hold.
40–50 _____ Pay close attention to question order
50–60 _____
Quite a lot of research has been directed at finding
would confuse anyone with an age of 30, 40, or 50, out how responses are affected by asking questions
as these answers each fit into two of the supplied at different points in an interview schedule. Results
options. have varied and few if any consistent effects have been
104 PART II Quantitative Research
found. Mayhew (2000) tells an interesting anecdote experiments do not always reveal clear-cut effects,
on question order in a crime survey. By mistake, half even in cases where effects might legitimately have
of the respondents were asked the question “Taking been expected. Nonetheless, two general lessons
everything into account, would you say the police in have emerged:
this area do a good job or a poor job?” twice: once
early on, and once again later, in the context of ques- • Unless the study is designed to test q
uestion-
tions on contact with the police. Almost a quarter order effects, all respondents should receive
(22 per cent) of those respondents gave a more positive questions in the same order.
rating the second time. Mayhew suggests that as the • Researchers should be sensitive to the possible
interview wore on, respondents became more sensi- effects that a question could have on subse-
tized to crime-related issues and more sympathetic quent questions.
to the pressures on the police. As appealing as this
explanation sounds, it cannot account for the 13 per In addition, the following rules about question
cent who gave a lower rating the second time around. order are sometimes proposed:
That is quite a reliability issue. (See also Box 5.3.)
It’s hard to draw any general lessons from • Early questions should be directly related to
question-order research, at least in part because the announced research topic. This removes
Imagine that there are two pollsters: one who Set Two:
favours the federal New Democratic Party and 1. The recent free trade deal the federal govern-
another who favours the Conservatives. The key ment signed with the European Union will help
question in both their surveys is: “In the upcoming create jobs in Canada.
federal election, which party do you intend to vote 2. Canada’s mounting debt is mortgaging the
for?” That question would be followed by an alpha- future of its children.
betical list of the main political parties. Look at the 3. Lower taxes would benefit everyone, espe-
two sets of Likert items that follow (assume there cially hard-working Canadian families.
are five response options ranging from “Strongly 4. Our laws should protect law-abiding citizens,
agree” to “Strongly disagree”), and imagine that not criminals.
the items immediately preceded the voting ques-
For those who do not follow politics closely, the
tion. Try to determine whether these items would
first set would probably increase the number of re-
influence responses to the voting question and, if
spondents saying they would vote NDP, while the
so, how.
second would likely do the same for the number
Set One: declaring their intention to vote Conservative.
1. Canada must do more to reduce global warming. Although most pollsters avoid such blatant biases,
2. The environment should be the federal gov- sometimes more subtle influences on respondents
ernment’s top priority. go unnoticed. In any event, this is an interesting
3. Corporate taxes should be increased to pay for exercise because it illustrates the importance of
social assistance for Canada’s poorest families. question order in survey research: if it is necessary
4. Canada must do more to protect Indigenous to use these items, they ought to be presented
rights. after the voting question.
5 | Survey Research: I nterviews and Questionnaires 105
the risk that respondents will be put off by ap- the interviewer arrives at the scheduled point.
parently irrelevant questions asked at an early A different answer may be offered at the latter
stage in the interview. It also means that per- juncture because of question-order effects.
sonal questions about age, social background,
and so on should not be asked at the beginning Avoid provoking response sets
of an interview. This issue (see also Chapter 4) is especially relevant
• Questions likely to be important or meaning- to multiple-indicator measures, where respondents
ful to respondents should be asked early in the reply to a battery of related questions or items. A re-
interview, to stimulate their interest and atten- sponse set exists if responses are motivated by some-
tion. This suggestion may conflict with the thing other than the person’s actual feeling about the
previous one if questions about the research items. They can result from psychological predispos-
topic are not of interest to respondents. itions like acquiescence (a desire to be “cooperative”
• If possible, questions that may cause embar- or to please the researcher) or laziness, and in such
rassment or anxiety should be left for later in cases are evident when someone responds to a series
the schedule—although not to the very end. of items in the same way, regardless of the content of
Respondents should not leave the interview (or the items. To avoid acquiescence, researchers using
complete the questionnaire) with misgivings multiple-item measures may include some items
or negative feelings about the research. that logically call for opposite positions, so that if re-
• With a long schedule or questionnaire, ques- spondents are consistent in their attitudes, they can
tions should be grouped logically in related neither agree with all the items nor disagree with all
sections, and should not jump back and forth of them. This also requires the respondent to care-
from one topic to another. fully read each question. Similar procedures may be
• Within each group of questions, general ques- used to avoid the effects of laziness or boredom. For
tions should precede specific ones. If a specific example, when visiting a doctor’s office for the first
question comes first, the aspect of the general time, patients usually have to fill out a form with a
question that it highlights may be discounted long list of diseases, indicating whether they have ever
by respondents who feel they have already cov- suffered from them. Someone may begin carefully,
ered it. For example, if a question asking how reading the list and checking off “No, No, No, . . . ,”
people feel about their salary precedes a gen- but eventually tire of the task and just tick “No” for
eral question about job satisfaction, some re- the remaining items without reading them closely.
spondents may discount the issue of pay when Those who persist in choosing the same answer re-
responding to the job satisfaction question, gardless of the question may have to be removed from
since they have already addressed it. the study. Response sets may also stem from a desire
• It is sometimes recommended that questions to be liked or respected by the interviewer, so in-
dealing with opinions and attitudes precede structing interviewers to avoid both becoming overly
questions about behaviour and knowledge. friendly with respondents and appearing judgmental
This is because questions that tap opinions and about their replies can help. Using vignettes can also
attitudes may be affected by the latter sorts of help reduce answers motivated by social desirability.
questions. For instance, if a husband reports
that he does only 20 per cent of the house- Consider using vignette questions
work, making that statement can affect how he when appropriate
responds when asked if housework should be Closed questions that are used to examine ethical
shared equally between spouses. standards and beliefs may be employed in conjunction
• If a respondent provides an answer to a ques- with the vignette technique. It involves presenting
tion that is to be asked later in the interview, the respondents with one or more scenarios and asking
question should still be asked later on, when them how they would respond if confronted with the
106 PART II Quantitative Research
circumstances depicted in the scenario. For example, paid work for unpaid care?). The specificity of the
Kingsbury and Coplan (2012) used vignettes to exam- situation facing Jim and Margaret increases as the
ine how some Ontario mothers of preschool chil- vignette develops. The first question (a) does not
dren reacted to hypothetical accounts of their child’s say whether they are prepared to move; the second
shyness and aggression. The researchers hypoth- (b) says that they are; and in the last question (d) they
esized that mothers would look more favourably on have in fact moved and are facing a new dilemma.
gender-congruent behaviours (such as shyness in girls) Many aspects of the issues tapped by the ques-
than gender-incongruent ones (physical aggression in tions in Box 5.4 can be accessed through attitude
girls), especially if they held more traditional attitudes items. For example:
toward sex roles. The findings provided some support
for the hypotheses, although the results were mixed. When two heterosexual working spouses decide
Box 5.4 outlines a vignette designed to tease out that one of them should quit work to care for
respondents’ norms concerning several aspects of ailing parents, the wife should be the one to give
family obligations, including the nature of the as- up her job.
sistance required (direct involvement or simple pro- Strongly agree _____
vision of resources); geographical considerations; Agree _____
the choice between paid work and unpaid care; and Undecided _____
among heterosexual couples, the gender question Disagree _____
(should it be the man or the woman who gives up Strongly disagree _____
5 | Survey Research: I nterviews and Questionnaires 107
Jim and Margaret Robinson are a married couple in moving right now could have a bad effect on
their early forties. Jim’s parents, who live several their children’s education.
hundred kilometres away, have had a serious car What should Jim and Margaret do? Should
accident and they need long-term daily help. Jim they move or should they stay?
is their only son. He and his wife both could get • move
transfers to work nearer his parents. • stay
a. What should Jim and Margaret do? c. Why do you think they should move/stay?
• move to live near Jim’s parents d. Jim and Margaret do go to live near Jim’s
• have Jim’s parents move to live with parents. A year later Jim’s mother dies and his
them father’s condition gets worse so that he needs
• give Jim’s parents money to help them full-time care.
pay for daily care Should one of either Jim or Margaret take an
• let Jim’s parents make their own arrange extended leave from work to take care of Jim’s
ments father? IF YES: Who should, Jim or Margaret?
• do something else (specify) • Yes, Jim should give up his job.
• don’t know • Yes, Margaret should give up her job.
b. In fact, Jim and Margaret are prepared • No, neither should give up their jobs.
to move and live near Jim’s parents, but • don’t know/depends
teachers at their children’s school say that Source: Adapted from Finch (1987, p. 108).
The advantage of the vignette over this kind of It is also difficult to establish how far respondents’
attitude item is that it anchors the choice in a more answers reflect their own normative views or indeed
realistic situation and thus reduces the risk of an how they themselves would act when confronted
unreflective reply. Finch (1987) also argued that on with the kinds of choices outlined in the scenarios.
a sensitive topic like this, some respondents may Often what people say they would do in a particu-
feel uncomfortable or fear that they will be judged lar situation turns out to be very different from what
on their replies. The fact that the vignette is about they actually do. People are not necessarily being
other people (and imaginary ones at that) creates a dishonest—predicting behaviour, whether our own
certain distance between the questioning and the re- or someone else’s, is often more difficult than we
spondent, and hence the chance for a more candid would imagine.
answer. One obvious requirement of the vignette
technique is that scenarios must be believable. Run a pilot study
Finch also pointed to some limitations with this It is always desirable to conduct a pilot study before
style of questioning, however. It may be impossible collecting data from respondents: not just to ensure
to establish what assumptions are being made by that individual questions operate well but also to
the respondents about the characters in the scenario ensure that the research instrument as a whole is ap-
(such as their age, ethnicity, and number of children propriate. Pilot studies may be particularly crucial in
at home) and the significance of those assumptions research using questionnaires, since there will be no
for the validity and comparability of people’s replies. interviewer present to clear up any confusion.
108 PART II Quantitative Research
Here are some specific uses of pilot studies in • A pilot study may also offer an opportunity to
survey research: consider the overall structure of the study. If
the questions do not flow well, it may be neces-
• If the study is going to employ mainly closed sary to move some of them around.
questions, a researcher can pose open ques-
tions in pilot qualitative interviews and then The pilot should not be carried out on people who
use them to generate the fixed-choice answers. may become members of the sample in the full study,
• Piloting an interview schedule can give inter- since participation in the pilot could affect their re-
viewers some experience in using it and help sponses in the study proper. If the population you
them develop confidence. are investigating is small, you don’t want to be forced
• If everyone (or virtually everyone) who an- to exclude certain people, since that could affect the
swers a question gives the same answer, the re- representativeness of the sample. If this is an issue,
sulting data are not likely to be useful. A pilot do the pilot study on respondents from a different
study gives you an opportunity to identify but similar population. Beagan (2001) did this in her
such questions and modify them as necessary study of medical students.
to elicit more varied answers.
• In survey interviews, it may be possible to Consider using existing questions
identify questions that make respondents feel Finally, it is sometimes possible to use questions that
uncomfortable and to detect any junctures have been employed by other researchers for at least
where interest tends to flag. part of the questionnaire or interview schedule. This
• Questions that respondents tend to skip or may seem like stealing, but it is a legitimate research
have difficulty understanding (more easily practice so long as the original source is cited. One
recognizable in an interview than in a ques- advantage of using existing questions is that they
tionnaire context) should become apparent. If may have already been piloted and tested for their
certain questions are frequently skipped, it may reliability and validity. Robinson et al. (1999), for ex-
be because of unclear or threatening phrasing, ample, provide an extensive set of survey questions
poorly worded instructions, or confusing po- that have been used to measure political attitudes,
sitioning in the interview schedule or ques- along with measures of their reliability and validity.
tionnaire (see Box 5.5). Whatever the cause, A further advantage is that existing questions
missing data are undesirable and a pilot study allow for comparisons with other research. For in-
may be very helpful in identifying problems. stance, in psychology it is quite common to use
5 | Survey Research: I nterviews and Questionnaires 109
How would you fix this questionnaire? Now there are just a few more questions.
“Hello. I am taking a sociology course on research 7. If one were planned, would you fight having an
methods and would like to ask you some ques- AIDS hospice on your street?
tions. Would that be ok?” Yes No
1. Were you ever scared that you were 8. How does your spouse feel about AIDS victims?
HIV-positive? Compassionate Sympathetic
Yes No Concerned Angry Don’t know
For questions 2 to 6, please answer Agree or 9. Elizabeth Taylor has raised much money for
Disagree. AIDS research. Should people be doing more
fundraising in this area?
2. As a religious person, I feel sorry for AIDS victims. Yes No
3. I think victims of any disease deserve com-
10. Do you know any AIDS victims?
passion.
Yes No
4. I think the new legislation will be a boon to
AIDS victims. Finally I need a few facts for comparison purposes.
5. The government should not allocate more
11. How old are you?
money to AIDS research.
12. Did YOU graduate from high school?
6. People with AIDS and other sexual diseases
Yes No
should be quarantined.
13. Did you EVER have yourself tested for AIDS?
If you agree, has your viewpoint changed since the
Yes No
1980s when AIDS first widely emerged?
Thank you for your cooperation.
Yes No
standardized personality and intelligence tests; these examining questions used by others may provide
permit comparisons between different samples and ideas about how best to approach your own ques-
allow different researchers to make consistent meas- tions, even if you eventually decide to modify those
urements. Similarly, some of the questions used in used by other researchers.
Bell et al.’s (2009) study of the heritability of political
orientations were taken from earlier studies. They
were either used verbatim or slightly revised to make
Issues related to conducting
them suitable for the sample to which they were ad- interviews
ministered. Walklate (2000, p. 194) described how in In addition to crafting good questions, there are
developing a survey instrument for victims of crime, other issues to be considered when doing research
she and her colleagues used “tried and tested ques- using structured interviews and questionnaires.
tions taken from pre-existing criminal victimization These are addressed below.
surveys amended to take account of our own more
localized concerns.” Existing questions can also be Interview contexts
used to investigate whether change has occurred In a traditional interview, an interviewer stands
over time by comparing current results with those or sits in front of the respondent, asks a series of
arrived at in earlier time periods. At the very least, questions, and writes down or keys in the answers.
110 PART II Quantitative Research
Checklist
Checklist of issues to consider for a structured interview schedule or questionnaire
☐☐ Is a clear and comprehensive introduction to ☐☐ Are questions relating to the research topic
the research provided for respondents? asked near the beginning of the interview or
☐☐ Are there any questions used by other re- questionnaire?
searchers that would be useful? ☐☐ Have the following been avoided?
☐☐ Will the questions provide answers to all the • ambiguous terms in questions or response
research questions? choices
☐☐ Are there any questions not strictly relevant • long questions
to the research questions that could be • double-barrelled questions
dropped? • very general questions
• leading questions
☐☐ Has the questionnaire been pre-tested with
• questions that include negatives
some appropriate respondents?
• questions using technical terms
☐☐ If a structured interview schedule is used, are
the instructions clear? For example, with filter ☐☐ Do respondents have the knowledge required
questions, is it clear which question(s) should to answer the questions?
be omitted? ☐☐ Is there an appropriate match between ques-
☐☐ Are instructions about how to record re- tions and response choices?
sponses clear (for example, whether to tick ☐☐ Are the response choices properly balanced?
or circle; whether more than one response is ☐☐ Do any of the questions depend too much on
allowable)? respondents’ memories?
☐☐ Has the number of open questions been
limited? If using a Likert scale approach:
☐☐ Can respondents indicate levels of intensity ☐☐ Are some items that have to be
in their replies, or are they forced into “yes or r everse-scored included, in order to identify
no” answers? response sets?
☐☐ Have questions and their answers been kept ☐☐ Is there evidence that the items really do
on the same page? relate to the same underlying cluster of atti-
☐☐ Have socio-demographic questions been tudes, so that the items can be aggregated?
left until near the end of the interview or ☐☐ Are the response choices exhaustive and not
questionnaire? overlapping?
However, there are several possible departures from as well, and to administer a structured interview to a
this pattern. focus group would be very unusual. In most survey
research a specific individual is the object of ques-
More than one interviewer or interviewee tioning by one interviewer. Indeed, it is usually ad-
Multiple interviewers are rare in social research be- visable to discourage as far as possible the presence
cause of the considerable cost involved in dispatch- or intrusion of others during the interview. Studies
ing more than one person to conduct an interview. in which more than one person is being interviewed
In the case of focus groups it is more common, but tend to be conducted by qualitative researchers, al-
in that instance there is more than one interviewee though that is not always the case. For example,
5 | Survey Research: I nterviews and Questionnaires 111
Pahl’s (1990) study of patterns in the control of part of the study. Since lack of a phone is most
money among couples employed structured inter- likely to be a feature of poorer households, the
viewing first of the couples and then of husbands and potential for sampling bias exists. Also, many
wives separately. people choose to be ex-directory (they pay to
have their telephone numbers not listed in a
Alternatives to in-person interviews telephone book) or they use only cellphones
It is customary in academic social research not in- and for that reason are not listed. This pre-
volving national samples to use face-to-face inter- sents a problem if the researcher is using a
views. However, telephone interviewing is the norm directory to select respondents. A solution
in fields such as market or government research. to the latter problem is random-digit dial-
Telephone interviews have several advantages over ling. With this technique (used in the annual
the face-to-face variety: A lberta Survey conducted at the University of
A lberta, as well as in the General Social Survey
• They are far cheaper and quicker to adminis- administered by Statistics Canada), a com-
ter than in-person interviews, which may re- puter randomly selects telephone numbers
quire the interviewers to spend a great deal of within a predefined geographical area. This
time travelling to meet with respondents; this method can give you access to ex-directory
is a particular problem when a sample is geo- households, though it’s important to note that
graphically dispersed. in some areas it is prohibited by law to contact
• Telephone interviews are easier to supervise cellphones in this way.
than the personal kind. This is a particular ad- • Respondents with hearing impairments are
vantage when several interviewers are calling likely to find telephone interviews more diffi-
from the same location at the same time, since cult than personal interviews.
supervisors may be able to detect interviewer • A telephone interview is unlikely to be sustain-
errors, such as rephrasing questions or probing able beyond 20 to 25 minutes, whereas per-
inappropriately. Face-to-face interviews can sonal interviews can be conducted for longer
be recorded for the purpose of assessing data periods of time (Frey, 2004).
quality, but the recording of interviews raises • There is some evidence that telephone inter-
issues of confidentiality and so must be treated views fare less well than personal interviews in
with caution. obtaining data on sensitive issues such as drug
• Telephone interviewing can also reduce bias and alcohol use, income, tax payments, and
arising from the characteristics of the inter- health. The evidence is not entirely consistent
viewers or interviewees (for example, gender, on this point, but when many questions of this
class, and ethnicity). The remoteness of the kind are to be asked, a personal interview may
interviewer in telephone interviewing removes be superior (Shuy, 2002).
this potential source of bias to a significant • Telephone interviewers cannot see the re-
extent, since the interviewers’ “race” and gen- spondents and therefore cannot respond to
eral appearance cannot be seen, although their signs of puzzlement or unease on their faces.
gender is usually apparent. In a personal interview, the interviewer may
respond to such signs by restating the question
On the other hand, telephone interviewing suffers or attempting to clarify its meaning, though
from certain limitations when compared to the per- this has to be handled in a standardized way as
sonal interview: much as possible. A further issue is that inter-
viewers may be able to collect subsidiary infor-
• People who do not have a telephone or other- mation, such as whether the home in which the
wise cannot be contacted by phone cannot be interview was conducted is in need of repair.
112 PART II Quantitative Research
That sort of information cannot be acquired (CAPI) and computer-assisted telephone inter-
over the telephone. viewing (CATI). The use of CAPI has been growing
• Frequently, the target of the research is a specific lately, mainly because of the increased portability
individual in a household or firm: for example, and affordability of laptop computers and other elec-
the person in a certain role or position, or some- tronic devices, and the availability of quality soft-
one with particular characteristics. It is probably ware packages.
more difficult to ascertain by telephone inter- With computer-assisted interviewing, the ques-
view whether the correct person is replying. tions in the interview schedule appear on the screen,
• The telephone interviewer cannot employ the interviewer keys in the reply, and the computer
visual aids such as “show cards” from which shows the next question. This approach is espe-
respondents select their replies (see Box 5.6). cially useful when filter questions are asked. In the
Similarly, diagrams or photographs cannot be 2015 Canadian Election Study (funded by Elections
used. Canada), respondents were asked the filter question,
“The federal election was held on Monday, O ctober 19.
There is some evidence that the quality of the data In any election, some people are not able to vote be-
gathered through telephone interviews is inferior to cause they are sick or busy, or for some other reason.
that produced by face-to-face interviews. According Others do not want to vote. Did you vote in the recent
to Holbrook et al. (2003), a series of experiments in the federal election?” Only those who said yes were then
US using long questionnaires found that respondents asked, “Which party did you vote for?” Once the re-
interviewed by telephone were more likely to express sponse to the filter question has been keyed in, the
no opinion or give “don’t know” as an answer; give the computer will automatically go to the next relevant
same answer to a series of linked questions; express question, eliminating inappropriate items where ne-
socially desirable opinions; be apprehensive about the cessary. This removes the possibility of interviewers
interview; and be dissatisfied with the time taken by inadvertently asking meaningless questions or fail-
the interviews (even though they were shorter than ing to ask some that should be asked. In this way,
the face-to-face kind). Also, telephone interviewees computer-assisted interviewing improves standard-
tend to be less engaged in the interview process. ization in the asking and recording of questions.
Computer-assisted interviewing makes it easier to
Computer-assisted interviewing perform other tasks as well. For example, sometimes
Today it is common to use computers in the inter- it is beneficial to randomize the order in which cer-
view process. There are two main formats for such tain items are presented to the respondent in order
interviews: computer-assisted personal interviewing to determine whether the responses given depend
on the order in which the items are listed. Automatic
randomization was used in the Goyder et al. (2003)
study of occupational prestige done in southwestern
BOX 5.6 Two show cards Ontario. As the authors put it: “The deck was elec-
Card 4 (age) Card 6 (for various items) tronically shuffled for each new respondent.” An-
a. Less than 20 1. Strongly agree other advantage of computer-assisted interviews is
b. 20–29 2. Agree
that if interviewers are out in the field all day, the
data can be sent to the research office electronically.
c. 30–39 3. Undecided
d. 40–49 4. Disagree Using online (email) personal interviews
e. 50–59 5. Strongly disagree
Online interviews run a higher risk of respondent
f. 60–69 dropout than other forms do. However, Mann and
g. 70 and over Stewart (2000, pp. 138–139) have suggested that it
is possible to develop a relationship of mutual trust
5 | Survey Research: I nterviews and Questionnaires 113
when using them. This can be accomplished by regu- replies are often more detailed than in face-to-
larly sending messages to respondents assuring them face interviews; and
that their responses are helpful and significant, espe- • online interviewees’ answers tend to be more
cially since online interviewing is still an unfamiliar considered and grammatically correct because
experience for most people and takes longer than they have more time to ponder and tidy up
other forms. It is worth the trouble because online answers. Whether this is a positive feature is
interviews make it relatively easy for the researcher debatable. There is the obvious advantage of a
to go back to interviewees for further information or “clean” transcript, but there may be some loss
reflections—something difficult to do with face-to- of spontaneity.
face interviews.
A further issue for the online personal interviewer On the other hand, Curasi found that some online
is whether to send all the questions at once or to con- responses were short on detail, perhaps because re-
duct the interview on a question-followed-by-reply plies had to be typed, not just spoken. The full signifi-
basis. The problem with sending all at once is that cance of the difference between email and face-to-face
respondents may read through them and reply only interviewing is still being explored by researchers.
to the most interesting ones. Bampton and Cowton Webcam technologies like Skype may offer fur-
(2002) reported that conducting email interviews by ther possibilities for online personal interviews,
sending questions in small batches takes the pressure making the online interview similar to the telephone
off interviewees to make a quick reply, gives them the version and quite comparable to an in-person inter-
opportunity to provide considered replies (a lthough view because those involved in the exchange can see
there can be a loss of spontaneity), and gives the each other. However, one of the main advantages
interviewers greater opportunity to respond to inter- of the email interview is lost: respondents’ answers
viewee answers. need to be transcribed.
There is evidence that prospective online inter-
viewees are more likely to agree to participate if agree- Conducting interviews
ment is solicited prior to sending the actual questions. Know the interview schedule
Another way to encourage respondents to participate Before interviewing anybody, an interviewer should
is for the researcher to use some form of self-disclosure, have a thorough knowledge of the interview sched-
such as directions to a website that contains detailed ule. Interviewing can be stressful, and it is possible
contact information. It also helps to provide personal that under strain standard interview procedures can
material about the researcher (such as a photo), as well cause interviewers to get flustered and leave questions
as information relevant to the research topic (Curasi, out or ask the wrong questions. All interviewers need
2001; O’Connor & Madge, 2001, 2003). Such steps are to be fully trained to reduce interviewer variability in
necessary because unsolicited emails (“spam”) are asking questions, a potential source of error. Fowler
often seen as a nuisance and can result in an immedi- (1993) cited evidence to suggest that a training period
ate refusal to take the message seriously. shorter than one full day is rarely sufficient.
Curasi (2001) conducted a comparison in which
24 interviews carried out through email corres- Introducing the research
pondence were contrasted with 24 parallel face-to- Prospective respondents have to be offered a credible
face interviews. The interviews were concerned with rationale for participating in the research, in par-
shopping online. She found that: ticular for giving up their valuable time. This aspect
of conducting interviews is of special significance
• face-to-face is better than online for main- at a time when survey response rates appear to be
taining rapport with respondents; declining. The introductory rationale may be either
• because greater commitment and motivation spoken by the interviewer or printed for the re-
are required to complete an online interview, spondent to read. It is usually spoken if interviewers
114 PART II Quantitative Research
“cold call” potential respondents at their homes, is important not to take it too far. Too much rapport
either in person or by telephone. A written ration- can result in the interview going on too long and can
ale is common to alert respondents that someone also lead respondents to tailor their answers to please
will be contacting them to request an interview. the interviewer. Thus, achieving the appropriate rap-
In many cases, respondents may be presented with port between interviewer and respondent is a deli-
both modes—for example, researchers will send re- cate balancing act.
spondents a letter inviting them to participate in a
study and then call or speak to them in person. Topics and issues to include in an
introductory statement
Rapport In an introductory statement to a prospective
As suggested earlier, interviewers need to establish interviewee:
a rapport with respondents. A relationship must be
forged fairly quickly to encourage respondents to • Clearly identify yourself.
participate in and persist with the interview. While • Identify the auspices under which the research
this injunction essentially invites the interviewer to is being conducted (for example, a university, a
be friendly with respondents and put them at ease, it government agency).
5 | Survey Research: I nterviews and Questionnaires 115
• If you are a student doing research for a thesis, Some general tactics with regard to probes are as
make that fact clear. follows:
• Indicate what the research is about in broad
terms and why it is important, and give an • If further information is required, usually in
indication of the kind of information to be the context of an open-ended question, stan-
collected. dardized probes should be employed, such as
• Indicate how the respondent has been selected “Can you say a little more about that?” or “Are
(for example, by a random process, by conven- there any other reasons why you think that?”
ience, because of special characteristics). • If, with a closed question, the respondent re-
• Provide reassurance about the confidentiality plies in a way that does not match one of the
of any information given by the respondent. pre-designed answers, the interviewer should
• Explain that participation is voluntary. repeat the fixed-choice alternatives and make
• Reassure respondents that they will not be it clear that the answer must be chosen from
identified or identifiable. This can usually be those provided.
achieved by pointing out that when the data • When the interviewer needs to know about
are aggregated or analyzed at the group level, something that requires quantification, such
individual participants cannot be identified. as the number of visits made to a doctor in
• Give the respondent the opportunity to ask the last four weeks or the number of banks in
questions; be sure to provide a contact tele- which the respondent has accounts, but the
phone number or email address. If in person, respondent answers in general terms (“quite
simply ask if the respondent has any questions. often” or “I have several”), the interviewer
needs to persist for a more precise answer.
These suggestions are also relevant to the cover- With fixed-response surveys, this will usually
ing letter that should accompany online and mailed entail repeating the response options. The
questionnaires. The latter should also include a interviewer should not suggest an answer on
stamped, pre-addressed return envelope. the basis of the respondent’s reply since the re-
spondent may not be comfortable with the idea
Probing of disagreeing with the interviewer.
A highly problematic area for researchers em-
ploying a structured interview method is probing In a face-to-face interview, the interviewer may
respondents who need help with their answers. This use “show cards” (see Box 5.6) instead of reading
may occur if respondents do not understand the out a series of fixed-choice alternatives. Sometimes
question and then struggle to provide an adequate called “flash cards,” they display all the answers from
answer, or if they do not provide a complete answer which the respondent is to choose and are handed to
and have to be probed for more information. For ex- the respondent at different points in the interview.
ample, in the 2013 Alberta Survey, respondents were There are several situations in which it may be bene-
asked “What is your religion, if any?” If they an- ficial to use show cards:
swered “Christian,” the probe available for all inter-
viewers was: “Any particular denomination?” • Sometimes there is a very long list of possible
A potential problem with some probes is that answers; for example, respondents may be
some interviewers may be more inclined to use it asked about which social media sites they use
than other interviewers, and the probe itself may most frequently. To read out a list of sites would
affect the response given, which could lead to reli- be tedious and it is probably better to hand the
ability problems. A bigger problem arises if different respondent a list from which to choose.
interviewers use different probes, although proper • Some people are not keen to divulge personal
organization and training should preclude that. details such as their age or income. One way of
116 PART II Quantitative Research
Advantages of the questionnaire over especially true with questionnaires, since there is no
the structured interview interviewer to help respondents with questions they
cannot understand.
Cheaper, quicker, more convenient
to administer Greater risk of missing data
Interviewing is expensive and time-consuming. The Incomplete questionnaires are more common. It is
cost is obviously less pronounced when compared also easier for respondents to decide not to answer a
to telephone and email interviews, but even there question when they are on their own than when they
the questionnaire enjoys cost advantages. A class of are with an interviewer; questions that respondents
400 students can fill out questionnaires in one class find boring or irrelevant are especially likely to be
period, and the number of people accessing a ques- skipped. Questionnaire respondents are more likely
tionnaire online is virtually unlimited. Even with a than interview participants to become tired of an-
team of telephone interviewers, it takes a long time to swering questions that are not fully salient to them,
conduct personal interviews with such large sample and to abandon the project entirely. Put positively,
sizes. However, it is important to remember that when a research issue is important to the respondent,
questionnaires may not be filled out immediately a high response rate is possible. This means that
(respondents usually complete them at their conven- when the questions are highly relevant, a question-
ience). In many situations it is necessary to send out naire may be a good choice, especially because of its
follow-up letters and/or duplicate questionnaires to much lower cost.
those who fail to complete them in a timely manner.
Cannot probe
Absence of interviewer effects There is no opportunity to probe respondents to elab-
We have already noted that interviewer characteris- orate on an answer. However, this problem mainly
tics may affect respondent answers. Obviously those applies to open questions, which are not widely used
sorts of effects are not an issue with questionnaires, in questionnaire research.
since there is no interviewer present. Similarly, with
a questionnaire, no one is there to read the questions Difficult to ask a lot of questions
to the subject in the wrong order, to present them in Long questionnaires are rarely feasible because of the
different ways to different respondents, or to state the risk of respondent fatigue. They may also discourage
items with varying emphases from person to person; prospective respondents from participating at all.
the problems those practices can cause are precluded.
Probably of greater importance is the tendency Difficult to ask other kinds
for people to exhibit a social desirability bias when of questions
an interviewer is present, giving “politically correct” Long or complex questions should be avoided as far
rather than genuine responses. Research summar- as possible, since some respondents find them diffi-
ized by Tourangeau and Smith (1996) strongly sug- cult to follow.
gests that respondents tend to report more drug use
and alcohol consumption and higher numbers of Questionnaire can be read
sexual partners and abortions when responding to as a whole
questionnaires than in structured interviews. When respondents are able to read the whole ques-
tionnaire even before answering the first question,
Disadvantages of questionnaires none of the questions asked is truly independent of
versus structured interviews the others. It also means that the researcher cannot
be sure that questions have been answered in the cor-
Cannot explain the question rect order, raising the possibility of q
uestion-order
It’s always important to ensure that the ques- effects, although some software packages can mini-
tions asked are clear and unambiguous, but this is mize this problem.
118 PART II Quantitative Research
Not appropriate for some kinds embedded questionnaire, respondents must select
of respondents and/or type their answers. To return the question-
naire, it must be attached to a reply email.
Respondents whose literacy is limited or whose fa- The chief advantage of the embedded question-
cility with the language used is restricted may not be naire is that it requires less computer expertise.
able to answer a questionnaire. The second of these Knowing how to read and then return an attachment
difficulties cannot be entirely overcome with an requires a certain facility with online communica-
interview, but the difficulties are likely to be greater tion that is still not universally possessed. Also,
with a questionnaire. the recipients’ operating systems or software may
not be able to handle attachments, or respondents
One last problem: Who filled may refuse to open them because of concerns about
out the questionnaire? viruses. On the other hand, since most email soft-
With mailed or online questionnaires, the researcher ware allows only limited formatting, the appearance
can never be sure whether the questions were an- of embedded questionnaires tends to be rather dull
swered by the designated respondent or someone and featureless, although this situation is rapidly
else. It is also impossible to prevent other members changing. Furthermore, it is slightly easier for re-
of a household (for example) from helping the re- spondents to type answers into an attachment that
spondent answer the questions. uses well-known software such as Microsoft Word
because if the questionnaire is embedded in an email
Online social surveys the alignment of questions and answers may be lost.
As mentioned, the number of surveys being adminis-
tered online has grown considerably. There are both Web surveys
pros and cons with online surveys as compared to Web surveys invite prospective respondents to visit
mailed questionnaires (see Box 5.7). There is also a a website where the questionnaire can be found and
crucial distinction between surveys administered by completed online. The Web survey has an import-
email (email surveys) and surveys administered via ant advantage over the email survey in that it can
the Web (Web surveys). In the case of the former, use a much wider variety of embellishments, such
the questionnaire is sent via email to a respondent, as “radio buttons” and pull-down menus of possible
whereas with a Web survey, the respondent is answers. With open questions, the respondent is in-
directed to a website in order to answer it. vited to type directly into a boxed area.
However, the advantages of the Web survey go
Email surveys beyond appearance and format. As with interview
With email surveys it is important to distinguish software, questionnaires can be designed so that
between embedded and attached questionnaires. when there is a filter question, it skips automatically to
When the questions are embedded in the body of the the next appropriate question. The questionnaire can
email, there may be an introduction to the question- also be programmed either to show only one question
naire followed by a graphic that partitions the intro- at a time or to allow the respondent to scroll down
duction from the questionnaire itself. Respondents and look at all the questions in advance. If a question
may be asked either to indicate their replies using is missed, the software may provide a prompt telling
simple notation, such as an “X,” or to delete alterna- the respondent that it was not answered and that a
tives that do not apply to them. If a question is open, response is needed. Finally, respondents’ answers
they are asked to type in an answer. When finished, can be automatically programmed to download into
they simply hit the reply button to return the com- a database, eliminating the daunting task of coding
pleted questionnaire. a large number of questionnaires. Software packages
An attached questionnaire arrives as an attach- designed to produce questionnaires with all the fea-
ment to an email that introduces it. As with the tures just described are now widely available.
5 | Survey Research: I nterviews and Questionnaires 119
This box summarizes the main advantages and dis- population still differs in significant ways from
advantages of online surveys, both email and Web the non-online population, it remains a difficulty.
varieties, compared to mailed questionnaires. • Confidentiality and anonymity issues. It is
Note that all three share one disadvantage rela- standard practice for survey researchers to
tive to personal and telephone interviews: namely, indicate that replies will be kept anonymous.
that the researcher can never be certain of who is However, with email surveys the addresses of
actually answering the questions. the respondents are known, which may make it
difficult for people to believe that their identity
Advantages really will be protected.
• Low cost. Even though mailed questionnaires • Multiple replies. With Web surveys, there is a
are inexpensive to administer, online surveys risk that some people may complete the ques-
cost even less. tionnaire more than once, although software is
• Faster response and processing. Online surveys available that can reduce that possibility. There
can be returned considerably faster than mailed is much less risk of this with email surveys.
questionnaires. Researchers also appreciate
the automatic skipping of irrelevant questions
that is possible when they use filter questions,
as well as the opportunity for immediate down-
loading of replies into a database.
• Fewer unanswered questions. There is evi-
dence that online questionnaires are com-
pleted with fewer unanswered questions than
cnythzl/iStockphoto
mailed questionnaires.
• Better response to open questions. Open ques-
tions are more likely to be answered online,
and online replies are usually more detailed.
Have you ever participated in an online survey,
Disadvantages either through email or via the Web? If so, were you
• Low response rate. Typically, response rates concerned about a possible lack of anonymity? Did
this affect how you answered the questions?
for online surveys are lower than those for
comparable mailed questionnaires.
Sources: Adapted from Cobanoglu et al. (2001); Kent and Lee (1999);
• Restricted to online populations. This restric- S chaeffer and Dillman (1998); Sheehan and Hoy (1999); Sinclair et al.
tion should ease over time, but since the online (2012); Tse (1998).
Web surveys are also advantageous in that they such as people with eating disorders or extreme pol-
may give researchers access to unique populations, itical views (Wright, 2005).
such as people who share a particular interest. Web Potential respondents need to be directed to the
surveys are particularly useful for reaching stigma- website containing the questionnaire. One way to do
tized populations who are difficult to reach offline, this is to email them, as Andrews et al. (2007) did
120 PART II Quantitative Research
than one answer? If not, this should be indicated in of adolescent risk behaviour that produced rich nar-
the instructions, for example: ratives of the participants’ everyday lives. Sometimes
this method is supplemented by a personal interview
(Please choose the ONE answer that best rep-
in which the researcher will follow up on points raised
resents your views by placing a tick in the ap-
in the diary: for example, by asking the diarist what
propriate box.)
he or she meant by certain remarks. Diary data may
If this is not made clear and some respondents also be used in conjunction with information derived
choose more than one answer, their replies cannot from surveys, as in Robinson and Lee’s (2014) examin-
be counted and will have to be treated as if they were ation of the effect of new information technologies on
missing. If choosing more than one category is ac- people’s social lives and mass media use.
ceptable, this too must be made clear, for example: Corti (1993) distinguished between “structured
diaries” and “free-text diaries.” Either can be em-
(Please choose ALL answers that represent
ployed by quantitative as well as qualitative re-
your views by placing a tick in the appropriate
searchers. Structured diaries, such as those used by
boxes. You may tick more than one box for each
O’Sullivan et al. (2006) to examine sexual practices
question—choose as many as apply to you.)
among college students, have the general appearance
It is a common error for such instructions to be of a questionnaire with largely closed questions. At
omitted and for respondents to be unsure about how the end of each day the students in the study indicated
to reply. whether they had engaged in sexual activity that day,
and if they had, they used a checklist to mark which
Keep question and answers together sexual activity had been performed, whether they
This simple rule may seem obvious, but it is often used a condom, how safe from HIV infection they felt
transgressed: a question, or the question and the an- during the sexual encounter, and so on.
swers that accompany it, should not be split between Sullivan’s (1996) research on the domestic division
two separate pages. A common error is for a question of labour provides another illustration of a structured
to appear at the bottom of a page and the responses diary. The kind of diary Sullivan used is often de-
on the next page. The danger here is that the re- scribed as a “time-use” diary because it is designed for
spondent will either (a) fail to answer the question or recording, more or less at the time of the actual behav-
(b) pay insufficient attention to the question and give iour, how long people engage in activities such as food
a superficial answer (this problem is especially likely preparation, child care, or eating. Sullivan also asked
when a series of questions with a common answer couples to record on a five-point scale the amount of
format is being used, as with a Likert scale). enjoyment they derived from those kinds of activities.
Recording events as they happen is often con-
Researcher-driven diaries as a sidered a more accurate approach than recording
form of questionnaire them at the end of the day, largely because there is
When the researcher is interested in precise ac- less risk of forgetting. However, it is more intrusive
counts of behaviour, what Elliott (1997) called the than end-of-day reporting or answering a question-
researcher-driven diary can be used as a kind of naire because it is a constant interruption and may
questionnaire. The researcher asks participants to even change people’s behaviour. For example, one
record their perceptions, feelings, or actions with can become preoccupied with the behaviours under
regard to certain experiences shortly after they occur. study and thus engage in them in a more thoughtful
Laurenceau and Bolger (2005) used the diary method or premeditated way.
to study marital and family processes. Hessler et al. Sometimes diaries can be constructed as part of
(2003) used researcher-driven diaries in an email study a larger structured interview. This was the approach
122 PART II Quantitative Research
taken in the 2015 Time Use study, which was part Advantages and disadvantages of the diary
of the General Social Survey conducted for Statistics as a method of data collection
Canada. A 24-hour diary was produced by asking
people what they were doing on a certain day, start- The studies illustrating the use of diaries discussed
ing at 4:00 a.m. The first question was: “Last [day of above suggest potential advantages:
the week] at [4:00 a.m.], what were you doing?” (Most
people say “sleeping,” and are later asked how much • When fairly precise estimates regarding fre-
time they spent sleeping, and what time they went quency and/or amount of time spent in differ-
to sleep.) Sometimes questions such as “Where were ent forms of behaviour are required, the diary
you?” or “Who was with you?” are asked as well. may provide more valid and reliable data than
An example of a free-text diary is provided in questionnaires.
Crook and Light’s (2002) study in which university • When information about the sequencing of
students were asked to keep a diary for a week, div- different types of behaviour is required, it is
ided into 15-minute intervals, of the different kinds likely to perform better than questionnaires or
of study and learning activity in which they engaged, interviews.
where the activity took place, and the study resources • A diary is useful for collecting data on behav-
used (for example, the library). The various activities iour that is personally sensitive, such as sexual
were grouped into three types: classes, private study, activity.
and social study (that is, study with a peer). They were
able to show the very different amounts and patterns On the other hand, the diary method may present
of study typically undertaken during a day. the following problems:
Using free-text recording of behaviour carries
the same kinds of coding problems as those associ- • Diaries tend to be more expensive than per-
ated with open-ended survey questions—namely, the sonal interviews because of costs related to
time-consuming nature of the exercise and the risk recruiting diarists and ensuring that they are
of introducing error while coding answers. However, using their diaries properly.
the free-text approach is less likely to be problematic • There is a risk of attrition as diarists tire of the
when, as in the case of Crook and Light’s (2002) re- task, which raises reliability issues.
search, diarists are given instructions about what is • Failure to record details quickly enough can
required and the behaviours of interest are specific and result in errors or omissions.
concrete. It would be much more difficult to code free-
text entries relating to more general behaviours such as However, diary researchers such as Sullivan
the domestic activities studied by Sullivan (1996). argue that the information collected through diaries
Corti (1993) recommended that people doing re- is more accurate than that collected through inter-
search involving diaries should: views or questionnaires on the same topic.
Many important studies have been conducted used by Walters (2004) and Marin and Hayes
through secondary analysis of existing data. Among (2017).
the high-quality data sets used by researchers in 4. The Longitudinal Immigration Database (IMDB)
Canada are the following: developed by Citizenship and Immigration
Canada and Statistics Canada to provide data
1. The National Longitudinal Survey of Children
for the study of immigrants’ long-term earnings;
and Youth, which contains data on young
used by Li (2003) and Mata and Pendakur (2017).
people up to age 25; used by Brannigan et al.
5. The Canadian Election Studies, which pro-
(2002), Kerr (2004), Beran et al. (2008), and
vide data for all federal elections from 1965
Ames et al. (2015).
to the present; they are available through the
2. The National Population Health Survey, begun
Inter-University Consortium for Political and
in 1994–5 and planned to continue for 20 years;
Social Research; scores of publications have
it examines health and health care in Canada;
been produced using these surveys, for example,
used by Légaré et al. (2014).
Gidengil et al. (2006), Goodyear-Grant and Cros-
3. National Graduates Surveys, which ask
kill (2011), and Kevins and Soroka (2018).
graduates about their employment two years
after finishing their post-secondary education; These are only a few of the available data sets.
university-affiliated research centres (see Box 5.9). tiny fraction of the cost involved in collecting
The latter include the Institute for Social Research at new data.
York University, and the Institute of Urban Studies • High-quality data. Most of the data sets em-
at the University of Winnipeg. Sources such as these ployed for secondary analysis are of extremely
make raw statistical data available on topics such as high quality. First, the sampling procedures
income, fertility, crime, and unemployment, as well have often been rigorous, in many cases pro-
as a host of other social and political issues. Using ducing samples that are as close to represen-
secondary data rather than collecting your own has tative as is reasonably possible. A lthough
the additional advantage of sparing an already over- those responsible for these studies face the
surveyed public yet another round of questions. For same problems of non-response as anyone
this reason, secondary analysis should be considered else, well-established procedures are usually
not just by students but by all social researchers. in place for following up n on-respondents
Indeed, some granting agencies require applicants and thus keeping this problem to a minimum.
proposing to collect new data to demonstrate that Second, the samples are f requently national in
relevant data are not already available in an archive. scope, or at least cover a wide variety of regions,
which is a highly desirable—and costly—
Advantages of secondary analysis feature. Third, many data sets have been gener-
There are several reasons for considering secondary ated by highly experienced researchers and, in
analysis a serious alternative to collecting new data. Its the case of some of the large data sets, gathered
advantages have been enumerated by Dale et al. (1988): by social research organizations with strong
control procedures to check on data quality.
• Cost and time. As we noted above, secondary • Opportunity for longitudinal analysis. An-
analysis often offers good-quality data for a other valuable feature of secondary analysis is
124 PART II Quantitative Research
that it can provide an opportunity for longi- • Subgroup analysis. When samples are large (as
tudinal research. Sometimes, as with the in the GSS) there is an opportunity to study sub-
General Social Survey (GSS ; see Box 5.10), a groups. A sample of 100 people, for example,
panel format (in which the same subjects are might include only three people over age 85.
examined at different times) is used to chart No quantitative researcher wants to talk about
trends over time. Similarly, because certain those three seniors, as in “only 33 per cent of
interview questions are recycled and asked of the aged are in good health”: that amounts to
different samples each year, shifting opinions one person. With a national sample of 2000,
or changes in behaviour can be identified. the numbers are increased twenty-fold to 60.
Even though this number is still small, it will Cross-cultural research can address that issue.
allow a more meaningful estimate of seniors’ A study of religiosity by Kelley and De Graaf
health. It is impossible to reach such a number (1997) is an example. The authors described
in a small study unless the group in question is the process as follows:
the specific focus of the research.
• Opportunity for cross-cultural (international) Data are from the 1991 “Religion” module of
analysis. It’s easy to forget that many findings the International Social Survey Programme
may not apply to countries other than the (ISSP) . . . a module containing exactly the same
ones in which the research was conducted. questions, answer categories, and sequencing
Availability
GSS findings form the basis of a series of publica-
tions that present national and regional summary Source: Statistics Canada (2013), accessed 3 November 2018.
126 PART II Quantitative Research
for all countries surveyed. . . . The samples • Reanalysis can offer new interpretations. It’s
are all large, representative national samples easy to think that once a set of data has been
of adults. The most common procedure is to analyzed, the data have in some sense been
hold face-to-face interviews . . . followed by a used up. In fact, data can be analyzed in so
leave-behind self-completion questionnaire many different ways that there is almost always
containing the ISSP module . . . (1997, p. 642) additional insight to be gained from them. For
example, a secondary analyst may look at rela-
Kelley and De Graaf’s results were based on a tionships between variables that have not been
secondary analysis of data from 15 nations. Oppor- previously considered, prompting a reconsider-
tunities for such cross-cultural analysis appear to ation of the data’s relevance (see Box 5.11). Also,
be increasing. For example, common core questions new methods of quantitative data analysis, with
may be used in national surveys conducted in several the potential of allowing different interpreta-
countries. Both the US and Britain have the equiva- tions of the data, are continuously emerging. As
lent of our GSS. This allowed Grabb and Curtis (2004) awareness of such techniques spreads, and their
to create a body of literature comparing the US and potential relevance is recognized, researchers
Canada, revealing cross-national regional differences can apply them to existing data sets.
where national differences were widely expected. • It fulfills the wider obligations of the social
researcher. For all types of social research,
• More time for data analysis. Precisely because research participants give up some of their
data collection is time-consuming, the analysis time, usually for no reward. It’s not unreason-
of the data is often rushed. It might seem that able that the public should expect the data they
data collection would be the difficult phase and provide to be mined to the fullest. Making data
the analysis relatively straightforward, but this available for secondary analysis enhances the
is not always the case. Working out what to chances that they will be put to maximal use.
make of the data requires considerable thought
and, often, willingness to learn unfamiliar sta- Limitations of secondary analysis
tistical techniques. Freedom from the task of The foregoing list of benefits sounds almost too good
collecting fresh data means that data analysis to be true. But there are some drawbacks to analyz-
can be better planned and executed. ing data gathered by others:
Secondary analysis can involve topics that in all may be related to feelings of guilt, irregular eating
likelihood were not envisaged by those responsible patterns, and concerns about body shape. Similarly,
for the data collection. In a secondary analysis of Bell and Kandler (2017) did a secondary analysis
data on dietary choices, attitudes, and practices, of American twin data that showed that political
Beardsworth et al. (2002) added gender as a focus orientations, party identification, and interest in
that had not been examined in previous analyses. politics could not be fully explained by socio-
The study showed that women were more likely logical factors, and that those political characteris-
than men to adopt a “virtuous” pattern of eating, tics were still highly heritable after the sociological
one in keeping with Western ethical and nutritional variables were taken into account. Both of these
principles. The article also showed that adherence studies provide examples of profitable extensions
to these principles had a negative side in that it of prior research using existing data.
5 | Survey Research: I nterviews and Questionnaires 127
• Lack of familiarity with the data. With data col- Similarly, people could be coming from adjoin-
lected by others, a period of familiarization is ing neighbourhoods to commit crimes in a com-
necessary to come to grips with a wide array of munity they perceive as more vulnerable than
variables, the ways in which they were coded, their own. Group data cannot evaluate these
and various aspects of their organization. possibilities. To avoid the ecological fallacy, the
With large data sets this period can be quite unit of analysis (individual, group, neighbour-
prolonged. hood, etc.) of the data must be the same as the
• Complexity of the data. Some of the best-known unit of analysis of the statement or hypothesis.
data sets employed for secondary analysis, such For example, to test the hypothesis that people in
as the GSS, are very large both in the numbers group X are more likely to commit crimes than
of cases and in the numbers of variables they people not in group X, you would need data on
contain. Sometimes the sheer volume of data individuals, not neighbourhoods or regions.
can present problems, and here too a period of • No control over data quality. The point has
acclimatization may be required. Also, some of been made on several occasions that second-
the most prominent data sets employed for sec- ary analysis offers researchers the opportunity
ondary analysis are hierarchical, meaning that to examine data of far higher quality than they
the data are collected and presented at the level could collect themselves. However, this point
both of the household and of the individual applies mainly to reputable data sets such as the
(and sometimes at other levels as well). Different GSS and others. With lesser-known data sets,
data may apply to each level. Thus at the house- more caution may be necessary with regard
hold level, data on variables such as the number to data quality, although certain fundamental
of cars may be included, while at the individual checks on quality are usually undertaken by
level, data on income and occupation are found. the archives in which the data are deposited.
The secondary analyst must decide which level • Absence of key variables. Because secondary
of analysis to use; if the decision is to analyze analysis entails the analysis of data collected by
individual-level data, the individual-level data others for their own purposes, one or more of
must then be extracted from the data set. the secondary analysts’ key variables may not be
• The ecological fallacy. This can be a problem if present, or may be measured differently in dif-
data gathered by region or neighbourhood (such ferent years. Walters (2004) had this problem in
as census data) are used to make statements a study of incomes earned by graduates of vari-
about individuals. The term “ecological fallacy” ous kinds of post-secondary programs; therefore
comes from the common practice of acquiring he had to modify the original data. Similarly, an
data for a geographical area, such as the crime analyst might want to see if a known relation-
rate in a particular neighbourhood. Coleman ship between two variables holds even when
and Moynihan (1996) provided an example of other variables are taken into account, but might
this fallacy as it relates to the relationship be- find that those other variables are not part of the
tween ethnicity and crime. They observed that data set. Considering more than one independ-
findings showing a higher incidence of crime ent variable at a time is a form of multivariate
in neighbourhoods with high concentrations of analysis, a topic we will touch on in Chapter 8.
ethnic minorities had been used to imply that
members of such minority groups were more The need for survey data as a check
likely than others to commit crimes. However, and counterpoint to official statistics
individual data are needed to examine this hy- Official government statistics are important sources
pothesis, which may or may not be true. For ex- of data for secondary analysis. Some of these sources,
ample, it may not be the members of the minority such as the census, are based on interviews and ques-
groups who are responsible for the high levels of tionnaires conducted by state-run organizations.
offending, but their n on-minority neighbours. Others, such as the police data sometimes used to
128 PART II Quantitative Research
calculate the crime rate, are better understood as in- intoxicated, official estimates of the incidence of those
formation gathered by state agencies to record what behaviours will almost certainly increase. Moreover,
they have done in the process of conducting their during such crackdowns police officers may be less
duties. Here we explore potential deficiencies in the likely than usual to let perpetrators off with a warning.
latter and how survey data may sometimes provide a Zero-tolerance programs have the same result. This
needed correction to them. illustrates how variations over time in official crime
To be sure, official statistics offer the social re- statistics may not be the result of variations in the
searcher certain advantages: frequencies of the activities tracked, but may simply
reflect changes in the willingness to expend resources
• The data are often based on whole populations on surveillance or to proceed with prosecutions.
rather than samples, making it possible to Problems with official statistics may also stem
obtain a complete picture of the population in from factors related to “race,” ethnicity, and class.
question. For example, there may be variations in the likeli-
• There is a greater prospect of analyzing the hood that members of the public will report a crime
data both longitudinally and c ross-culturally. when the perpetrator is a member of a particular
Because the data are compiled over many ethnic group (Are Indigenous people more likely
years, it’s possible to chart trends over time than others to be noticed and reported?), and for a
and perhaps to relate them to broader social variety of reasons police may be less inclined to in-
changes. As well, official statistics from differ- vestigate white-collar crime than street crime.
ent nations can be compared for a specific area These deficiencies in official statistics may be offset
of activity. Durkheim’s (1897/1952) famous by using other sources of data when performing sec-
study of suicide, for example, was the result ondary analysis, most notably those derived from
of a comparative analysis of official statistics interviews and questionnaires that are based on rep-
from several countries. resentative samples. A good example of this is the
utilization of victimization surveys (such as the afore-
However, as mentioned in the previous chapter (see mentioned GSS victimization studies) in criminology,
Box 4.4), some forms of official statistics can be very which may be used in conjunction with official police
misleading if they record only those individuals who statistics. Both types of data have their strengths and
are processed by the agencies responsible for compiling weaknesses, so a judicious use of both may be required
the statistics, such as the police. Crime rates have been to understand the full extent and consequences of crim-
a particular focus of attention and concern among inal activity. Boivin and Leclerc (2016), for instance,
critics of the use of official statistics, but other official point out that survey research indicates that only a
statistics suffer from the same problem. For example, small fraction of domestic violence cases is reported to
official unemployment statistics may misrepresent the the police, and seek to explain why by analyzing inci-
“real” level of unemployment. People who have given dent reports compiled by Quebec police officers.
up trying to find work are often missed in the statistics,
while those who work in the “underground or informal
economy” (and thus are not actually unemployed) may
The feminist critique
be included in the unemployment statistics. In addi- Some feminist and other anti-oppressive social re-
tion, the definition of “unemployment” used by those searchers maintain that standard survey research
who compile official statistics may not be exactly the methods create an asymmetrical relationship in
same as the ones used by social researchers. which the researcher extracts information from the
Measurement problems may also arise if policies respondent and gives nothing in return. For example,
regarding the phenomenon to be counted vary over textbook advice of the kind offered in this chapter says
time. For example, if a local government decides to that it is important to establish a rapport with par-
devote more police resources to monitoring activ- ticipants, but that interviewers should guard against
ities such as drug use, the sex trade, or driving while becoming too familiar with them. Thus, participants
5 | Survey Research: I nterviews and Questionnaires 129
who ask questions (for example, about the topic of the special training in the handling of sensitive topics
study or the research process) should be politely but has contributed to this change. Moreover, survey re-
firmly rebuffed lest the answers bias their subsequent search in several countries has proven to be highly
responses. Feminist scholars in the past have criticized instructive about issues of concern to feminists, such
such practices as hierarchical and even exploitative. as the frequency and causes of phenomena such as
On the other hand, as we noted in Chapter 1, there violence against women and children (Brownridge
has been some softening of attitudes toward quan- et al., 2017; Walby & Myhill, 2001), and the challenges
titative research among feminist researchers. This faced by sexual assault and rape crisis centres (Beres
has come about in part because the way research et al., 2009; see the “Methods in Motion” box in this
is conducted has evolved in response to criticisms chapter). Such research would seem to be consistent
from feminists and others. Paying more attention to with the goals of most feminist researchers, and is
issues such as privacy in the interview and providing potentially of great value for both women and men.
Key Points
Survey research is one of the most widely used • While open questions undoubtedly have advan-
data-gathering techniques in the social sciences. tages in survey research, closed ones are typically
Surveys include interviews, in which a researcher en- preferable. They facilitate the asking of the ques-
gages in a verbal discussion with the participant, and tions, the recording of answers, and coding. This
point applies especially to questionnaires.
questionnaires, which respondents read and fill out
on their own. This kind of research permits standard- • Open questions of the kind used in qualitative
ization in the asking of questions and the categoriz- interviewing can play a useful role in formulating
ation of answers. fixed-choice answers.
130 PART II Quantitative Research
• Learning the rules of question asking will help you • Online surveys take two major forms: Web surveys
avoid some serious problems. and email surveys.
• Question order is very important and some gen- • The researcher-driven diary is an alternative to
eral rules should be followed. using questionnaires and interviews, especially
• Response sets are a potential problem in both when the research questions are concerned
structured interviews and questionnaires, but with specific behaviours rather than attitudes or
steps can be taken to lessen their impact. opinions.
• Always put yourself in the position of the re- • The visual presentation of closed questions and
spondent when devising questions. their general layout are important considerations
in designing questionnaires.
• Ensure that the survey questions generate data ap-
propriate to the research questions. • Secondary analyses allow researchers to conduct
their inquiries without having to collect new data.
• A pilot study can identify problems in question
formulation. • Very often, secondary analysis involves high-
quality data sets taken from large, representative
• Structured interviews can be administered in
samples.
person, over the phone, or online.
• Secondary analysis presents a few disadvantages,
• It is important for interviewers to keep to the
such as the absence of theoretically important
wording and order of questions. Training in the
variables.
asking of questions, the recording of answers,
and how to establish rapport with respondents is • Official statistics, especially those relating to
essential. crime, can be problematic for secondary analysis.
• Questionnaires reduce some of the problems en- • The problems associated with official statistics
countered in structured interviews, but they also may be offset by using survey data taken from rep-
have weaknesses. resentative samples.
AA Assume you are doing a study on the frequency of What disadvantages do they have compared to
tobacco smoking among university students, and traditional research methods?
that there are already widely used questions on AA If you were asked to participate in an online survey
this topic that have been administered to the gen- concerning cheating on university exams, would
eral public. Would it be necessary to devise your you agree to do it? Explain why or why not, then
own questions, or could you rely on the existing explain how researchers could increase the re-
ones? Explain. sponse rate for the study.
each person is interviewed, he or she comments revised versions solve all the problems contained
on the experience—e.g., on whether the questions in the original question, or if further revision
were clear, whether they felt embarrassed, whether would be appropriate.
the fixed responses adequately covered what came 4. The class is divided up into small groups. Each
to their mind when they were asked the questions, group is asked to come up with 10 survey ques-
and so on. The exercise is then repeated, with people tions designed to measure attitudes toward a
in Group B interviewing people in Group A. Once all controversial topic (e.g., assisted suicide, gay mar-
class groups are finished, the instructor leads an riage, animal rights, etc.) and to gather basic socio-
open discussion that starts with the question “What demographic data. Once that has been done, each
did you learn from this exercise?” group puts its questions up on the board/screen,
3. The instructor reads each question, one at a time, and the rest of the class provides a critique. Once
from Box 5.5. After each question is read, the class all groups have had their turn, the groups get
is asked to point out its flaws. All members of the together again and revise their questions to ad-
class are then given two minutes to come up with dress the concerns raised by their classmates. Each
a revised version of the question. Individual stu- group presents its questions to the class a second
dents are called on to produce their revised ver- time, with commentary on how the previous weak-
sions, which are written on the board/screen. The nesses have been corrected. The class again offers
class is then asked to comment on whether the a critique to each group.
Relevant Websites
The Canadian Election Study site gives you access justice system and experiences of criminal victimiz-
to structured interviews and questionnaires and ation in Canada.
the numerical data derived from them for Canadian www23.statcan.gc.ca/imdb/p2SV.
elections going back as far as 1965. pl?Function=getSurvey&SDDS=4504
http://ces-eec.arts.ubc.ca/
The Inter-University Consortium for Political and
This YouTube video explains how to make an online Social Research is an international association of
survey using Google Docs. about 700 academic institutions and research organ-
www.youtube.com/watch?v=DEhD0m1fhAQ izations that “provides leadership and training in data
access, curation, and methods of analysis for the social
In the General Social Survey on Time Use, 2015
science research community.” Click on “Find Data” to
Questionnaire, you can find an example of a research
access over 500,000 digital files for examples of ques-
diary that forms part of a larger structured interview.
tions used in a wide variety of social research projects.
www23.statcan.gc.ca/imdb/p3Instr.
www.icpsr.umich.edu/icpsrweb/ICPSR
pl?Function=getInstrumentList&Item_
Id=217656&UL=1V& The UK Data Service provides a searchable databank
of questions used on social surveys in the UK.
The General Social Survey on Victimization, 2014
http://ukdataservice.ac.uk
Questionnaire contains the questions used by Sta-
tistics Canada to assess perceptions of our criminal (Websites accessed 20 October 2018)
Chapter Overview
Structured observation is a relatively underused method in social research. It entails the direct obser-
vation of behaviour, which is analyzed using categories that are devised before the observation begins.
This chapter explores:
• the limitations of survey research for the study of behaviour;
• different forms of observation in social research;
• the potential of structured observation to contribute to our understanding of social behaviour;
• how to devise an observation schedule;
• different strategies for conducting structured observation;
• issues of reliability and validity in structured observation;
• field studies, in which a researcher intervenes in actual social life and records what happens as a
consequence of the intervention;
• ethical issues in the above; and
• some criticisms of structured observation.
Kate has just arrived at a “meet and greet” reception She notices that the people who look middle-aged
held by her university’s social sciences division as part tend to be talking to other people who look
of Orientation Week. The invitation said that professors middle-aged. Are they profs? A group has gathered
and even the dean would be in attendance. Laid out on around a middle-aged woman in a business suit who
a long table in the centre of the room are fruit and vege- is holding forth about student grades in the social
table snacks, cheese and crackers, wine, beer, and fruit sciences compared to the humanities. Is she the dean?
juices, as well as tea and coffee. Being naturally curious, Kate also notices that in the past 10 minutes, the woman
Kate decides that instead of schmoozing she will take who might be the dean has talked to seven people who
a glass of wine and make her way to a deserted part of appear to be profs, but only two who look like students.
the room. The people who look like students seem far more likely
▲ rawpixel/123rf
134 PART II Quantitative Research
to be talking to other students than to profs. Shifting her to one. As a test of the latter observation she sets the
attention to the table in the centre of the room, Kate timer on her watch so she can do a formal count for
notices that the middle-agers who are drinking alco- 10 minutes. To her horror, the woman who appears to be
hol have almost all chosen wine rather than beer, and the dean walks all the way across the room and asks her
that the women prefer white to red by a factor of two brusquely, “Are you doing some kind of study?”
• Problems of meaning. People may vary in their For example, men who recently underwent sur-
interpretations of key terms in a question. Does gery for prostate cancer may not want to admit
“watching television” include having it on in the to any incontinence.
background while you are making dinner? • Gap between stated and actual behaviour.
• Problems of memory. Respondents may not re- How people say they would behave and how
member certain aspects of their behaviour, or they actually behave is often inconsistent (see
they may have false memories of it. For e
xample, Box 4.6). Husbands, for example, traditionally
people may underestimate the number of alco- overestimate how much time they devote to
holic drinks they consume in a typical week. housework. Perrucci et al. (2000) found two
• Social desirability bias. Perhaps the most im- quite different kinds of racial relations on a
portant shortcoming of survey research is predominantly white university campus, be-
that respondents will often give answers that haviours they described as “front stage” and
they think will reflect well on them; for ex- “back stage.” The “back stage” behaviour,
ample, overestimating donations to charity or which showed less than full acceptance of
under-reporting traffic tickets. racial minorities, did not become apparent
• Threatening or embarrassing questions. These until questionnaires were supplemented by
may lead to replies that are not fully truthful. direct observation.
6 | Structured Observation 135
So why not just observe possible to examine the frequency with which each
group asks questions and answers them, talks or
behaviour directly? keeps silent, and so forth. Similarly, age and gender
An obvious solution to the problems identified in differences could be explored in relation to Bales’s
Box 6.1 is not to rely on research instruments such as categories 11 (“Shows tension”) and 2 (“Tension re-
surveys, but to observe people’s behaviour directly. lease”). Or the same categories could be used to com-
Structured observation (sometimes called “system- pare “authoritarian” and “shared” leadership styles.
atic observation”) is one way to do that. (See Box 6.2 For example, is there less tension and thus less need
for other types of observation research.) Using this for tension relief with an “authoritarian” leadership
method, the researcher formulates explicit rules out- style, or with a “shared” one? A scheme like this
lining what behaviours are to be observed and how could also be used to help leadership trainees im-
the observations are to be recorded. Each person in prove their leadership skills.
the study is observed for a predetermined period of If an observer coded what was happening in
time following rules set out in a document (similar the group as a whole every 15 seconds, the coding
to a structured interview schedule) that is usually for a 12-minute period might look something like
called an observation schedule. The data collected in Figure 6.2. Notice how often a question tends to
this way, like survey data, can then be treated as vari- be followed by an answer and a negative affect by a
ables for analysis. smoothing positive affect (Bales organized his cat-
One of the classic schedules for the observation egories so that the codes assigned to linked behav-
of small-group behaviour was developed by Bales iours such as these add up to 13). In many cases,
(1951) (see Figure 6.1). Suppose you wanted to com- two types of leaders emerge: a task specialist and a
pare leaders and followers in terms of their behav- socio-emotional leader who is more responsible for
iour: using Bales’s observation schedule, it would be group members’ feelings.
• Participant observation. This is one of the setting. Bell’s (2007) study of western Canadian
best-known methods of social science re- separatists is an example of this sort of obser-
search, especially in sociology and anthro- vation. If the subjects are not aware that they
pology. It is primarily qualitative and entails are being observed and analyzed, such re-
the relatively prolonged immersion of the search can be called unobtrusive observation.
observer in a particular social setting (for ex- However, Bell’s subjects knew that he was
ample, a group, organization, or community). studying them. Non-participant observation
The main goal is usually to elicit the meanings may be either structured or unstructured (see
that the people being observed attribute to below); Bell’s approach was unstructured.
their environment and behaviour. Participant • Unstructured observation. As its name implies,
observers vary considerably in how much they unstructured observation does not follow
participate in the social settings in which formal rules for either making or recording ob-
they locate themselves. (For a more detailed servations. Instead, the observer records the
discussion, see Chapter 10.) subjects’ behaviour in as much detail as pos-
• Non-participant observation. This term ap- sible, and then develops a narrative account
plies to situations in which the observer does (story) of that behaviour. Most participant ob-
not take part in what is going on in the social servation is unstructured.
136 PART II Quantitative Research
7 7 6 6 6 7 6 6 8 5 10 2
5 8 5 12 1 9 9 4 4 5 4 4
10 11 12 12 2 1 1 11 2 8 8 5
5 10 10 2 1 3 2 1 3 6 2 1
of the observer. For example, in Bales’s scheme ORACLE (Observational Research and Class-
it can be difficult to distinguish between a cat- room Learning Evaluation) project (Galton
egory 10, “Withholds help,” and a category 11, et al. 1980). In this research, eight children
“Withdraws.” If interpretation is needed, clear (four of each gender) and their teacher were
guidelines and considerable training and ex- observed at regular intervals. A mechanical
perience are required. device made a noise every 25 seconds, signal-
ling the observer to record what the teachers
or pupils were doing (in accordance with the
Strategies for observing observation schedule).
behaviour
There are different ways to observe and record
behaviours:
Issues of reliability and validity
Compared with interviews and questionnaires,
• Recording incidents means waiting for some- structured observation “provides (a) more reliable
thing to happen and then recording what fol- information about events; (b) greater precision re-
lows from it. Essentially, this is what LaPiere garding their timing, duration, and frequency; (c)
(1934; see Box 4.6) did when he waited for a greater accuracy in the time ordering of variables;
Chinese couple to negotiate entry to a hotel or and (d) more accurate and economical reconstruc-
restaurant and then noted whether they were tion of large-scale social episodes” (McCall, 1984,
allowed in or not. p. 277). This is a very strong endorsement, but there
• A wide variety of behaviours can be observed are issues of reliability and validity to consider. Some
and recorded in either short or long periods of them are similar to those faced by any researcher
of time. In research reported in Chapter 2, seeking to develop measures in social research (es-
children in St Helena were videotaped over pecially survey research), while others are specific to
a two-week period during their morning, structured observation.
lunch, and afternoon breaks. The tapes were
then coded using “the Playground Behav- Reliability
iour Observation Schedule, an instrument One thing practitioners of structured observation
for recording the occurrence of 23 behaviours have been concerned with is inter-observer con-
(e.g., games; fantasy play; character imitation; sistency. Essentially, this entails considering how
anti-social and pro-social behaviour) and their closely two or more observers of the same behaviour
behaviour groupings (i.e., whether the behav- agree on how to code it (see Boxes 6.3 and 6.4).
iour was undertaken by an individual, a pair, A second consideration is intra-observer con-
by 3 to 5 children, or 6 or more). . . . A separ- sistency: the degree of consistency in the appli-
ate schedule was completed for each 30-second cation of the observation schedule by a single
segment” (Charlton et al., 1998, p. 7). observer over time. Consistency is difficult to
• Time sampling is another approach to the achieve because people tend to behave differently
observation of behaviour. Since it is usually depending on the context. Reliability may also
not possible to observe a situation continu- be threatened by factors such as observer fatigue
ously for extended periods of time, research- and lapses in attention. However, observers can be
ers generally sample the times at which they trained to get highly reliable results using complex
make their observations. The times at which coding schedules. The procedures for assessing
the observations are made are chosen in ad- intra-
observer reliability across all possibilities
vance, either systematically or randomly. An are broadly similar to those used to assess inter-
example is a study of schools known as the observer consistency.
138 PART II Quantitative Research
Normand and colleagues (2013) studied how The results indicated that children with ADHD
attention deficit hyperactivity disorder (ADHD) committed a higher proportion of illegal man-
may affect friendships among children aged 7–13. oeuvres in the car race, and that, unlike those
Children from the Ottawa–Gatineau region were without ADHD, increased the proportion of rule
recruited to participate, about two-thirds of whom violations over time. Similarly, in the card-sharing
had a diagnosis of ADHD. The recruited children task, ADHD children made a greater proportion
were asked to invite their best friend to participate of self-centred and insensitive proposals, and
as well. The research was conducted at two differ- increased their level of insensitive proposals.
ent time periods, six months apart. One aspect of When choosing games, children with ADHD made
the study involved structured observation of the a higher proportion of insensitive proposals. All
friendship pairs engaging in three dyadic activ- in all, it appeared that the ADHD children were
ities: a rule-based car race in which each child at a marked disadvantage with regard to the be-
controlled a toy vehicle; a sharing activity involv- haviours associated with creating and sustaining
ing trading cards; and a game-choice task in which healthy friendships.
the friendship pair had to decide on which of sev- Given the importance of high-quality friend-
eral games offered they would play at the end of ships for personal well-being, the researchers
the session. During these interactions, observers recommended a number of interventions to help
counted various types of behaviour that correlate children with ADHD, including having parents
with childhood friendship quality and satisfaction, coach their children on the importance of rule
such as the number of rule violations committed in following, sensitivity to others, and comprom-
the first task, and the number of self-centred or in- ise when interacting with friends or potential
sensitive negotiation tactics used in the other two. friends.
Validity
BOX 6.4 Cohen’s kappa
Faulty administration of any measure can obviously
Cohen’s kappa is a measure of the level of affect its validity. However, ensuring measurement
agreement between two people’s coding de- validity—that is, ensuring that an indicator is meas-
cisions that takes into consideration the possi- uring what it is supposed to measure—can be diffi-
bility that agreement will occur by chance. cult even when the measure is administered properly.
It can be applied to any coding. Much like The second point simply means that structured
Cronbach’s alpha (see Chapter 4), the coeffi- observation demands attention to the same kinds
cient varies from 0 to 1. The closer it is to 1, of issues (assessing face validity, concurrent valid-
the higher the agreement and the better the ity, and so on) that interviews and questionnaires
inter-observer consistency. As a rough rule do. The first—error in implementation—relates to
of thumb, a coefficient over .75 is considered two matters in particular:
“very good”; from .6 to .75, “good”; “fair” ranges
from .4 to .59. • Is the observation schedule being administered
as directed? This is the equivalent of ensuring
6 | Structured Observation 139
that structured interviewers follow instruc- • Do people change their behaviour when they
tions exactly. Variability between observers know they are being observed? If they do, that
or over time makes the measure unreliable would constitute one of several kinds of “reactive
and therefore invalid. Observers should have effect” that can occur in various forms of research
a thorough understanding of how the observa- (see Box 6.5). If people adjust the way they behave
tion schedule is to be implemented. because they know they are being observed
Webb et al. wrote about “reactive measurement quasi-experiments showed that each time work-
effect” (1966, p. 13). It occurs when subjects’ ing conditions improved (e.g., through improved
awareness that they are participating in research lighting), productivity went up, at least tempor-
leads to changes in their behaviour that confound arily. But it also went up after working conditions
the investigator’s data. Response sets, social desir- deteriorated! It appeared that performance im-
ability bias, and political correctness can all cause proved simply because the workers were aware of
reactive effects. If a respondent reacts to some being observed and analyzed, not because of the
characteristic of the interviewer (e.g., gender or experimental manipulation itself: this is known
ethnicity), that is another example of a reactive as the “Hawthorne effect.” A well-known general
effect. Here are a few more: problem in experimental studies is that some in-
dividuals will seek out cues about the aims of the
• The Hawthorne effect. The classic studies illus- research, and then adjust what they say and do to
trating this effect were done at the Hawthorne conform with what they believe to be the purpose
Works (located near Chicago) in the 1920s (see of the study. That is a problem, even if such partici-
Roethlisberger & Dickson, 1939). A series of pants are mistaken about the aims of the research.
continued
140 PART II Quantitative Research
• Role selection. Webb et al. argued that par- knows certain crucial details, such as which
ticipants may be tempted to adopt a particu- subjects have received the treatment and
lar kind of role in research. Some people with which have not (single blind means that only
no strong opinion on a particular item may the subjects are kept in the dark). Here, hiring
develop one when asked, in order to appear others to collect the data can help.
knowledgeable. Even telling them how import-
ant they are to the research can encourage Reactive effects are likely to occur in any re-
some display of “expertise.” Other, fairly opin- search in which participants know they are the
ionated individuals may back off a bit, scared focus of investigation. Webb et al. called for greater
by the audio recorder, perhaps, or not wanting use of what they called unobtrusive measures:
to come on too strong to a stranger. non-reactive methods that leave participants un-
• Researcher presence as a change agent. The aware that they are being studied. Records can be
simple fact that the researcher is present may checked at hospitals and courts, for instance, re-
change the situation that is being observed. For moving the reactivity, but for ethical reasons per-
example, if an observer is sitting in a classroom, mission must first be obtained, and getting such
that means a chair is being occupied that other- permission may be difficult or impossible.
wise might have been empty, which in turn may Alternatively, the usual tools of social research
cause changes in space and privacy for those can be used. A confidentiality guarantee backed up
sitting nearby. When Whyte (1955), in his study by reminders may help, as may anonymous ques-
of “Cornerville,” joined one of the neighbour- tionnaires and telephone (rather than in-person)
hood’s bowling teams, he in effect removed one interviews, although these have their own down-
of his subjects from the team and turned that sides. It also helps to phrase questions so that they
person into a rival on an opposing team. are as non-threatening and non-judgmental as pos-
• Trying to help or to be nice to the researcher. sible; for example: “Many people did not have time
Although it is often assumed that research par- to vote in the referendum or could not make up their
ticipants are always passive, this is not true, even minds. Did you vote?” If you want to allow more men
in experiments. Some research subjects may try to admit they fear being out late at night, don’t ask
to please the researcher. An old saying holds how afraid they are. Use a vignette: “John does not
that if you ask friends to do push-ups, they will like to walk alone at night in his neighbourhood.
ask “what?” or “why?” But subjects in an experi- There is a lot of crime there, and police officers are
ment (especially students, the usual pool from rarely if ever seen. Is John almost a copy of you, a lot
which subjects are drawn) will ask “where?”! In like you, a bit like you, not at all like you?”
fact, subjects generally try to do what they think Finally, although it is good to be aware of the
they are supposed to do, although a few may shortcomings resulting from reactive effects,
deliberately do the opposite. Some will even remember that no study is perfect, and that an
ask “Was I good?” at the end of the experiment. imperfect study can still be an important source of
Such behaviour can be problematic if it affects knowledge. Also, maintain the same level of skep-
the phenomenon being studied. ticism when the results you encounter please you
• Different researchers may elicit different as when they displease you. For example, research
reactions. Some may get the data they want showing that university men are more violent than
by communicating, often unconsciously, their university women in heterosexual romantic rela-
expectations. This is why experiments are sup- tionships should be held to the same methodo-
posed to be double blind, so that no one (not logical standards as research showing no gender
even the person administering the experiment) differences (for example, Katz et al., 2002).
6 | Structured Observation 141
(perhaps because they want to be seen in a fa- Another famous field experiment was conducted by
vourable light by the observer), their behaviour Rosenhan (1973). He was one of eight researchers who
would have to be considered atypical or not au- posed as patients and were admitted to mental hos-
thentic. However, as McCall (1984) noted, there pitals in the US. Each pseudo-patient was instructed
is evidence that participants become accustomed to claim that he or she was “hearing voices,” and each
to being observed and that researchers tend to was accordingly diagnosed with schizophrenia. As
become less intrusive the longer they are in the soon as they were admitted, the pseudo-patients were
field. However, these effects are likely to persist instructed to cease exhibiting any symptoms. Even
when the behaviour under observation involves though they were all “sane,” it took many of them
sensitive issues such as sexuality or deviance. quite a long time to be released. The length of hospital-
ization varied between 7 and 52 days, with an average
of 19 days. In four of the hospitals, pseudo-patients
Field experiments as a form of approached psychiatrists and nurses with a request
structured observation for release and recorded the nature of the response:
A field experiment is a study in which a researcher 71 per cent of psychiatrists moved on with their heads
directly intervenes in a natural setting to observe the averted and 88 per cent of nurses did likewise. To
consequences of that intervention. That’s what LaPiere Rosenhan this indicated that a mental patient becomes
(1934) did when he arranged for a Chinese couple to powerless and depersonalized. This study is highly
seek entry to hotels and restaurants in order to ob- controversial, in part because of its use of deception,
serve the effects (although his study was an imperfect the ethical implications of which were discussed in
experiment because there was no control group of Chapter 3. Also, in this sort of research, opportunities
non-Chinese people). In a field experiment—unlike to employ an observation schedule are limited because
most types of structured observation—participants excessive use will blow the observer’s cover. The most
do not know that they are being studied. that can usually be done is limited coding.
Checklist
Checklist for structured observation research
☐☐ Are the research questions stated clearly? ☐☐ Are the categories of behaviour exhaustive?
☐☐ Does the observation schedule indicate pre- ☐☐ Do the different categories of behaviour allow
cisely the kinds of behaviour to be observed? you to answer the research questions?
☐☐ Have observation categories been designed ☐☐ Has a pilot study using the observation sched-
to minimize the need for observers to inter- ule been conducted?
pret what is going on? ☐☐ Are the coding instructions clear?
☐☐ Has any overlap in the categories of behav- ☐☐ Is it easy to log the behaviour while it is
iour been eliminated? happening?
Key Points
• Structured observation is an alternative to survey- • Reactive effects must be taken into account, but
based measures of behaviour. should not be exaggerated.
• It involves explicit rules for recording behaviour. • The field experiment is a form of structured obser-
• Structured observation has generally been ap- vation, but can involve ethical difficulties.
plied to a narrow range of behaviours, such as that • Problems with structured observation include de-
occurring in schools, courts, and hospitals. termining the meaning that people give to their
• It shares with survey research many common behaviour and ensuring that the framework for
problems concerning reliability, validity, and gen recording the observations is valid.
eralizability.
and cons of using structured versus unstructured research that is used to examine how parents treat
observation to conduct this research. their children while they shop in a large depart-
ment store?
The observation schedule
RR What goals and considerations should be kept in Field experiments as a form of structured
mind when constructing an observation schedule? observation
AA Construct an observation schedule containing five RR What are field experiments and what ethical con-
items that could be used to observe fan behaviour cerns do they pose?
at a hockey game, and a second schedule to be AA What ethical issues should you consider if you do
used for observing people attending a figure skat- a field experiment that involves posing as a con-
ing competition. Then explain how and why they fused traveller who does not speak English or
are different. French at a busy Canadian airport?
methodologies. Group 3 makes the case that there groups 2 and 3. Once all groups have presented,
is no methodology that can be ethically applied each group is given some time to assess and cri-
to research Rosenhan’s topic, and that therefore tique the other groups. Each group then presents
no one should be allowed to conduct the study. its critique to the rest of the class, with a general
After some time for preparation, group 1 pre- discussion to follow.
sents its case to the rest of the class, followed by
Relevant Websites
At Bonkers Institute, Rosenhan describes the famous This is a short YouTube video of structured observa-
field experiment he conducted in mental hospitals. tion being done on children.
www.bonkersinstitute.org/rosenhan.html www.youtube.com/watch?v=VgTpAoJEXaE
Chapter Overview
Sampling principles are not exclusively the concern of survey and other quantitative researchers. The
qualitative work you will read about in Part III also involves working with samples, but the nature of
qualitative sampling is often rather different from the quantitative kind, so a discussion of the former has
been placed in that part of the book. This chapter explores:
• the role of sampling in relation to the overall process of doing quantitative research;
• why generalizable findings (that is, findings that can be applied to some larger population) require
a representative sample;
• the idea of a probability sample: one using a random selection process;
• the main types of probability sample: simple random, systematic, and stratified random, and the
multi-stage cluster sample;
• the main issues involved in deciding sample size;
• different types of non-probability sampling, including quota sampling, which is widely used in
market research;
• potential sources of sampling error in survey research;
• sampling in structured observation research, where times and contexts are sampled; and
• raising response rates.
Do you like to spend more time and money than is ac- costs involved make that impossible. In other areas of
tually necessary to accomplish something? Most people life, reducing the time and money you devote to a pro-
would say no, including social researchers, who inevit- ject normally means a reduction in the quality of the
ably face the prospect of doing their work with limited final product—think of building a house on a limited
time and physical resources. Ideally, researchers would budget. Fortunately, this is not necessarily the case with
study huge populations, such as all women in Canada social research. With proper sampling techniques you
or all adults living in the Atlantic provinces, but the can vastly reduce the time and resources needed to
▲ Nickbeer/iStockphoto
7 | Quantitative Sampling 147
complete the research without seriously compromising even calculate the probability that sample-based statis-
the findings. You might think of sampling as “how to do a tical findings are in error. This chapter introduces you to
lot with very little.” Under the right conditions, you can the techniques that make this possible.
Introduction these samples would not do. Why? There are various
reasons, but the following stand out:
The term “sampling” refers to the selection of a
subset of a population for research. Since only in • If you distribute questionnaires in your classes,
very rare instances can an entire population be anyone not taking a course with you will be
studied, sampling has an impact on virtually all excluded. How many music or engineering
the research discussed in this book. As we will see majors take sociology?
below, how the sampling is conducted determines • If you go out looking for people to interview,
what information is acquired and influences the your sample will exclude anyone who happens
researcher’s interpretation of the results in import- to be somewhere else at that particular time.
ant ways. Some students will be at work, others will have
The key terms used in sampling are explained skipped classes that day, and many will have a
in Box 7.1. We’ll begin with a discussion of sam- timetable different from yours. Some students
pling in studies involving structured interviews may never need to go where you are.
or questionnaires, and then consider the tech- • Your decisions about which people to ap-
niques used for other methods of quantitative re- proach may be influenced by your judgments
search. Finally, ways to increase response rates are about how friendly or cooperative they appear
presented. to be. Gender, “race,” and age are only a few of
Suppose you are wondering about your fellow the factors that could have an impact on your
students’ attitudes on certain matters, or perhaps selection of prospective respondents.
about their behaviours or backgrounds. To exam-
ine any of these areas, either structured interviews The problem with these sampling methods is that
or questionnaires would be appropriate. However, they make the choice of possible participants de-
let’s say that there are around 30,000 students at pendent on non-universal criteria: availability in one
your university. It would be a huge undertaking for case and personal judgment in the other. In the lan-
you to arrange for all of them to fill out a question- guage of sampling, this means that the sample would
naire, and the time and energy required to interview be biased: it would not represent the population
30,000 people would be prohibitive. It’s almost cer- from which it was selected. It is extremely difficult
tain that a sample of students would have to suffice. to remove bias altogether and get a truly represent-
In fact, the need to sample is almost universal in ative sample, but steps should be taken to keep it to
quantitative research. a minimum.
But would any sample do? What if you gave ques- Three sources of bias can be identified:
tionnaires to everyone in your classes? Or if you
stood in a busy location on campus and then inter- • Not using a random method to pick the sample.
viewed anyone who agreed to speak with you? Essentially, a random method is one in which each
If your goal is simply to get some experience in element has the same chance of being selected,
data gathering and analysis, or to conduct a pilot like numbers from a lottery cage (although in
study, these samples would be fine. However, if the some circumstances that principle may be vio-
object is to get a representative sample, one that can lated for technical reasons; for example, in order
be used to make inferences about all 30,000 students, to get a sufficiently large number of people from
148 PART II Quantitative Research
• Element or unit: A single case in the popu- • Probability sample: A sample selected using
lation. In the social sciences that element a random process such that each unit in the
or unit is usually a person, but many other population has a known chance of being
things can be sampled as well: nations, cities, selected. The aim of probability sampling is to
regions, schools, firms, and so on. Finch and keep sampling error (see below) to a minimum.
Hayes (1994), for example, based part of their • Non-probability sample: A sample selected
research on a random sample of wills of de- using a non-random method. Essentially, this
ceased people. implies that some units in the population are
• Population: All the cases about which you are more likely to be selected than others.
seeking knowledge, or all the cases to which • Sampling error: An error of estimation that
your conclusions are meant to apply. For ex- occurs if there is a difference between the char-
ample, if you are studying voting behaviour, acteristics of a sample and those of the popula-
the population might be all the people in a tion from which it was selected. Sampling error
particular jurisdiction who are eligible to vote. can occur even when a random method is used.
If the topic is hyperactivity among young chil- For example, if valid measures indicate that
dren, the population might be all children aged 45 per cent of a national probability sample is
2 to 11. (Note that “population” in this context in favour of more public funding for the arts, but
means something different from “population” the actual figure in the population is 49 per cent,
in the everyday sense.) the difference is caused by sampling error.
• Sampling frame: The list of elements from • Non-response: A situation that occurs when-
which the sample will be selected. ever some unit selected for the sample refuses
• Sample: The elements selected for investiga- to participate in the study, cannot be contacted,
tion, a subset of the population. The method or for some other reason does not supply the
of selection may involve probability or non- required data.
probability sampling (see next column). • Census: Data collected from all elements in
• Representative sample: A sample that is a the population rather than from a sample.
microcosm of the population, one that “repre- The phrase “the census” typically refers to the
sents” its essential characteristics. A sample enumeration of all (or nearly all) members of
such as this is most likely to be selected when a the population of a nation-state—that is, a na-
probability sampling process is used (see next). tional census.
a small subgroup into the sample). As we saw cases or is otherwise inaccurate, the sample de-
above, if a random method is not used, there is rived from it may not represent the population,
a risk that the selection process will be affected even if a random sampling method is employed.
by human judgment, so that some members of • Some people in the sample refuse to participate
the population are more likely to be selected than or cannot be contacted—in other words, there is
others. This source of bias is eliminated through non-response. The problem with non-response
the use of probability sampling, described later is that those who agree to participate may differ
in this chapter. from those who refuse or cannot be reached in
• The sampling frame or list of potential subjects is ways relevant to the subject matter being inves-
inadequate. If the sampling frame excludes some tigated. If that is the case, the sample may not be
7 | Quantitative Sampling 149
representative. For example, if you are studying If the sample is representative, the sample of 50
rates of criminal victimization and people who should also be equally split in terms of this variable
have been victims of violent crime decline to (see Figure 7.1). If there is a small amount of sampling
participate, that would skew the results. error, so that the sample contains one person too
many who does not watch reality TV and one too few
If the data are available, a researcher can check to who does, the distribution would look like Figure 7.2.
see how non-respondents differ from the population. Figure 7.3 shows a more serious over-representation
It is often possible to do this in terms of characteristics in the sample of people who do not watch reality pro-
such as gender or age, but these are not the only pot- grams. This time there are three too many who do
entially relevant factors; in the case of a sample of uni- not watch them and three too few who do. Finally,
versity students, for example, you might want to find Figure 7.4 shows a very serious over-representation of
out whether non-respondents reflect the entire stu- people who do not watch reality TV: 35, as opposed to
dent body in terms of faculty of r egistration. H
owever, 25 if the sample were perfectly representative.
it is usually impossible to check c haracteristics that It’s important to appreciate that the possibility of
go beyond basic demographic categories, such as sampling error can never be completely eliminated.
attitudes toward separatism or patterns of smoking Even with a well-crafted probability sample, a degree
behaviour, because population data are generally not of sampling error is likely to creep in, just as flipping
available on these topics. a coin 100 times is likely to yield more 49Heads/
51Tails and 51Heads/49Tails results combined than
50Heads/50Tails outcomes, even though the latter is
Sampling error the most probable single result. Ask your instructor
To appreciate the significance of sampling error to take a minute for an in-class experiment in which
for achieving a representative sample, consider a coin is flipped 20 times and you’ll see the surpris-
Figures 7.1 through 7.4. Imagine a population of 200 ingly large number of outcomes that are not 10Heads/
people and a sample of 50. Imagine as well that the 10Tails. However, probability sampling stands a
topic is whether people watch reality TV programs, better chance than non-probability sampling of
and that the population is divided equally between minimizing sampling error, so it’s very unlikely that
those who do (100) and those who do not (100). the sample will end up looking like Figure 7.4 when
Watch reality TV Do not watch reality TV Watch reality TV Do not watch reality TV
FIGURE 7.1 A sample with no sampling error FIGURE 7.2 A sample with very little sampling error
150 PART II Quantitative Research
Watch reality TV Do not watch reality TV Watch reality TV Do not watch reality TV
FIGURE 7.3 A sample with a fair amount of FIGURE 7.4 A sample with a lot of sampling error
sampling error
it is used. Moreover, probability sampling allows you This is known as the sampling ratio and is expressed as
to employ tests of statistical significance that will
permit you to make inferences about the population n/N
from which the sample was selected, with a known
probability of error. This topic will be addressed in where n is the sample size and N is the population size.
Chapter 8. The key steps in devising a simple random sample
are as follows:
are worth noting here. First, there is almost no on to the next one. This means that the student as-
opportunity for bias, since in this case the selec- signed the number 45 will be the first to be included
tion process is entirely random: students are not in the sample. Moving down the first column,
selected on the basis of subjective criteria such as the student assigned the number 3189 will be next;
whether or not they look friendly and approach- the student assigned 5768 next; and so on.
able. Second, selection does not depend on To avoid interviewing the same person twice,
availability: students do not have to walk by the ignore any random number that appears more than
interviewer in order to be included in the sample. once and continue down the list. This procedure pro-
The process of selection takes place without their duces what is called a simple random sample without
knowledge, and they do not know they will be replacement: no number is placed back in (replaced
asked to be part of a social survey until they are after having been selected) for a second chance at
contacted by someone associated with the study. inclusion. Since virtually all simple random samples
in social research are like this, the qualifier “without
Step 5 mentions the possible use of a table of replacement” is usually omitted. Strictly speaking,
random numbers, found in the back of many statis- “without replacement” means that the result is not a
tics books or generated by computer software. Such true simple random sample, as those chosen later have
tables are usually made up of columns of five-digit a greater chance of being selected than those selected
numbers; for example: earlier. (The first person chosen has one chance in 9000
of being selected; removing that person means person
09188 08358 two has one chance in the remaining 8999. Thus the
90045 28306 450th person chosen will have one chance out of the
8551 people remaining; a chance of one in 8551 is
73189 53840
greater than a chance of one in 9000.) R esearchers
75768 91757 generally overlook this p roblem, since it is a small one
54016 87941 compared with other sampling problems.
The first thing to notice is that, since these are Systematic sample
five-digit numbers and the population size is just A systematic sample is selected directly from the
9000 (a four-digit number), the only random num- sampling frame, without using random numbers.
bers that might be appropriate are the two that begin In the present case, 1 student in 20 is to be selected
with zeros, 09188 and 08358 (and the former is al- (450/9000). That “1 in 20” figure is called the sam-
ready larger than the largest possible case number pling interval. A random start begins the process,
for this study). To deal with this, drop one of the which in this case can be achieved by selecting at
digits in each number—say the first digit. Using only random a number from 1 to 20, possibly by using the
the last four digits gives us the following: last two digits in a table of random numbers. With
the 10 random numbers above, the first relevant one
9188 8358 is 54016, since it is the first one where the last two
0045 8306 digits yield a number in the desired range (1–20),
namely 16. This means that the sixteenth student in
3189 3840
the sampling frame is the first one to be included
5768 1757 in the sample. Thereafter, take every twentieth stu-
4016 7941 dent on the list; the sequence of case numbers would
be: 16, 36, 56, 76, 96, 116, and so on. Who is the last
Still, one of the resulting numbers—9188— person in the sample? Case number 8996.
exceeds 9000; no student has that case number. In systematic sampling, it is important to ensure that
The solution is simple: ignore that number and go there is no inherent ordering or pattern in the sampling
152 PART II Quantitative Research
frame, a feature called periodicity. For example, imagine criterion (in this case faculty) and selecting either a
that the sampling frame was set up such that case 1 was simple random sample or a systematic sample from
male, case 2 female, case 3 male, case 4 female, and so each of the resulting strata. In the present example,
on for all cases. In the example above, that would yield five faculties mean five strata, with the sample in each
a sample of 450 females and 0 males, in all likelihood a stratum being one-twentieth of the total for each fac-
very unrepresentative sample. If there is any pattern to ulty, as in Table 7.1, which also shows a hypothetical
the list, arrange the cases in random order or choose a outcome of a simple random or systematic sample,
different sampling method. one that does not exactly mirror the population.
Note that with systematic sampling, not every The advantage of stratified sampling is clear: it
possible combination of cases has an equal chance ensures that the sample is distributed in the same
of being selected. Once that first case number (here way as the population in terms of the stratifying
16) is chosen, every case number in the sample must criterion. Using a simple random or systematic sam-
end in 6, and no even-numbered prefixes will qual- pling approach may result in a distribution like that
ify, such as 26, 86, or 166. of the stratified sample, but it is unlikely. On the
other hand, this strategy requires that the relevant
Stratified random sampling criteria for stratification be known in advance of the
In the imaginary study of university students dis- research, which is not always the case.
cussed earlier, the student’s discipline may be relevant Two more points are relevant here. Stratified sam-
to what you will be researching. Generating a simple pling is feasible only when it is relatively easy to iden-
random sample or a systematic sample may yield one tify and allocate units to strata. Five strata based on
in which, for example, the proportion of kinesiology physical activity level, from “competitive athletes”
students in the sample is the same as that in the student at one end to “couch potatoes” at the other, would
population, but usually it is not an exact match. Thus be very hard to set up, because it would require an
if there are 1800 students majoring in kinesiology, a initial study just to get the population data on activ-
sampling ratio of 1 in 20 should produce 90 students ity level that you would need to create these strata.
in the sample from this faculty. But because of sam- Second, more than one stratifying criterion can be
pling error there might be, say, 85 or 93. used at the same time. In our example, we could
So long as data are available on each student’s fac- stratify by faculty and gender, or faculty, gender, and
ulty of registration, it’s possible to ensure that students whether students are undergraduates or postgrad-
are exactly represented in terms of their faculty uates, and so on, as long as the criteria are easy to use
membership by using stratified random sampling. and relevant to the research question. But this ap-
In the language of sampling, this means stratify- proach is practicable only when data on the relevant
ing the population (dividing it into subgroups) by a stratifying criteria are available for the population.
Does your school have an accessible list of female Prince Edward Island. An alternative solution would
undergraduate engineering students? be to select a stratified cluster sample, which combines
the notion of stratified sampling discussed in the pre-
Multi-stage cluster sampling vious section with cluster sampling. In our example,
In the previous example, the students to be inter- we could first group all universities by region—for ex-
viewed all attend the same university. However, im- ample, Atlantic Canada, Quebec, Ontario, the Prairies,
agine a national sample of students. A major problem and British Columbia—and then randomly sample
with large populations is the general lack of an ad- two universities from each region. The next step would
equate sampling frame. Is there a list somewhere of be to sample 500 students from each of the 10 uni-
all students registered at Canadian universities? No. versities, yielding a total sample size of 5000. (Again,
Without a sampling frame, it’s impossible to select a technical adjustments would be required because the
simple random or systematic sample. clusters are not of equal size.) These are the stages:
One way to deal with these problems is to employ
cluster sampling. With cluster sampling, the primary • Group the universities into five regions.
sampling unit (the first stage of the sampling proced- • Sample two universities from each region.
ure) is not the individuals or units of the population • Sample 500 students from each of the 10
to be studied, but an aggregate of them, known as a universities.
cluster. Imagine you want a nationally representative
sample of 5000 students. One solution is to sample Cluster sampling if done properly can be very ef-
universities and then students from each of the sam- fective. Its greatest merit is that it solves the problem
pled universities, with a probability sample at each of having no adequate sampling frame. The student
stage. For example, you could randomly sample 10 study described above does not require a list of all
universities from the entire population of universi- university students in Canada, something that would
ties, yielding 10 clusters, and then interview 500 ran- be very hard to come by. With cluster sampling, no
domly selected students at each of the 10 universities such list is ever required.
selected, to create a sample of 5000 students. (This With large populations, cluster sampling is very
would require “probability proportionate to size” economical compared to simple random or system-
sampling because each university has a different atic sampling when doing in-person interviews be-
number of students, but that technique goes beyond cause of reduced travel costs—research participants
the scope of our discussion here.) tend to cluster geographically. Using this technique
Suppose the result of sampling 10 universities is for research based on telephone interviews has less
the following: of an economic advantage now than it once did be-
cause long-distance phone charges are so low, and of
• Alberta course if the research is to be done on the Internet,
• McGill travel expenses are not an issue.
• Simon Fraser The advantages of multi-stage cluster sampling
• Winnipeg should be clear by now. However, as suggested above,
• Dalhousie even a very rigorous sampling strategy cannot elimin-
• Guelph ate all sampling error, as the example in Box 7.2 shows.
• Brandon
• Western
• Nipissing
The qualities of a
• Toronto probability sample
Many researchers prefer probability samples be-
Note the absence of universities in Saskatchewan, cause they allow one to make inferences from the
Newfoundland and Labrador, New Brunswick, and sample to the population from which it was selected.
154 PART II Quantitative Research
If an average of 9.7 units of alcohol is consumed in This consideration is important because sam-
the previous seven days by respondents in a prob- pling theory tells us that about 95 per cent of all
ability sample, would a similar figure be found in sample means lie within (+ or −) 1.96 standard
the population? The answer is complex but it is errors of the population mean (see Figure 7.5). In
sketched out here and in Box 7.3. Assume for the other words, we can be 95 per cent confident that
sake of illustration that it would be possible to our sample mean lies within 1.96 standard errors
take an infinite number of random samples of the of the population mean.
same size from a population. Not all the resulting To illustrate further, if a sample has been
sample means would be perfect estimates of the selected according to probability sampling princi-
population mean. We can never know the popu- ples, we can be 95 per cent certain that the popu-
lation figure with 100 per cent certainty, but we lation mean lies between
can imagine another study with a sample mean of
(a) [the sample mean] – [1.96 times the stan-
9.6 units, another with 9.8, and so on. These out-
dard error of the mean]; and
comes, as long as there are enough of them and
(b) [ the sample mean] + [1.96 times the stan-
the sample size is sufficiently high, will take the
dard error of the mean].
form of a bell-shaped curve known as a normal
distribution (see Figure 7.5) with the sample This is known as the 95 per cent confidence
means clustering around the population mean. Of interval. If the mean level of alcohol consumption
the sample means not exactly at the population in the previous seven days in our sample of 450
mean, half will be below the population mean students is 9.7 units and the standard error of the
and half above it. Moving to the left or to the right mean is 1.3, we can be 95 per cent certain that the
(away from the population mean), the curve tails population mean lies between
off, reflecting the smaller and smaller number of
9.7 – (1.96 × 1.3)
samples generating means that depart consider-
ably from the population mean. The variation of and
sample means around the population mean is the
9.7 + (1.96 × 1.3)
sampling error and is measured using a statistic
known as the standard error of the mean. This that is, between 7.152 and 12.248 units of alcohol.
is an estimate of the amount by which a sample Under certain conditions, the standard error of
mean is likely to differ from the population mean. the mean in a stratified sample is smaller than in
other probability samples. In other words, stratifi-
cation may inject an extra increment of precision
Number of samples
In other words, with probability samples, the sample sample sizes are 100 (minimum) and then 400, 900,
findings can be generalized to the population. This is 1600, and 2500. These sample size increases cut the
not to say that the sample and population character- sampling error by half, then by one-third, then by
istics are exactly the same, but the first can be used one-fourth, and then by one-fifth respectively. (The
to estimate the second with a known probability of reason has to do with the square roots involved in
error. In the example of alcohol consumption in the the denominator of the calculation; thus the ori-
sample of 450 students (in the previous section), the ginal square root, 10, for a sample of 100 becomes
mean number of units consumed by the sample (X) 20, 30, 40, and 50 [the square root of 2500], yield-
can be used to estimate the population mean (µ) with ing the one-half, one-third, one-quarter, and one-
a known margin of error. Greek letters are used to fifth.) Notice that the increases in precision become
denote population characteristics; the Latin alpha- less pronounced (going from 100 to 400 reduces
bet (the one used to write in English), for sample sampling error by half, but 400 to 900 reduces it by
characteristics. only one-third) and thus the rate of reduction in the
To expand on this point it is necessary to use standard error of the mean declines. Sample size is
some basic statistical ideas. These are presented in likely to be profoundly affected by considerations of
Boxes 7.2 and 7.3 but these can be skipped if all you time and cost at such a juncture, since striving for
need is a broad idea of sampling procedures. smaller and smaller increments of precision becomes
increasingly uneconomic as ever-larger samples
become less and less cost efficient.
Sample size Fowler (1993) warned against simple acceptance of
“How large should a sample be?” “Is the sample big this size criterion and argued that it is not normal for
enough?” No issue in sampling raises more ques- researchers to be able to specify in advance “a desired
tions than sample size. Decisions about sample size level of precision” (p. 34). Moreover, since sampling
are not straightforward, as we will see below. At the error is only one component of any error entailed in
same time, the quest for the perfect sample is usually an estimate, the notion of a desired level of precision
constrained by the practical considerations of cost is not realistic. Instead, to the extent that this notion
and time. affects decisions about sample size, it usually does so
in a general rather than a calculated way. It is usually
Absolute and relative sample size cost that ultimately determines sample size.
People unfamiliar with the field may be surprised to
learn that it is the absolute size of a sample that is Non-response
important, not the proportion of the population that Considerations about sampling error do not end
it makes up. This means that a national probability with issues of sample size. It’s also important to bear
sample of 1000 Canadians has about as much valid- in mind the problem of non-response.
ity as a national probability sample of 1000 Amer- The response rate can be defined as the percent-
icans, even though the latter involves a much larger age of the sample that actually participates in the
population. study. It is often a good deal less than 100 per cent,
Increasing the size of a sample increases the pre- partly because there are usually some people in the
cision of the estimates derived from it; for instance, sample who will either refuse or be unable to partici-
a larger sample size means that the 95 per cent con- pate, who cannot be contacted, or who turn out to be
fidence interval referred to in Box 7.2 is narrowed. unsuitable for the study. Others will be excluded if
However, a large sample cannot guarantee precision, they leave a large number of questions unanswered,
so it might be better to say that increasing the size of or if they do not take the interview or questionnaire
a sample probably increases the precision of the esti- seriously.
mates it can create. In other words, as sample size in- Unfortunately, survey response rates have been in
creases, sampling error tends to decrease. Common “steep decline” for decades (Johnson & Wislar, 2012),
156 PART II Quantitative Research
Assume a population of five cases (scores = 6, 8, appears where 12 and 14 meet: [12 + 14]/2 = 13.
10, 12, 14) from which a random sample of two Now count the number of the sample means in the
cases is selected. The five scores sum to 50 and table; there are 25. There is only one mean of 14,
therefore the mean of the population is 50/5, or in the bottom right corner, but five means of 10, on
10. Taking all possible random samples of two the diagonal. Thus five of the 25 sample means, or
cases with replacement (meaning that after a case 20 per cent of them, have a value of 10, the same
is chosen it goes back into the pool, so that two as the population mean. Sample means of 9 and
6s or two 14s are possible), and calculating the 11 are the next most frequent at 32 per cent (a
means of each sample of size two, the following combined total of 8 of 25 means). With 8 and 12,
set of means results: another 24 per cent (6 of 25) are now counted. So
sample means of 8, 9, 10, 11, and 12 represent 76
Value of first observation per cent of all possible sample means. The sample
in sample means furthest from the population mean, the 6
6 8 10 12 14 and 14 (in each corner), are the rarest, at 4 per
6 6 7 8 9 10 cent (1 of 25) each.
Value of second Those percentages are really probabilities.
8 7 8 9 10 11
observation in The probability of getting a sample mean of 7
10 8 9 10 11 12
sample is 2 chances in 25 or, as it is usually expressed, 8
12 9 10 11 12 13 chances in 100 or 8 per cent.
14 10 11 12 13 14 The logic of a confidence interval should now
be more apparent. A researcher normally takes
Thus the 7, where 6 and 8 meet, is the mean only one sample, but most random sample results
of those two values, namely: [6 + 8]/2 = 7; the 13 stay close to the population value. In our example,
with the added problem that there has been no con- research topic) and may have to be taken into ac-
sensus among researchers regarding how to define count in the analysis. Researchers sometimes take
and calculate response rates. In their research on steps to determine the level of non-response bias, as
sustainable consumption in the Edmonton area, Schieman and Narisada (2014) did when they con-
Kennedy et al. (2013) reported a response rate of cluded that it was not an issue in their study.
69 per cent. Schieman and Narisada (2014), who Given that 100 per cent response rates are rare,
used a national sample of the Canadian workforce sample size may have to be adjusted. For example, if
to examine people’s sense of mastery over their lives, the goal is a sample of 450 students and earlier surveys
achieved a response rate of about 40 per cent. How- suggest that the response rate will be approximately
ever, non-response bias (the extent to which people 80 per cent, it may be advisable to sample 575 indi-
included in the sample differ from the population as a viduals, on the assumption that approximately 115
whole) is increasingly considered to be more import- (20 per cent) will not participate, leaving 460 and
ant than the response rate itself (Johnson & Wislar, still room for some additional refusals.
2012). Are the people who cannot be contacted richer How far should researchers go to boost response
and on vacation? Are they ill and in hospital? Such rates? Smith (1995) showed that response rates are af-
differences may be important (depending on the fected by things like the subject matter of the research,
7 | Quantitative Sampling 157
76 per cent of the sample means are within (±) 2 example, suppose that the population mean value
units of the population mean. When plotted, these for testosterone is 10 and athlete X comes in at 14.
results don’t exactly take the shape of the bell While that is a possible (non-drugged) outcome, it
curve shown in Figure 7.5, but you can see the is rare. So if the doping committee is willing to be
broad outline. Including more and more cases in wrong 4 per cent of the time (the frequency with
the sample shapes the lines into a normal curve. which an honest 14 occurs), the athlete can be ac-
With a real sample, an extension can be cre- cused. With a reputation at stake, however, a 4 per
ated on each side of the sample mean to make it cent chance of error is too high. That is why the
more likely to be correct than if the sample mean tests allow for only much less likely errors.
by itself were used. The extension for a 95 per
cent confidence interval is equal to 1.96 times Sample
the standard error of the mean. (In Box 7.2 the Mean Frequency %
sample mean was 9.7, and the extensions were 14 1 4
± [1.96 × 1.3] or ± 2.548). The extensions around 13 2 8
the sample mean create what is called an interval. 12 3 12
There is only a 5 per cent chance that the interval
11 4 16
does not include the population mean for a 95 per
10 5 20
cent confidence interval.
9 4 16
This is basically how athletes using banned
8 3 12
substances are caught. They provide one bio-
7 2 8
sample (actually two, but one is a backup), which
the authorities then check against a more so- 6 1 4
phisticated version of the table to the right. For ∑ 25 100
the type of respondent, and the effort expended on respond to a questionnaire usually results in an im-
obtaining the cooperation of prospective respondents. proved response rate, at little additional cost.
Later in the chapter, a number of steps that can im-
prove response rates to questionnaires (which are par- Heterogeneity of the population
ticularly prone to poor response rates) are discussed. Yet another consideration is the heterogeneity of the
However, boosting response rates can be expensive. population from which the sample is to be taken.
Teitler et al. (2003) discussed the steps taken to in- When a population is heterogeneous with regard
crease the response rate of a US sample that was hard to characteristics relevant to the topic under study,
to reach, namely both unwed parents of newborn chil- as a whole country or city normally is, the samples
dren. They found that although increasing the sample drawn are likely to be highly varied. When it is rela-
from an initial 68 per cent to 80 per cent of all couples tively homogeneous, such as a population of students
meant that the final sample more closely resembled or members of a particular occupation, the amount
the population from which it had been taken, dimin- of variation is less. Generally, the greater the hetero-
ishing returns set in. This does not mean that steps geneity of a population, the larger a sample should be
should not be taken to improve response rates, how- in order to maximize the chances that all groups will
ever. For example, following up on people who do not be adequately represented.
158 PART II Quantitative Research
want in their principals, and who has several classes in Florida was based on a convenience sample, as
made up of teachers working part-time toward mas- was MacKinnon and Luke’s (2002) study of cultural
ter’s degrees. Now suppose that she decides to admin- change, and Beagan’s (2001) study of medical stu-
ister a questionnaire to those students—a sample of dents mentioned in earlier chapters.
convenience. Chances are that she will receive all or Convenience samples are also good for pilot
almost all of the questionnaires back, thus achieving studies. Imagine that part of the education professor’s
a high response rate. However, the use of the conven- research involves developing a battery of questions to
ience sampling strategy makes it impossible to gen- measure the leadership preferences of teachers that
eralize the findings to teachers as a whole: the very she plans to use later, with a different sample. Since it
fact that her students are taking this degree program is highly desirable to pre-test such a research instru-
marks them off as different from teachers in general. ment before actually using it, administering it to a
(In addition, of course, it’s ethically questionable for group that is not part of the main study is a legitimate
a researcher to ask people who are dependent on her way of exploring issues such as whether respondents
for a grade to take part in her research.) will reply honestly to questions on sensitive topics,
This is not to suggest that convenience samples and whether the questions are clear and comprehen-
should always be avoided. In fact, they are used more sible. A convenience sample may also be used to test
often than you might think. Mairs and Bullock’s newly created scales for reliability and to generate
(2013) research on sexually risky behaviour and ideas for further research. Box 7.4 contains another
HIV testing on the part of Canadian “snowbirds” illustration of how a convenience sample may be used.
Suppose you want to study university under- convenience sample, because the choices of univer-
graduates to find out the extent of their part-time sity, faculty, degree program, and course are made
employment while at school. To save time and purposively rather than randomly. Because of the
effort you decide to conduct the study at your own way the questionnaires are administered (in class),
university. You could use probability sampling to there is a very high response rate. On the other
get participants, but you think the research is not hand, students who are absent from class do not get
important enough to justify the cost. Instead you a chance to fill out a questionnaire. An important
choose five faculties, and within each, one or two question is whether absence from class is connected
of the specific degree programs it offers: anthro- in some way to part-time work, the latter variable
pology, visual arts, chemistry, electrical engin- being the focus of your study. In other words, is
eering, French, kinesiology, media studies, and absence higher among students who work part-time,
psychology. This choice of subjects is designed to perhaps because they are working at the time of the
maximize variety in the types of degree programs class? Are some students too tired to go to class be-
represented while providing similar numbers of cause of their part-time work? If so, you will probably
males and females (since in particular degree underestimate the proportion of students who work
programs one gender frequently predominates). part-time. Also, because you have selected a con-
Questionnaires can then be given to students in venience sample, you must say so in your research
first-, second-, third-, and fourth-year courses. report, acknowledging that the results cannot be
Together, these procedures represent a good generalized to Canadian students as a whole or even
attempt at generating a varied sample. It is a to the student population at your university.
160 PART II Quantitative Research
Becker reported on how he generated a sample of users who would be willing to discuss their
marijuana users: experiences with me . . . Although in the end
half of the fifty interviews were conducted
I had been a professional dance musician . . . with musicians, the other half covered a
and my first interviews were with people I wide range of people, including labourers,
had met in the music business. I asked them machinists, and people in the professions.
to put me in contact with other [marijuana] (1963, pp. 45–46)
7 | Quantitative Sampling 161
The problem with snowball sampling is that it is political opinion polling in some countries. The aim
very unlikely to be representative of the population, of quota sampling is to produce a sample that reflects
although, as Becker’s experience suggests, the very a population in terms of the relative proportions of
notion of a population is problematic in some cir- people in different categories (gender, ethnicity, age,
cumstances. However, snowball sampling is gener- socioeconomic status, region of residence, etc.), or in
ally used not within a quantitative research strategy, combinations of these categories. Unlike a stratified
but within a qualitative one: both B ecker’s work on sample, however, a quota sample is not random, since
marijuana users and Bryman’s (1995, 1999) studies the final selection of people is left up to the inter-
on theme parks used a qualitative research frame- viewer. Information about the stratification of the
work, and concerns about the ability to generalize Canadian population or about certain regions can be
do not loom large in qualitative research. Indeed, obtained from sources such as the census and from
Tastsoglou and Miedema (2003) warned their read- surveys based on probability samples such as the
ers not to generalize from their 40 women to all im- General Social Survey.
migrant women in the Maritimes, much less to those Once the categories and the numbers of people
in the rest of Canada. This is not to suggest that to be interviewed within each category (the quotas)
snowball sampling is entirely irrelevant to quantita- have been decided, interviewers merely have to
tive research: when the researcher needs to focus on select people who fit those categories. The quotas
relationships between people, tracing connections are typically interrelated. As in stratified sampling,
through snowball sampling can be a better approach the population may be divided into strata covering
than conventional probability sampling. Statistically several characteristics (for example, gender, employ-
small groups—gay francophone lawyers in New ment, and age) all at once (see Table 7.2). Census
Brunswick, for example—are researchable with a data can identify the number of people who should
snowball approach. Taking a random sample from be in each subgroup; thus the numbers to be inter-
that population would not be feasible. viewed in each subgroup reflect the population.
Interviewers can then seek out individuals who fit
Quota sampling the subgroup quotas. Accordingly, an interviewer
Quota sampling is not common in academic may be told to find and interview five 25- to 44-year-
social research, but it is used intensively for com- old u
nemployed females at an assigned location,
mercial purposes such as market research and in such as a mall.
TABLE 7.2 | S
trata for gender, age, and Structured observation and sampling
employment
Structured observation, like survey research, neces-
Older, female, Younger, female, sitates decisions about sampling. However, the issues
unemployed (5%) unemployed (4%) that have to be considered are not limited to sampling
Older, female, Younger, female, people. Several other sampling issues may be involved:
employed (17%) employed (20%)
Older, female, Younger, female, • If the structured observation is to take place in
other (7%) other (1%) public areas, there will be no sampling frames
Older, male, Younger, male,
(lists of people who frequent those areas). It would
unemployed (3%) unemployed (6%) be very difficult, for example, to get a complete
list of people who walk along a particular street,
Older, male, Younger, male,
or of men who frequent certain “tearooms.” One
employed (15%) employed (15%)
can, however, choose the people by using a table
Older, male, other (6%) Younger, male, other (1%) of random numbers: for example, the third,
then seventh, then eighteenth person seen in the
A number of criticisms are frequently levelled at public area. The same table can be used to estab-
quota samples: lish random samples of time, place, or activity.
• Time sampling involves an observer recording
• The proponents of probability sampling whatever is happening at a particular time,
argue that because the choice of respondent for example, in a pub every 15 minutes after
is left to the interviewer, a quota sample is not a random start time. This can be done on one
likely to be representative. As noted earlier, in person or on several individuals at once and
their choice of people to approach, interviewers the 15-minute figure can be varied using a
may be unduly influenced by their perceptions table of random numbers. It may be necessary
of friendliness or willingness to make eye con- that individuals watched on more than one day
tact (many people will look at the ground and not be observed at the same time of the day; the
shuffle past as quickly as possible, not wanting observation periods may have to be randomly
to be bothered). selected. For example, it would be an error to
• People who are in an interviewer’s vicinity at observe a certain pupil in a school classroom
interview times, therefore available to be ap- always at the end of the day, when most stu-
proached, may not be typical. There is a risk, dents are tired, since tiredness could give a
for example, that people not in full-time paid false impression of that pupil’s behaviour.
work may be over-represented, especially in
malls, making the sample unrepresentative. Time sampling should be combined with the next
• Judgments about eligibility may sometimes strategy:
be incorrect, for instance regarding the re-
spondent’s age, and even more so for social • Random sampling of specific places: for ex-
class. For example, someone who is actually ample, in a pub near the entrance, at the bar, by
young enough to be eligible but looks older the pool table, outside where the smokers con-
may not be asked because the quota of the gregate, the washroom area, and the booths.
older group is filled. In such cases an element • “Behaviour sampling,” whereby an entire group
of bias is introduced. is watched and the observer records a particular
• It is not appropriate to calculate a standard kind of behaviour. Thus in that same pub, the re-
error term from a quota sample; in fact, most searcher can observe the nth pick-up (second, then
statistical estimates are meaningless unless fifth, etc.) after the previous one was observed,
random sampling has been used. using a random number table to determine the n.
7 | Quantitative Sampling 163
sample can be generated by randomly selecting one November only, there might have been a preoccupation
day and then selecting every nth day thereafter. with Christmas fare, while findings from a summer
The time span analyzed in content analysis often issue might have been affected by the greater availabil-
depends on the research questions. Warde (1997) was ity of certain foods, such as particular types of fruit.)
interested in changes in representations of food (what
should be eaten and how it should be eaten) in food
columns in five of the most widely read weekly maga-
Reducing non-response
zines that explore topics related to household cuisine. There is little consistent evidence on whether response
He looked at two years, 20 years apart, and within rates are lower for interviews conducted over the tele-
each year at the February, May, August, and November phone rather than face-to-face (see Table 7.3). How-
issues to ensure that seasonal factors did not overly in- ever, there is a general belief that telephone interviews
fluence the findings. (If he had selected magazines from achieve slightly lower response rates than personal
TABLE 7.3 | C
omparing the strengths of different ways of contacting members of the
chosen sample
MODE OF SURVEY ADMINISTRATION
Face-to-face Telephone Posted
interview interview questionnaire Email Web
Resource issues
Is the cost of the mode of ✓ ✓✓ ✓✓✓ ✓✓✓ ✓✓✓
administration relatively low?
Is the speed of the mode of ✓ ✓✓✓ ✓✓✓ ✓✓✓ ✓✓✓
administration relatively fast?
Is the cost of handling a dispersed ✓ (✓✓ if ✓✓✓ ✓✓✓ ✓✓✓ ✓✓✓
sample relatively low? clustered)
Sampling-related issues
Does the mode of administration ✓✓✓ ✓✓ ✓ ✓ ✓
tend to produce a good response
rate?
the questionnaire to fit a booklet format. As with formats, although fewer and fewer people lack these
structured interviewing, begin with questions characteristics. Other factors that make such surveys
most likely to be of interest to the respondent. problematic include the following:
• Follow up individuals who do not reply at
first, possibly with two or three further mail- • Many people have more than one email
ings. The importance of reminders cannot be address.
overstated—they do work. One approach is to • Some households have one computer but sev-
send out a reminder letter to non-respondents eral users.
two weeks after the initial mailing, reasserting • Internet users are a biased sample of the popu-
the nature and aims of the survey and sug- lation in that they tend to be better educated,
gesting that the person contact a member of the wealthier, younger, and more urban than non-
research team to obtain a replacement copy of users, although these differences are dimin-
the questionnaire should the original one have ishing in importance as the Internet becomes
been lost. Then, two weeks after that, those who more accessible (Hallett & Barber, 2014).
still have not responded should be sent another • Few sampling frames exist for the general online
letter along with a second copy of the question- population, and most of these are likely to be
naire. These reminders have a demonstrable either confidential or too expensive to acquire.
effect on the response rate. Some writers argue
for even more than two mailings of reminder Such issues make it difficult to satisfy all the
letters. If a response rate is worryingly low, fur- criteria for conducting probability sampling. The
ther mailings are certainly desirable. chief problem with virtual sampling strategies is
• Monetary incentives increase the response rate that the representativeness of the sample is almost
but can be deemed unethical. They are most ef- always in question. However, open source software
fective if they arrive with the questionnaire (as like Amazon Mechanical Turk is making it easier
opposed to after its return). Apparently, most for social researchers to gain access to online par-
respondents will not take the money and dis- ticipants, and one could argue that virtual sampling
card the questionnaire. The evidence also sug- provides better access to large populations than
gests that quite small amounts of money have other forms of convenience sampling. Regardless of
a positive impact on the response rate, but that the type of sampling chosen, it is always necessary to
larger amounts do not necessarily improve the discuss the limitations of the method used.
rate any further. Online surveys are becoming more popular and
have potential if only for their low cost. Several prob-
Virtual sampling issues lems have been identified with Web and email sur-
A major limitation of online surveys is that not veys, but methodologists are beginning to come to
everyone is online and has the technical ability grips with them and may gradually develop ways of
to handle questionnaires in either email or Web overcoming their limitations.
Key Points
• Probability sampling is a mechanism for reducing • Randomly selected samples are important be-
bias in sample selection. cause they permit generalizations to the popula-
• Key technical terms in sampling include repre- tion of interest.
sentative sample, random sample, non-response, • Sampling error generally decreases as sample size
population, and sampling error. increases.
168 PART II Quantitative Research
• Under certain circumstances, quota samples can • Sampling and sampling-related error are just two
be an alternative to random samples, but they sources of error in social survey research.
have some deficiencies. • Response rates vary by the medium of communi-
• Convenience samples can provide useful data, but cation and most can be improved by persistent
have limited generalizability. follow-up procedures.
of the class then uses a random number table to 3. Each person in the class calculates the gender break-
select simple random samples of increasing size, down for each sample selected in exercise 1, and
for example, n = 1, n = 10, n = 20, n = 40, and so compares it to the overall gender breakdown of the
on, which are used to calculate sample means. population. The class is asked: How do these break-
The class is then asked: How does the size of the downs differ by sample size? What does it mean to
sample affect its accuracy? Which sample size has “stratify” a population by gender? Would stratifying
the least amount of sampling error? Why? be necessary in order to have the gender breakdown
2. Each member of the class then selects a sys- of the samples match that of the population?
tematic sample (every nth case, using a random 4. The mean number of kilometres travelled by all
start) the size of which matches that of the largest females in the population is calculated, as is the
simple random sample selected in exercise 1. For mean for all males. The means for women and
example, if the population size was 80 and the lar- men are then calculated for each sample selected
gest simple random sample selected had a size of in exercise 1. The class is asked: How is the accur-
40, every second case (80/40) would be selected acy of the means affected by sample size? Would
for the systematic sample. The class is asked: How it be appropriate to stratify by gender for this
close is the mean of the systematic sample to that variable? Why or why not? How could a gender-
of the largest simple random sample? Why are the stratified sample be selected to calculate the
two sample means not always the same? mean number of kilometres travelled to campus?
Relevant Websites
Statistics Canada provides a good general intro- These YouTube videos illustrate how to select a
duction to sampling. Both probability and non- simple random, stratified, and cluster sample.
probability sampling techniques are described here. www.youtube.com/watch?v=yx5KZi5QArQ&NR=1
Test your knowledge of sampling by doing the exer-
www.youtube.com/watch?v=sYRUYJYOpG0
cises provided.
www.statcan.gc.ca/edu/power-pouvoir/ www.youtube.com/watch?v=QOxXy-I6ogs&
ch13/5214895-eng.htm feature=related
Statistics Canada provides the sampling techniques (Websites accessed 23 October 2018)
Chapter Overview
This chapter presents some of the most basic and frequently used methods for analyzing quantitative
data. To illustrate them, we’ll use an imaginary data set on attendance at a gym—the kind of data set that
could be generated from a small research project suitable for most undergraduates.
The chapter explores:
• the importance of anticipating questions of analysis early in the research process, before all the
data have been collected;
• the different kinds of variables generated in quantitative research (knowing how to distinguish
them is crucial in deciding which statistics to use);
• methods for analyzing a single variable (univariate analysis);
• methods for analyzing relationships between two variables (bivariate analysis); and
• analysis of relationships among three or more variables (multivariate analysis).
If you’re like many students in the social sciences, the for a while. Feel better now? If not, remember that the
idea of learning about statistics doesn’t fill you with joy. best way to learn a new language is to start with the
How about communing with nature? Yes, that would be basics. That’s what we’ll do in this chapter. In fact, we’re
much better—open fields, wild winds, the bright sky, not going to go beyond a basic level, and we won’t even
a riot of colour provided by the plants and animals be using any formulas—we’ll leave that for your statis-
around you. On the other hand, it’s often said that tics courses. What we will do is introduce the logic of
mathematics is the language that nature speaks, and statistics, which will give you the conceptual founda-
statistics is a form of applied mathematics. Communing tions you need to do quantitative data analysis. We’ll
with nature via statistics can be just as invigorating as also discuss some important ways of organizing quan-
communing any other way, especially if you work at it titative data. Think of it as a new language class.
▲ Igor Stevanovic/123rf
8 | Quantitative Data Analysis 171
Questionnaire Code
1. Are you male or female? (please tick)
Male ✓ 1 2
Female
2. How old are you?
21 Years 21
3. Which of the following best describes your main reason for going to the gym?
(please circle one code only)
Relaxation 1
Maintain or improve fitness ✓ 2
Lose weight 3
Meet others 4
Build strength 5
Other (please specify) 6
4. When you go to the gym, how often do you use the cardiovascular
equipment (treadmill, step machine, bike, rower)? (please tick)
Always ✓ 1
Usually 2
Rarely 3
Never 4
5. When you go to the gym, how often do you use the weights? (please tick)
Always ✓ 1
Usually 2
Rarely 3
Never 4
6. Generally, how frequently do you go to the gym? (please tick)
Every day 1
4–6 days a week 2
2 or 3 days a week ✓ 3
Once a week 4
2 or 3 times a month 5
Once a month 6
Less than once a month 7
8 | Quantitative Data Analysis 173
7. Do you usually go with someone else to the gym or usually on your own?
(please circle one code only)
On my own ✓ 1
With a friend 2
With a partner/spouse 3
1 1 21 2 1 1 3 1 2 8 33 17 5
2 2 44 1 3 1 4 3 1 2 10 23 10
3 2 19 3 1 2 2 1 1 1 27 18 12
4 2 27 3 2 1 2 1 2 8 30 17 3
5 1 57 2 1 3 2 3 1 4 22 0 15
6 2 27 3 1 1 3 1 1 3 34 17 0
7 1 39 5 2 1 5 1 1 5 17 48 10
8 2 36 3 1 2 2 2 1 1 25 18 7
9 1 37 2 1 1 3 1 2 8 34 15 0
10 2 51 2 2 2 4 3 2 8 16 18 11
11 1 24 5 2 1 3 1 1 1 0 42 16
12 2 29 2 1 2 3 1 2 8 34 22 12
13 1 20 5 1 1 2 1 2 8 22 31 7
14 2 22 2 1 3 4 2 1 3 37 14 12
15 2 46 3 1 1 5 2 2 8 26 9 4
16 2 41 3 1 2 2 3 1 4 22 7 10
17 1 25 5 1 1 3 1 1 1 21 29 4
18 2 46 3 1 2 4 2 1 4 18 8 11
19 1 30 3 1 1 5 1 2 8 23 9 6
20 1 25 5 2 1 3 1 1 1 23 19 0
21 2 24 2 1 1 3 2 1 2 20 7 6
22 2 39 1 2 3 5 1 2 8 17 0 9
23 1 44 3 1 1 3 2 1 2 22 8 5
24 1 999 1 2 2 4 2 1 4 15 10 4
25 2 18 3 1 2 3 1 2 1 18 7 10
26 1 41 3 1 1 3 1 2 8 34 10 4
27 2 38 2 1 2 5 3 1 2 24 14 10
28 1 25 2 1 1 2 1 2 8 48 22 7
29 1 41 5 2 1 3 1 1 2 17 27 0
30 2 30 3 1 1 2 2 2 8 32 13 10
31 2 29 3 1 3 2 1 2 8 31 0 7
32 2 42 1 2 2 4 2 1 4 17 14 6
33 1 31 2 1 1 2 1 2 8 49 21 2
34 2 25 3 1 1 2 3 2 8 30 17 15
35 1 46 3 1 1 3 1 1 3 32 10 5
36 1 24 5 2 1 4 1 1 2 0 36 11
37 2 34 3 1 1 3 2 1 4 27 14 12
38 2 50 2 1 2 2 3 2 8 28 8 6
39 1 28 5 1 1 3 2 1 1 26 22 8
40 2 30 3 1 1 2 1 1 4 21 9 12
41 1 27 2 1 1 2 1 1 3 64 15 8
42 2 27 2 1 2 4 2 1 4 22 10 7
43 1 36 5 1 1 3 2 2 8 21 24 0
44 2 43 3 1 1 4 1 2 8 25 13 8
45 1 34 2 1 1 3 2 1 1 45 15 6
46 2 27 3 1 1 2 1 1 4 33 10 9
47 2 38 2 1 3 4 2 2 8 23 0 16
8 | Quantitative Data Analysis 175
48 1 28 2 1 1 3 3 1 2 38 13 5
49 1 44 5 1 1 2 1 2 8 27 19 7
50 2 31 3 1 2 3 2 2 8 32 11 5
51 2 23 2 1 1 4 2 1 1 33 18 8
52 1 45 3 1 1 3 1 1 2 26 10 7
53 2 34 3 1 2 2 3 2 8 36 8 12
54 1 27 3 1 1 2 3 1 3 42 13 6
55 2 40 3 1 1 2 2 1 4 26 9 10
56 2 24 2 1 1 2 1 1 2 22 10 9
57 1 37 2 1 1 5 2 2 8 21 11 0
58 1 22 5 1 1 4 1 1 1 23 17 6
59 2 31 3 1 2 3 1 1 4 40 16 12
60 1 37 2 1 1 2 3 2 8 54 12 3
61 2 33 1 2 2 4 2 2 8 17 10 5
62 1 23 5 1 1 3 1 1 1 41 27 8
63 1 28 3 1 1 3 3 2 8 27 11 8
64 2 29 2 1 2 5 2 1 2 24 9 9
65 2 43 3 1 1 2 1 2 8 36 17 12
66 1 28 5 1 1 3 1 1 1 22 15 4
67 1 48 2 1 1 5 1 1 4 25 11 7
68 2 32 2 2 2 4 2 2 8 27 13 11
69 1 28 5 1 1 2 2 2 8 15 23 7
70 2 23 2 1 1 5 1 1 4 14 11 5
71 2 43 2 1 2 5 1 2 8 18 7 3
72 1 28 2 1 1 4 3 1 2 34 18 8
73 2 23 3 1 1 2 1 2 8 37 17 17
74 2 36 1 2 2 4 2 1 4 18 12 4
75 1 50 2 1 1 3 1 1 2 28 14 3
76 1 37 3 1 1 2 2 2 8 26 14 9
77 2 41 3 1 1 2 1 1 4 24 11 4
78 1 26 5 2 1 5 1 1 1 23 19 8
79 2 28 3 1 1 4 1 2 8 27 12 4
80 2 35 2 1 1 3 1 1 1 28 14 0
81 1 28 5 1 1 2 1 1 2 20 24 12
82 2 36 2 1 1 3 2 2 8 26 9 14
83 2 29 3 1 1 4 1 1 4 23 13 4
84 1 34 1 2 2 4 2 1 8 24 12 3
85 1 53 2 1 1 3 3 1 1 32 17 6
86 2 30 3 1 1 4 1 2 8 24 10 9
87 1 43 2 1 1 2 1 1 2 24 14 10
88 2 26 5 2 1 4 1 1 1 16 23 7
89 2 44 1 1 1 4 2 2 8 27 18 6
90 1 45 1 2 2 3 3 2 8 20 14 5
Source: Rojek (1995).
which in turn is a greater frequency than “2 or 3 days neither more nor less of something than “maintain
a week,” and so on. Compare this to questions 3, 7, or improve fitness” is.
and 9 where the categories can’t be rank ordered. In In short, different types of variables can be gener-
the case of question 3, for instance, “relaxation” is ated in the course of research. The three main types
176 PART II Quantitative Research
are distinguished by looking at the relationship be- the unit is the minute and the difference be-
tween the categories of the variable as manifested by tween the categories is a one-minute inter-
any one individual. The three are: val. Thus a person who spends 32 minutes
on cardiovascular equipment is spending one
• Nominal variables. These variables, also minute more than someone who spends 31.
known as categorical variables, are composed That difference is the same as the difference be-
of categories that have no relationship to one tween someone who spends 8 minutes and an-
another except that they are different. In the other who spends 9 minutes on the equipment.
case of religion, for example, the categories At the interval/ratio level of measurement, it
might be Roman Catholic, Protestant, Jewish, is possible to say that two scores are equal or
Muslim, Hindu, Other, and No Religion. Any not equal (as at the nominal level), for example,
two research participants will be either in 32 ≠ 31; that they have greater or lesser values
the same category or in different categories: (as at the ordinal level), for example, 32 > 31,
no other kind of comparison is possible. This 8 < 9; and that the distance or amount of dif-
means that the order of the categories is arbi- ference between them can be specified, for ex-
trary: it could have begun with No Religion ample, 32 − 31 = 1. This is the highest level of
and ended with Roman Catholic. Switching measurement, the one that allows the widest
the order has no implications for the way the range of analysis techniques, including addi-
data are interpreted. tion, subtraction, multiplication, and division.
• Ordinal variables. With these variables the Sometimes a distinction is made between
different categories can be rank ordered. That interval and ratio variables. Ratio variables are
means that “greater than” (>) and “less than” interval variables with a fixed and non-arbitrary
(<) statements can be made about the categor- zero point (as in a wind speed of 0 kilometres per
ies and the people in them. For example, in hour). Many social science variables exhibit this
question 5 “always” indicates greater frequency quality, for example, income, age, and years of
than “usually,” which indicates greater fre- school completed. In the discussions that follow
quency than “rarely,” and so on. Note also that we will use the more inclusive term “interval/
it would be illogical to re-order the categories ratio,” but to understand, take two incomes:
as, say, “always, never, sometimes, usually.” $30K and $45K. They are nominally unequal
The distance or amount of difference between and the first is less than the second, in fact
the categories, however, may not be equal across $15K less. But with that real 0, one can also say
the range. Thus the difference between the cat- that the first is 2/3 of the second (or the second is
egories “always” and “usually” is probably not 1.5 times the first). The first income can also be
the same as the difference between “usually” said to stand in a 2:3 ratio to the second.
and “rarely,” and so on. This is because no unit of
measurement is used for this question. Ordinal The three main types of variable and illustra-
variables are similar to words that have either a tions from the gym survey are provided in Table 8.1.
comparative “er” or a superlative “est” attached Strictly speaking, items that have Likert-style re-
to them, as in “slimmer” and “slimmest.” sponse categories (see Box 4.1) produce ordinal vari-
• Interval/ratio variables. These are variables ables. H
owever, many writers argue that they can be
for which a unit of measurement exists and treated as though they produce interval/ratio vari-
thus the distances or amount of difference ables because of the relatively large number of cat-
between the categories can be made identical egories they generate. For a brief discussion of this
across the range of categories. The values take issue, see Bryman and Cramer (2001, pp. 58–59).
the form of actual numbers (such as 1, 2, 7.5, etc.). Figure 8.1 provides guidance about how to identify
In the case of variables var00010 to var00012, variables of each type.
8 | Quantitative Data Analysis 177
TABLE 8.1 | T
ypes of variable
Type Description Examples in gym study Variable name in SPSS
Nominal Variables whose categories cannot be var00001 gender
rank ordered; also known as categorical var00003 reasons
var00007 accomp
var00008 othsourc
var00009 exercise
Ordinal Variables whose categories can be rank var00004 carduse
ordered but the distances between the var00005 weiuse
categories are not equal across the range var00006 frequent
Interval/ratio Variables where the distances between var00002 age
the categories are identical across the var00010 cardmins
range var00011 weimins
var00012 othmins
Frequency tables
Can the categories be rank ordered?
A frequency table provides the number and percent-
age of people (or whatever else constitutes a case in
Yes No Variable is nominal/ the study) belonging to each of the categories of the
categorical
variable in question and can be created for all three
Are the distances between the categories equal? variable types. An example of a frequency table for a
nominal variable is provided for var00003 (reasons)
in Table 8.2. The table shows, for example, that 33
Yes No Variable is ordinal members of the sample go the gym to lose weight and
that they represent 36.7 per cent (percentages are often
Variable is interval/
rounded in frequency tables) of the entire sample.
ratio Frequency tables are typically created using soft-
FIGURE 8.1 Deciding how to categorize a variable
ware. A very commonly used software program
for statistical analysis is IBM SPSS Statistics (SPSS).
TABLE 8.2 | Frequency table showing reasons for visiting the gym
REASONS FOR VISITING
Frequency Per cent Valid per cent Cumulative per cent
Valid Relaxation 9 10.0 10.0 10.0
Fitness 31 34.4 34.4 44.4
Lose weight 33 36.7 36.7 81.1
Build strength 17 18.9 18.9 100.0
Total 90 100.0 100.0
178 PART II Quantitative Research
Relaxation 50
Build strength 10.0 %
18.9%
40
30
Count
Fitness
34.4%
20
Lose weight
36.7% 10
0
FIGURE 8.3 Pie chart showing main reasons for 20 and 21–30 31–40 41–50 51 and
under over
visiting the gym (SPSS output) Ages of gym visitors
as one that does not discriminate: the brightest average amount of time spent on the cardiovascu-
and least bright students get fairly close marks.) In lar equipment (26.5 minutes) higher than for the
the gym study, is there more or less variability in the weights (14.9 minutes), there is more deviation from
amount of time spent on cardiovascular equipment the mean too. The standard deviation is also affected
compared with weights? by outliers, which are sometimes excluded.
The most obvious way of measuring dispersion is
by calculating the range. This is simply the differ- Bivariate analysis
ence between the maximum and the minimum value Bivariate analysis examines whether there is a rela-
in a distribution of interval/ratio scores. (Range can tionship between two variables. Several techniques
be applied to an ordinal variable too, but in that case are available for examining relationships, but their
it is more definitional than descriptive: for example, use depends on the level of measurement of the two
a range from “Working class” to “Upper class.”) The variables being analyzed. Figure 8.5 details the main
range is 64 minutes for the cardiovascular machines types of bivariate analysis according to the types of
and 48 minutes for the weights. This suggests that variable involved.
there is more variability in the amount of time spent
on the former, probably because some people spend Contingency tables
almost no time on cardiovascular equipment. How- Contingency tables are probably the most flexible of
ever, like the mean, the range is influenced by out- all methods of analyzing relationships in that they
liers, such as respondent 41, who spent 64 minutes can be employed in relation to any pair of variables,
on the cardio equipment. from nominal to interval/ratio. They are not the most
Another measure of dispersion is the standard efficient method, however, especially for interval/
deviation, which is a measure of variation around ratio data, which is the reason why the method is not
the mean. It is calculated by taking the difference recommended in all the cells in Figure 8.5. A con-
between each value in a distribution and the mean, tingency table is like a frequency table except that it
squaring it, dividing the total of these squared allows two variables to be analyzed simultaneously,
differences by the number of values, and then taking so that relationships between them can be examined.
the square root. The standard deviation for var00010 It is normal for contingency tables to include percent-
(cardmins) is 9.9 minutes and for var00011 (weimins) ages, since they make the tables easier to interpret.
it is 8.0 minutes. This indicates that not only is the Table 8.4 examines the relationship between gender
FIGURE 8.5 Methods of bivariate analysis, one variable on top, other variable on side
8 | Quantitative Data Analysis 181
and reasons for visiting the gym using our survey TABLE 8.4 | Contingency table showing
data. “Gender” is the presumed independent variable the relationship between
and for that reason becomes the column variable—a gender and reasons for
common preference among researchers. “Reasons” visiting the gym
is the presumed dependent or row variable (but see
GENDER
Box 8.3). In this case, gender is assumed to influence
reasons for going to the gym; reasons for going to the Male Female
gym cannot influence gender. The percentages are Reasons No. % No. %
column percentages, representing the independent Relaxation 3 7 6 13
variable. The number in each cell is calculated as a Fitness 15 36 16 33
percentage of the total number in its column. Thus in Lose weight 8 19 25 52
the case of the top left-hand cell, the 3 men—of 42 in Build strength 16 38 1 2
all—who go to the gym for relaxation represent 3/42
Total 42 100 48 100
or 7 per cent of the men in the sample.
be spurious or non-causal: that is, the relation- environment, influences both breastfeeding and
ship between two variables may be caused by a cognitive abilities. She found greater support for
third variable. For example, Rippeyoung (2013) the spurious effect position, observing that there
examined whether the effects of breastfeeding were no significant differences in breastfeeding
can explain the gaps in cognitive ability between rates between the three economic groups. Spur-
“non-poor,” “near poor,” and “poor” children in ious relationships are discussed in more detail
Canada (a purportedly causal relationship), or later in this chapter.
whether a third factor, a rich educational home
Contingency tables are generated to look for when plotted on a scatter diagram, the values of the
patterns of association. In this case, there are clear two variables will approximate a straight line (even
gender differences in reasons for visiting the gym. though they may be scattered as in Figure 8.9), not a
As our student anticipated, females are much more curve. A scatter diagram should be created to test for
likely than males to go to the gym to lose weight. linearity before Pearson’s r values are used in a study.
They are also somewhat more likely to go to the
gym for relaxation. By contrast, men are much more Scatter diagrams
likely to go to the gym to build strength. There is Scatter diagrams are shown in Figures 8.6 through
little gender difference in terms of fitness as a reason. 8.9. If one variable can be identified as likely to be
the independent variable, it is by convention placed
Pearson’s r on the X-axis, the horizontal axis.
Pearson’s r is a statistic used to examine relation- The scatter diagrams are based on the information
ships between two interval/ratio variables. Its chief in Box 8.4, which gives imaginary data for five inter-
features are as follows: val/ratio variables. The scatter diagram for variables
1 and 2 is presented in Figure 8.6 and shows a per-
• The coefficient has values from 0 (which indi- fect positive relationship, which yields a Pearson’s r
cates that there is no linear relationship what- correlation of +1. This means that as one variable in-
ever between the two variables) to +1 or −1 (the creases, the other variable increases as well, and the
first indicating a perfect positive relationship, value of one variable perfectly predicts the value of the
the second a perfect negative). other. If the correlation is below 1, at least one other
• The closer a positive coefficient is to 1, the variable is affecting the relationship between them,
stronger the relationship; the closer it is to possibly affecting variable 1 as well as variable 2.
zero, the weaker the relationship. The scatter diagram for variables 2 and 3
• Similarly, the closer a negative coefficient is to (Figure 8.7) shows a perfect negative relationship,
−1, the stronger the relationship is; the closer it which yields a Pearson’s r correlation of −1. This
is to zero, the weaker the relationship. means that as one variable increases, the other de-
• The sign of the coefficient (positive or negative) creases, and that the values of one variable perfectly
indicates the direction of a relationship. N
egative predict the values of the other. Here, as in the previ-
means that as one variable is going up, the other ous case, the line could be extended beyond the data
is going down; it doesn’t matter which is which. to predict what would happen to either variable if
Positive means that the two are going in the the other moved beyond the range shown, although
same direction, either both up or both down. it is risky to go beyond what is known. Also, because
social data usually include so many exceptions that
For Pearson’s r to be used, the relationship be- few cases fall exactly on the line, predictions based
tween the two variables must be broadly linear: on them tend to be general rather than specific.
8 | Quantitative Data Analysis 183
12 60
10 50
8 40
Variable 1
Variable 3
6 30
4 20
2 10
0 0
0 10 20 30 0 10 20 30
Variable 2 Variable 2
FIGURE 8.6 Scatter diagram showing a perfect FIGURE 8.7 Scatter diagram showing a perfect
positive relationship negative relationship
30
relationship
Variables 10
1 2 3 4 5
1 10 50 7 9
2 12 45 13 23 0
0 10 20 30
3 14 40 18 7 Variable 2
4 16 35 14 15
FIGURE 8.8 Scatter diagram showing two variables
5 18 30 16 6
that are not related
6 20 25 23 22
7 22 20 19 12
30
8 24 15 24 8
9 26 10 22 18
10 28 5 24 10
20
Variable 4
2 and 4, whose scatter diagram appears in Figure 8.9. var00006 (frequent; see Table 8.1). Using Kendall’s
There is clearly a positive relationship; in fact, the tau-b to calculate the correlation between the first
Pearson’s r is .88 (positive correlations are usually pre- two variables—frequency of use of the cardiovascu-
sented without the + sign). This means that the varia- lar and weights equipment—it turns out to be very
tion in the two variables is very closely connected, but low. A slightly stronger relationship is found between
that other variables also have some influence. var00006 (frequency of going to the gym) and
Going back to the gym survey, the correlation be- var00010 (cardmins, the amount of time spent on
tween age (var00002) and the time spent on weights, the cardiovascular equipment): close to .4. Another
weimins (var00011), is −.27 (see Table 8.5): a weak option for relating two ordinal variables is to use
negative relationship. This suggests that older people Spearman’s rho (though it can only accommodate a
tend to spend less time on such equipment than small number of ties: that is, cases with the same
younger ones, but that other variables also influence rank on a variable). It ranges from −1, a perfect nega-
the time people spend there. tive relationship, to +1, a perfect positive relation-
ship. For example, look at the participation rank (A)
Kendall’s tau-b and popularity rank (B) of a group of seven people.
Kendall’s tau-b is designed for pairs of ordinal vari-
ables, but is also used (as Figure 8.5 suggests) when Person Mary John Bill Sally Kim Susan Joe
one variable is ordinal and the other is interval/ratio. A 1 2 3 4 5 6 7
(Notice the general rule that data can be moved B 1 2 3 4 5 6 7
down a level, interval to ordinal, but not up.) It is
exactly the same as Pearson’s r in terms of possible Spearman’s rho is 1.0; knowing a person’s partici-
outcomes in that the computed value of Kendall’s pation rank perfectly predicts the popularity rank.
tau-b can be positive or negative, and varies from It is called a rank correlation coefficient to remind
0 to ±1. In the gym study there are three ordinal vari- you of its ordinal character and to distinguish it
ables: var00004 (carduse), var00005 (weiuse), and from Pearson’s r.
TABLE 8.5 | Correlations output for age, weimins, and cardmins (SPSS output)
Correlations of p < 0.05 are “flagged” with asterisks
CORRELATIONS
AGE WEIMINS CARDMINS
AGE Pearson Correlation 1.000 –.273 ** –.109
Sig. (2-tailed) . .010 .311
N 89 89 89
WEIMINS Pearson Correlation −.273 ** 1.000 –.161
Sig. (2-tailed) .010 . .130
N 89 90 90
CARDMINS Pearson Correlation −.109 –.161 1.000
Sig. (2-tailed) .311 .130 .
N 89 90 90
** Correlation is significant at the 0.01 level (2-tailed).
Cramér’s V is suitable for examining the strength Squaring eta, Kendall’s tau-b, Spearman’s rho, and
of a relationship between two nominal variables (see Pearson’s r produces statistics that measure how
Figure 8.5). Its coefficient ranges from 0 to 1. It is much of the variation in one variable can be ex-
always positive because nominal categories cannot plained or predicted by the other variable. Thus if r
be rank ordered; their values cannot go up or down. is −.27 (the correlation between weimins and age),
The value of Cramér’s V for the relationship be- r2 is .0729. This can be expressed as a percentage by
tween gender and reasons for going to the gym multiplying r2 by 100: thus 7.29 per cent of the varia-
(var00003, reasons) is .503, a moderate relationship. tion in the time spent on weights can be predicted by
Cramér’s V is usually reported with a contingency age. This also shows that a strong correlation like .7
table and a chi-square test (see below). It is not nor- explains only 49 per cent of the variance. For nom-
mally presented on its own. inal data, squaring Cramér’s V (.503 for gender and
reasons) provides an approximation of the explained
Comparing means and eta variance, so about 25 per cent of the variation in rea-
There are many other bivariate statistics but we will sons for visiting the gym is attributable to gender.
present just one more. To examine the relationship
between an interval/ratio variable and a nominal Statistical significance
variable (if the latter can be construed as the in- and inferential statistics
dependent variable), one can compare the means of One difficulty with working on sample data is a lin-
the interval/ratio variable for each subgroup of the gering worry about whether the findings can be gen-
nominal variable. As an example, consider Table 8.6, eralized to the population from which the sample
which presents the mean number of minutes spent was drawn. As we saw in Chapter 7, there is always
on cardiovascular equipment (var00010, cardmins) the possibility of sampling error, even when prob-
for each of the four categories of reasons for going to ability sampling procedures are followed (as in the
the gym (var00003, reasons). The four means shown gym survey). If this happens, the sample is to some
suggest that people who go to the gym for fitness or degree unrepresentative of the wider population.
to lose weight spend considerably more time on this To make matters worse, there is no practical way of
equipment than those who go to the gym to relax or finding out for sure how extensive the sampling error
to build strength. is, as no one can afford the time or money to study
The statistic eta can be calculated with these vari- every case in the population—that’s why a sample is
ables; it expresses the level of association between used in the first place. This is where various tests of
them. Since one variable is nominal (meaning no statistical significance come in.
rank ordering), its values are always positive; and What is a test of statistical significance? It provides
it has a range from 0 to 1. The eta for the data in an indication of the risk we are taking when we use
Table 8.6 is .48, a moderate relationship between the a particular sample statistic to estimate a population
two variables. characteristic (see Box 8.5). For example, the mean
TABLE 8.6 | C
omparing subgroup means: Time spent on cardiovascular
equipment by reasons for going to the gym
REASONS
Time Relaxation Fitness Lose weight Build strength Total
Mean number of minutes spent 18.33 30.55 28.36 19.65 26.47
on cardiovascular equipment
N 9 31 33 17 90
186 PART II Quantitative Research
There is always a chance that a rejected null hy- where none exists in the population. Therefore,
pothesis is actually true. The amount of risk as- we would have less confidence in generalizing
sociated with rejecting a true null hypothesis is with that level of significance than with p = .05.
called the level of statistical significance. The If we want a more stringent test than the latter
maximum acceptable risk in the social sciences is p-level provides (for instance, because of concern
5 chances in 100. This means that, in 100 samples, about the use that might be made of the results),
about 5 of them would exhibit a relationship when we can choose the p = .01 level. This means that
there is not one in the population. (We have to say the risk is 1 in 100 that the results could have
“about” 5 because this is based on probabilities, come about by chance alone (i.e., as a result of
not certainties.) Any one sample of the 100 sam- sampling error). If p = .01 has been chosen, the
ples might be one of those five, but the risk is fairly null hypothesis should not be rejected if the re-
small. This significance level is denoted by p = .05 sults are statistically significant at the p = .05
(p means “probability”). level but not the p = .01 level. In what instances
A significance level of p = .1 means that about would it be appropriate to choose p = .01 rather
10 in 100 samples would show a relationship than p = .05?
age of the people in the gym sample is 33.6 years. associated with rejecting the null hypothesis
Using the concept of the standard error of the mean, (implying that there is a relationship in the
the 95 per cent confidence interval for the population population) when in fact the null is correct
mean extends from 31.72 to 35.47 years of age. The (there is no relationship in the population).
risk that this range does not contain the population Levels of statistical significance are expressed
mean is 5 per cent. Chapter 7 (see Box 7.2), in its dis- as probability levels: that is, the probability of
cussion of the standard error of the mean, revealed rejecting the null hypothesis when it is in fact
some of the ideas behind statistical significance. The true. The convention among most social re-
rest of this section looks at tests of significance for searchers is that an acceptable level of statistical
measures of bivariate association. All the tests have a significance is p .05, meaning that there are at
common structure, illustrated in the following steps. most 5 chances in 100 that the sample shows a
relationship not also found in the population.
• Set up a null hypothesis. A null hypothesis is a This risk level varies, depending on the circum-
hypothesis that is to be disproved; for example, stances. For example, 5 chances of error in 100
that there is no association between two vari- would be too many in the case of a drug being
ables, or that two populations do not differ on tested for its efficacy, especially a drug with side
some characteristic. In the context of our gym effects. Probably p would be set at .00001, ac-
survey, a null hypothesis might state that there cepting only one chance in 100,000 of an error.
is no relationship between gender and reasons • Determine the statistical significance of the
for visiting the gym. If your data lead to a re- findings. Use a statistical test such as chi-
jection of the null hypothesis, you will have square (pp. 187–89).
indirect support for the research hypothesis • Decide whether to reject or not reject the null
that there is a relationship in the population. hypothesis. Reject the null if the findings are
• Establish an acceptable level of statistical sig- statistically significant at the .05 level—that is,
nificance. This is essentially the level of risk if the risk of error (i.e., getting a relationship as
8 | Quantitative Data Analysis 187
strong as the one that was found when in fact finding might be statistically significant, but is it im-
there is no relationship in the population) is no portant? Recall that statistical significance gets easier
higher than 5 in 100. Such results are unlikely to achieve as sample size goes up. Also, it is important
to have occurred by chance alone. to appreciate that tests of statistical significance can
be conducted only on probability samples.
Two types of error are possible when inferring
statistical significance (see Figure 8.10). A Type I Correlation and statistical significance
error occurs when a null hypothesis that is actually Examining the statistical significance of a correlation
true is rejected. This means that the results are the from a randomly selected sample provides informa-
product of chance, and that the researcher was mis- tion about the likelihood that the correlation exists
taken in concluding that there was a relationship in in the population. For instance, with a Pearson’s r of
the population. The level of significance is the prob- −.62 in the sample, what is the likelihood that there
ability of making a Type I error, so a p = .05 level of is no relationship between the two variables in the
significance means a greater likelihood of making a population? How likely is a −.62 to occur by chance
Type I error than a p = .01 level of significance does. alone, that is, because of sampling error?
The second type of error occurs when the null hy- Whether a correlation coefficient is statistically
pothesis should be rejected but is not. This is called a significant or not is affected by two factors:
Type II error. These errors are more likely when the
significance level is .01 than when it is .05, because • the size of the computed coefficient
using .01 makes it less likely that the null will be re- • the size of the sample
jected. The two types of error cannot be minimized
at the same time; if you decrease the chances of mak- This second factor may appear surprising, but it
ing a Type I error, you increase the chances of making is true for all statistics. The larger a sample, the more
a Type II error, and vice versa. Researchers usually likely it is that a computed correlation coefficient is
make the conservative choice and seek to minimize statistically significant. Thus, even though the cor-
Type I error; this makes it less likely that the null will relation between age and the amount of time spent
be rejected in error, and hence less likely that the re- on weight machines in the gym survey is just −.27—a
search hypothesis will be erroneously supported. fairly weak relationship—it is statistically significant
If an analysis reveals a statistically significant find- at the p = .01 level. This means that there is only one
ing, this does not mean that the finding is important. chance in 100 that there is no relationship between
It simply means that the results are probably not due age and weights in the population. Because the statis-
to chance alone. For example, a study may find a cor- tical significance of a correlation coefficient depends
relation between time spent jogging per week and the so much on the sample size, it is important always to
amount of money spent on running shoes. Such a examine both the correlation coefficient and the sig-
nificance level. This is true for any calculated statistic.
Error
This treatment of correlation and statistical sig-
Type I Type II
(risk of rejecting the (risk of failing to reject nificance applies to both Pearson’s r and Kendall’s
null hypothesis when the null hypothesis when tau-b. A similar interpretation can also be applied to
it should be confirmed) it should be rejected)
Cramér’s V and the chi-square test.
frequency—that is, one that would occur on the basis of less than one chance in 10,000 of rejecting a true null
chance alone. Think of the days of the week and crime: hypothesis (that is, inferring that there is a relationship
one might expect that 14.29 per cent (1/7) of all crimes in the population when none exists). We can be ex-
would occur on each day of the week. That is what would tremely confident that there is a relationship between
be expected if there were no relationship between day gender and reasons for visiting the gym among all gym
of the week and crime. The data say otherwise; more members, since the chance of obtaining a sample that
crimes are committed on Friday and Saturday. The chi- shows this c hi-square value when there is no relation-
square value, which for the data in Table 8.4 is 22.726, is ship among all gym members is less than 1 in 10,000.
calculated by taking the differences between the actual But a chi-square value is also affected by the
and expected values for each cell in the table and then number of cases. You want a large chi-square to
summing those differences (the operation is slightly reject a null hypothesis, and a larger n makes this
more complicated than this, but the details need not easier to achieve. This is why it is necessary to look
concern us here). The chi-square value means nothing at the data and not just at the final statistic. Exam-
by itself and can be meaningfully interpreted only in ine Table 8.7, which shows the level of self-esteem
relation to its associated level of statistical significance, for three categories of people (those who engage in
which in this case is p < .0001. This means that there is contact sports, those who engage in non-contact
8 | Quantitative Data Analysis 189
TABLE 8.7 | H
ow chi-square is affected by increasing the size of n and not affected by
changes to column headings
A B C
Contact Non-contact No Contact Non-contact No Contact Non-contact No
Self-esteem sports sports sports sports sports sports sports sports sports
H 7 3 10 14 6 20 24 26 50
M 5 10 15 10 20 30 10 20 30
L 12 13 25 24 26 50 14 6 20
x = 1.93, not sig.
2
x = 5.08, p < .05
2
x = 5.08, p < .05
2
sports, and those who do not engage in any sports). made up of two types: variation within each of the four
Because the independent variable is nominal, chi- subgroups that make up the independent variable, and
square is more appropriate than Kendall’s tau-b. variation between them. The latter is often called the
Note the information in the bottom row of the table. explained variance (explained by the group one is in)
The first two parts of Table 8.7, A and B, show and the former the error variance. A test of statistical
that just by doubling the size of the sample, the re- significance for the comparison of means entails relat-
sults change from being very likely due to chance in ing the two types of variance to form what is known as
A (thus the null hypothesis of no relationship cannot an F statistic, which expresses the amount of explained
be rejected) to less than 5 chances in 100 that they oc- variance in relation to the amount of error variance. In
curred by chance in B, and thus a rejection of the null. the case of the data in Table 8.6, the resulting F statistic
This is problematic because it suggests that one could is statistically significant at the p < .001 level. This find-
collect more and more data until the null can be re- ing suggests that there is less than one chance in 1000
jected. Now look at the last part of the table, C. Why that there is no relationship between the two variables
was it included? When the data from B for high and among all gym members.
low self-esteem are switched around in C, which in
most instances should mean something, chi-square Multivariate analysis
stays the same. This illustrates the point that statis- Multivariate analysis entails the simultaneous analy-
tical significance pertains only to whether the results sis of three or more variables, and can only be intro-
may have resulted from chance alone, and says noth- duced here. It is sometimes called elaboration as it is
ing about practical significance or importance. There more complicated and creates a more valuable picture
is a big practical difference between a situation in than bivariate analysis: one rarely finds an adequate
which half the respondents at all levels of sports par- explanation of a particular phenomenon by looking at
ticipation have low self-esteem and one in which half only one possible cause. (For more information on the
have high self-esteem, but that makes no difference techniques, consult a textbook on quantitative data
to the level of statistical significance—it remains the analysis: for example, Agresti & Finlay, 2009; Healey
same in each situation. The practical implications of & Prus, 2016.) There are three main contexts in which
a finding go beyond its level of statistical significance. multivariate analysis is used. Each is explained below.
produced because each variable is itself related to a Controlling in this instance might involve taking only
third variable. Think about the positive relationship people who had high levels of physical vibrancy, and
between the number of fire engines at a fire and the seeing whether there is still a positive relationship be-
fire damage: the more engines, the greater the devas- tween income and self-esteem among them.) If physical
tation. Does this positive association mean that the vibrancy were an intervening variable, that would
number of fire engines present causes the amount weaken or eliminate the association between income
of fire damage? Of course not. The size of the fire and self-esteem. Also, notice the “control” word? Yes,
accounts for both the number of engines respond- the purpose of controls is to make cross-sectional re-
ing and the damage. Variation in the two variables search more like an experiment in which random as-
(number of fire engines and amount of damage) is signment makes all other things equal or controlled.
caused by a third factor, the size of the fire.
Consider a social science example. Assume there Is there an interaction?
is a positive relationship between income and voting If the relationship between two variables holds for some
behaviour such that the more income a person has, groups or situations but not for others, an interaction
the more likely he or she is to vote for a conservative exists. The word “interaction” is used in a statistical
party. Does this mean that having more money causes sense here, one that is different from its everyday mean-
people to move to the right? Or could this relationship ing. In statistical terms, if the effect of one independent
be explained by age (see Figure 8.11)? The older one is, variable varies at different levels of a second independ-
the more likely one is to have a higher salary; and the ent variable, there is an interaction. In the gym study,
older one is, the more conservative one tends to be. If for example, is the relationship between age and other
the apparent relationship between income and voting sources of regular exercise (var00008, othsourc) dif-
behaviour is in fact a function of age, that relationship ferent for men and women? Table 8.8 shows the rela-
is spurious. See Box 8.3 for another example. tionship between age and other sources of exercise and
includes both men and women. (Age has been broken
Is there an intervening variable? down into just three age bands to make the table easier
An intervening variable suggests that the relation- to read.) The table suggests that the 31-to-40 age group
ship between the two original variables is not a direct is less likely to have other sources of regular exercise
one. Assume that there is a positive relationship be- than the 30-and-under and 41-and-over age groups.
tween income and self-esteem—the higher one’s However, Table 8.9, which breaks the relationship
income, the more positive one feels about oneself. But down by gender, suggests that the patterns for males
maybe there is more to it than that. Maybe income af- and females are somewhat different. Among males,
fects overall levels of physical vibrancy (richer people the pattern shown in Table 8.9 is very pronounced (the
tend to be more physically fit and to smoke less than
poorer people), which in turn affects self-esteem:
TABLE 8.8 | C
ontingency table showing
income → physical vibrancy → level of self-esteem the relationship between age
If this is true, what should happen to the relationship and whether gym visitors
between income and self-esteem if physical vibrancy have other sources of r egular
is controlled? (To control means to hold constant. exercise (percentages)
AGE
Age
Other source of 30 and 41 and
exercise under 31–40 over
Other source 64 43 58
Income Voting behaviour No other source 36 57 42
N 42 23 24
FIGURE 8.11 A spurious relationship
8 | Quantitative Data Analysis 191
TABLE 8.9 | C
ontingency table showing the relationship between age and whether
gym visitors have other sources of regular exercise for males and females
(percentages)
GENDER
Male Female
Other source of exercise 30 and under 31–40 41 and over 30 and under 31–40 41 and over
Other source 70 33 75 59 50 42
No other source 30 67 25 41 50 58
n 20 9 12 22 14 12
31-to-40 age group is less likely to have other sources The ANOVA section of the output indicates a signifi-
of exercise than the other two age groups), but for fe- cance or p-value of .010, which is the significance level of
males the likelihood of having other sources of exercise the model as a whole. Our model at this point consists of
declines with age. We can say that the relationship be- only one independent variable (age). The final section of
tween age and having other sources of exercise is mod- the output, “Coefficients,” has two subsections: “Unstan-
erated by gender. Another way to put this is that there dardized coefficients” and “Standardized coefficients.”
is an interaction between age and gender. The unstandardized coefficient for age is −.245, which
represents the estimated change in the dependent vari-
Other uses for multivariate analysis able for each unit change in the independent variable. In
Multivariate analysis can also be used to determine our example, the model predicts that for each one-year
how much of the variation in the dependent variable is increase in age, the time spent on weights decreases by
explained or predicted by the independent variables. .245 minutes. The standardized coefficient of −.273 indi-
In addition, it provides a test to find out which, if any, cates that for every standard deviation increase in age,
of the independent variables are significant predict- the time spent on weights decreases by .273 standard
ors after controlling for the others. These uses will be deviation units. Standardized coefficients (sometimes
illustrated by examining multiple linear regression. called beta weights) are useful in comparing the effect
Our brief discussion of this topic will also acquaint of independent variables that may be measured in dif-
you with some key concepts in multivariate analysis. ferent units: the higher the standardized coefficient, the
Consider the variable weimins, the number of more important the independent variable is considered
minutes spent on weights. What might cause it to to be. Finally, the significance level for age is shown to be
vary? One relevant factor might be age. Perhaps the .010, indicating that the probability that the association
older people get, the less likely they are to work out between age and weimins is due to chance alone is 1 in
using weights. To test this idea, we could perform a 100: this suggests it’s a pretty safe bet that there is a rela-
bivariate regression with weimins as the dependent tionship between those two variables in the population.
variable and age as the independent variable. As we have noted, a single cause or independent
The first part of Table 8.10 shows an R square variable is rarely if ever enough to provide a good ex-
value of .074, which indicates that age by itself ex- planation for anything. Let’s see what happens if we
plains 7.4 per cent of the variation in weimins—not put a second independent variable—gender—into the
very much. The value of R square tends to be some- model. Normally, the variables used in linear regression
what inflated, especially if the sample size is small (as have to be at the interval/ratio level of measurement, but
in our example) or the number of independent vari- a special procedure allows us to use nominal variables
ables is large (Agresti & Finlay, 2009, p. 366). A more like gender as well. Such variables have to be converted
realistic estimate is the adjusted R square, which in to indicator or “dummy” variables, a process that goes
our case has a value of .064 or 6.4 per cent.
192 PART II Quantitative Research
ANOVAa
Model Sum of Squares df Mean Square F Sig.
1 Regression 417.236 1 417.236 6.999 .010b
Residual 5186.719 87 59.617
Total 5603.955 88
a
Dependent variable: weimins
b
Predictors: (Constant), age
COEFFICIENTSa
Unstandardized Standardized
Coefficients Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) 23.209 3.218 7.213 .000
age −.245 .093 −.273 −2.645 .010
a
Dependent variable: weimins
beyond the scope of our discussion here. Assume that units in weimins for each standard deviation increase
the dummy variable we have created is called gender2. in age (controlling for gender2). The significance level
Table 8.11 shows the output generated after gender2 for age is .003, which means that age remains signifi-
was added to the model. The “Model Summary” cant even when gender2 is held constant. Had the sig-
shows that the R-square value has increased to .228 nificance level exceeded .050, we would have had to
(or that 22.8 per cent of the variation in weimins can consider dropping age from the model.
be explained or predicted by the new model), and that The “Coefficients” section shows an unstandard-
the more conservative adjusted R-square has risen to ized coefficient of 6.251 for gender2. Because of the way
.210 (or 21.0 per cent). This indicates that by adding this dummy variable was coded, the coefficient means
gender2 we have produced a much better model, al- that, holding age constant, the model estimates that the
though the majority of the variation in weimins is average time men spend on weights is 6.251 minutes
still unaccounted for. The ANOVA significance level greater than that spent by women. The standardized
is .000, which indicates that the findings for the coefficient of .393 tells us that according to the model,
new model as a whole (which includes both age and the average time men spend on weights is .393 standard
gender2 as independent variables) have a probability deviation units higher than that of women (controlling
of less than 1 in 1000 of being due to chance alone. for age). Also, gender2 has a p-value of .000, indicating
The “Coefficients” sections show an unstandard- that the association observed between it and weimins
ized coefficient of −.260 for age, which means that, (after controlling for age) has a probability of less than
controlling for gender2, for each one-year increase in 1 in 1000 of being due to chance alone. Had the signifi-
age, the model predicts a decline of .260 minutes spent cance level been greater than .050, we would have had
on weights. The standardized coefficient for age is good reason to exclude gender2 from the model. But
−.289, indicating a decrease of .289 standard deviation with these results, it stays in.
8 | Quantitative Data Analysis 193
TABLE 8.11 | M
ultiple regression output with gender2 (gender dummy variable)
in the model
MODEL SUMMARY
Model R R-square Adjusted R-square Std. Error of the Estimate
1 .478a .228 .210 7.09088
a
Predictors: (Constant), gender2, age
ANOVAa
Model Sum of Squares df Mean Square F Sig.
1 Regression 1279.830 2 639.915 12.727 .000b
Residual 4324.125 86 50.281
Total 5603.955 88
a
Dependent variable: weimins
b
Predictors: (Constant), gender2, age
COEFFICIENTSa
Unstandardized Standardized
Coefficients Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) 20.831 3.010 6.920 .000
age −.260 .085 −.289 −3.053 .003
gender2 6.251 1.509 .393 4.142 .000
a
Dependent variable: weimins
police that were lower than those for non-Indigenous trust, contact with police, total criminal victimiza-
people on all six measures, and all the differences tions, and several socio-demographic variables. Cao
were statistically significant. In the multivariate an- suggests that “the differential confidence between
alyses, the association between being Indigenous Aboriginal and non-Aboriginal people found here is
and having lower levels of confidence in the police only the tip of the iceberg in the complicated rela-
on the composite measure remained significant at tionship between the police and Aboriginal people”
p < .05 even after 13 relevant variables were con- (516), and that “the dominant mythology of Canada
trolled for, including visible minority status, social as a non-racist nation must be challenged” (517).
Checklist
Checklist for performing and writing up quantitative data analysis
☐☐ Have missing data codes been specified for inferences are included, are their limitations
all variables? outlined?
☐☐ Are the statistical techniques used appro- ☐☐ If the data come from a cross-sectional
priate for the level of measurement (that is, design, have unsustainable inferences about
nominal, ordinal, or interval/ratio)? causality been resisted?
☐☐ Are the most appropriate and powerful tech- ☐☐ Does the analysis go beyond univariate to
niques for answering the research questions include bivariate and even multivariate
used? analyses?
☐☐ If the sample is not randomly selected, are ☐☐ Are the research questions answered, and
inferences about a population avoided? If only the analyses relevant to them presented?
Key Points
• Think about data analysis before designing re- • Become familiar with computer software such as
search instruments. SPSS , which is discussed in the Appendix.
• Know the difference between nominal, ordinal, • Don’t confuse statistical significance with import-
and interval/ratio variables. Techniques of data ance or practical significance.
analysis are applicable to some types of variable
and not others.
2. How many times have you shopped in the last RR In what circumstances is each of the following used:
month? Please write the number here ____. Pearson’s r, Kendall’s tau-b, Cramér’s V, Spearman’s
rho, eta?
3. For what items do you most enjoy shopping?
AA Your study indicates that the Pearson’s r coeffi-
Please tick one only.
cient for the association between the variables
Clothes (including shoes) ____
“years of formal education” and “income” is 0.6.
Food ____ How much of the variation in income can be pre-
Things for the house/apartment ____ dicted by years of formal education?
Presents or gifts ____
Entertainment (CDs, videos, etc.) ____ Statistical significance
RR What does “statistical significance” mean and
4. How important is it to you to buy clothes with
how does it differ from “importance” or “practical
designer labels?
significance”?
Very important ____
AA What does it mean to say that an eta of .42 is statis-
Fairly important ____ tically significant at p < .05?
Not very important ____
Not at all important ____ RR What does the chi-square test achieve?
AA Your results indicate that the chi-square statis-
Univariate analysis tic for the variables “gender” and “highest level
RR What is an outlier? How does the presence of an of education achieved” has a significance level of
outlier affect the mean and the range? p = .12. What does that mean?
AA You have just gathered data that include a vari-
able measuring the participants’ ethnic identity. Multivariate analysis
Which measure of central tendency would be most RR Define and provide an example of each of the
appropriate for this variable? Explain. following:
What information is conveyed by the contingency ANOVA table, the unstandardized and standard-
table? What do the chi-square results indicate? ized coefficients for age and frequent, and the
g. Produce the correlation matrix shown in Table 8.5. significance levels for age and frequent.
The class is asked to interpret the correlation
coefficient and significance level shown in 2. This activity does not require access to SPSS . After
each cell. Then they are asked: Why is there a presenting an overview of multivariate analysis,
diagonal of “1s” in the table? Why is some of the instructor divides the class into small groups.
the information in the table redundant? Each group is to come up with a real or hypothet-
h. Produce the regression output shown in Table ical example of:
8.10. The class is asked to interpret the R a. a spurious relationship;
square and adjusted R square values, the sig- b. an intervening variable; and
nificance level shown in the ANOVA table, the c. a statistical interaction.
unstandardized and standardized coefficients
The examples must be original, i.e. they cannot be
for age, and the significance level for age.
taken from the textbook or lecture materials.
i. Introduce a second variable into the regression
analysis by including the variable that measures Each group must then explain, at least conceptually,
the frequency with which patrons go to the gym how a researcher would go about testing for each
(frequent). For the purposes of this exercise, one. That is, they are to explain how one can deter-
disregard any level of measurement issues. The mine if a particular relationship is spurious, whether
class is asked to interpret the R square and ad- a certain variable is in fact an intervening variable,
justed R square values, then compare them to and whether an interaction exists, and illustrate that
what was found for the bivariate regression with the examples they just came up with. Once the
done in exercise h above, explaining any sim- groups have completed their work, they present
ilarities or differences. The class is then asked their material to the rest of the class for comment
to interpret the significance level shown in the and critique.
Relevant Websites
Statistics Canada offers a useful introduction to To learn more about big data, visit the Surveillance
quantitative data gathering and analysis. Studies Centre (Queen’s University) website.
www.statcan.gc.ca/edu/power-pouvoir/ch3/ www.sscqueens.org/
5214783-eng.htm
Information on big data is also available from the
This YouTube video explains levels of measurement. website of Ryerson University’s Privacy and Big Data
www.youtube.com/watch?v=B0ABvLa_u88 Institute.
www.ryerson.ca/pbdi/
This YouTube video illustrates the notion of correlation.
www.youtube.com/watch?v=SaSpZdf1oHU (Websites accessed 23 October 2018)
PART III
Qualitative Research
© kendo_OK/Shutterstock.com
9
The Nature of Qualitative Research
Chapter Overview
Qualitative analysis uses mainly words and images as data rather than numbers. Qualitative researchers
tend to produce inductive, constructionist, and interpretivist studies, although they do not necessarily
subscribe to all three of those perspectives. This chapter will consider:
• the main steps in qualitative research (although they are not followed as closely as the stages of
quantitative research);
• the relationship between theory and research in qualitative studies;
• the role of concepts in qualitative research;
• criteria for evaluating qualitative research;
• the main goals of qualitative researchers: seeing through the eyes of research participants; rich de-
scription; presentation of context, including the presence of conflict; concepts and theories as the
outcomes of research; and seeing social life and interaction as processes rather than static events;
• some common criticisms of qualitative research; and
• the main contrasts between qualitative and quantitative research.
Imagine that you are a non-Indigenous person par- although important, doesn’t address the core issues
ticipating in the National Inquiry into Missing and involved. How important would it be for you, in order
Murdered Indigenous Women and Girls. While some to reach a full understanding of this event, to find out
estimates maintain that the number of victims since what the murder meant to the victim’s family and
1980 is well into the thousands, you have been ap- friends? What would be the most appropriate way to
pointed to provide an in-depth study of one particular get the information you need to conduct the study? If
woman who was killed in northern Saskatchewan five it involves interviewing people who knew the victim,
years ago. Imagine as well that you’re free to gather how could you get access to them? What sort of inter-
the “facts” of the crime—time of death, physical view would you use? What would you talk about?
cause of death, the victim’s whereabouts before the Would it be useful to weave a historical account of col-
murder, and so on—but you feel that such information, onialism in Canada into your analysis? How would you
▲ MarkHatfield/iStockphoto
9 | The Nature of Qualitative Research 199
construct your research so as to not harm those who to fully understand what happened? This chapter dis-
have already experienced hardship? Will consultation cusses how qualitative researchers deal with these
and community involvement improve the research? sorts of questions as they conduct their analyses.
Finally, is it even possible for a non-Indigenous person
11. General research question(s) physical posters, and word of mouth. There
were 22 people, 12 women and 10 men, chosen
for the study, all of whom were unmarried
and living with their same-sex partner in the
22. Selecting relevant site(s) and subjects Greater Toronto Area.
• Step 3: Collection of relevant data. The authors
conducted in-depth, open-ended interviews in
33. Collection of relevant data
which the participants were encouraged to talk
about their current and former romantic rela-
5b.
5b Collection of further data tionships, their views on marriage, how their
relationships were viewed by friends and rela-
44. Interpretation of data tions, their levels of social support, and their
future relationship goals. The interviews typ-
ically lasted from 90 to 120 minutes.
• Step 4: Interpretation of data. One of the key
55. Conceptual and theoretical work
findings to emerge from the data analysis was
5a.
5a Tighter specification of that participants felt that their relationships
the research question(s) were not viewed as having the same level of legit-
66. Write up findings/conclusions imacy as those of married same-sex couples.
One participant stated that “if I were straight
FIGURE 9.1 An outline of the main steps of qualita- I might not be that into marriage. But for our
tive research queerness, I think for some of our families . . .
it’s necessary to take us seriously as a couple”
notions of marriage will be used as an example of (Lyon & Frohard-Dourlent, 2015, p. 413). The
how this is done: researchers also found that several of the people
interviewed explained their qualms about mar-
• Step 1: General research question(s). Since 2005, riage in personal rather than ideological terms.
all same-sex couples in Canada have had the Instead of discussing the social or political im-
right to get married. While some people in the plications of marriage, opposition to it often
LGBTQ community view the change in the law took the form of personal preferences. For ex-
as a victory for marriage equality, others have ample, one interviewee said that “it’s nice that
reservations. For instance, the legal right to people ask [if we will get married], but I say ‘No,
marry may bring with it conventional assump- we’re not into it,’” (p. 417) and left it at that.
tions of what marriage is supposed to entail, • Step 5: Conceptual and theoretical work.
such as the belief that child-rearing and mon- No new concepts emerged from Lyon and
ogamy are key features of the institution. Lyon Frohard-Dourlent’s work, but their article pro-
and Frohard-Dourlent (2015) conducted quali- vided good illustrations of key terms used by
tative interviews to explore how unmarried researchers in this area, such as “heteronorma-
gay couples who cohabit deal with the tension tive,” “heterosexist,” “homonormativity,” and
that exists between the acceptance of marriage “queer-based meaning systems.” In terms of
as a civil right and how having that right may theory, they proposed that the hard-fought
impact the nature of LGBTQ relationships. political battle for marriage equality that was
• Step 2: Selecting relevant site(s) and subjects. won by mainstream LGBTQ movements may
Participants were recruited through blogs and have made people in same-sex relationships
websites (including the University of Toronto reluctant to criticize marriage as an institu-
online announcements board), email lists, tion, which in the long run could influence
9 | The Nature of Qualitative Research 201
regard their concepts not as “definitive” but as of a concept can be as a guide to empirical inquiry. If
“sensitizing”—providing “a general sense of r eference it is too general, it will fail to provide a useful start-
and guidance in approaching empirical instances” ing point because its guidelines are too broad. If it is
(1954, p. 7). For Blumer, concepts should provide a too narrow, it is likely to repeat some of the difficul-
very general idea of what to look for, allowing the ties he identified with “definitive” concepts. How-
researcher to discover the variety of forms that the ever, his view is important in that it illustrates the
phenomenon identified by the concept can take. value of beginning with a broad definition of a con-
But Blumer’s distinction is not without problems. cept and then narrowing it down during the course
It is not clear how useful a very general formulation of data collection (for an example, see Box 9.1).
Is this woman’s smile spontaneous and sincere, or is she just doing her job?
Hochschild’s concept of emotional labour—labour and in the developing area of the sociology of emo-
that “requires one to induce or suppress feelings in tions. She gathered data on emotional labour by ex-
order to sustain the outward countenance that pro ploring how airline workers manage to keep smiling
duces the proper state of mind in others” (1983, p. 7)— at truly obnoxious customers. Hochschild’s initial
has become very influential in the sociology of work conceptualization emerged from a questionnaire
continued
204 Part III Qualitative Research
she had created on a related topic. To develop the attendants in the San Francisco Bay Area”
idea of emotional labour she gained access to Delta (Hochschild, 1983, p. 15).
Airlines, and in the course of her investigations she: For a contrasting occupational group also in-
• watched training sessions for flight attendants volved in emotional labour, she interviewed five
and had many conversations with both trainees debt collectors. Her book explored topics such as
and experienced attendants; the human costs of emotional labour and the issue
• interviewed airline personnel such as man- of gender in relation to it. It’s clear that Hochschild’s
agers and advertising agents; concept of emotional labour began with a some-
• examined Delta advertisements spanning 30 what imprecise idea that was gradually developed
years; to address its wider significance. The concept has
• observed the flight attendant recruitment pro- been picked up by other qualitative researchers in
cess at Pan American Airways (because she had the sociology of work. For example, Leidner (1993)
not been allowed to do this at Delta); and did an ethnographic study of a McDonald’s restau-
• conducted “open-ended interviews last- rant and an insurance company to investigate how
ing three to five hours each with thirty flight organizations seek to “routinize” emotional labour.
the people observed. The establishment of credibility • Researchers may show pre-publication ver-
entails both following proper research procedures and sions of their articles or books to their subjects.
submitting the findings to the people studied for con- Skeggs (1994), for example, asked the young
firmation that the account is consistent with the way working-class women who were the focus of
they see their world. The researcher is, after all, de- her ethnographic study to comment on draft
scribing their reality. This practice of getting feedback chapters (see Box 10.9 for further details).
on the study from the participants involved is referred • Researchers may provide the study partici-
to as respondent validation, member validation, pants with an opportunity to ask questions
or member reflection. about the methods and findings of the re-
The aim of respondent or member validation is search, offer feedback and criticism, and even
to seek corroboration or criticism of the researcher’s collaborate with the researcher on producing
observations and interpretations. It can take several the final product (Tracy, 2010).
different forms:
In each case, the goal is to confirm that the re-
• Each research participant may be provided searcher’s findings and impressions are congruent
with an account of what he or she said to the with those of the people on whom the research was
researcher or others. For example, in a study conducted and, if there are discrepancies, to find
that explored issues of trust between moth- out why. However, the idea is not without practical
ers of children with disabilities and school difficulties:
principals, Shelden et al. (2010) gave all re-
spondents a summary of the findings by either • Respondent validation may lead some research
telephone or email. They also asked each par- participants to respond defensively and even
ticipant if their views were represented in the demand censorship.
findings, and whether verbatim quotes were • Bloor (1997, p. 45) observed that research par-
accurate. ticipants who develop relationships of “fondness
and mutual regard” with the researcher may be A team of three researchers collected data over four
reluctant to criticize the study. days through observation, interviews, photography,
• It’s doubtful that research participants can and video records. The researchers conducted sev-
validate all of a researcher’s analysis, since it eral trustworthiness tests, including respondent
will be written for social scientists and will validation. In addition, they submitted their draft
therefore include many specialized references, manuscript and data set to three peers whose task
concepts, theories, and contextual issues it “was to criticize the project for lack of sufficient
unfamiliar to them. Skeggs (1994, p. 86), for data for drawing its conclusions, if they saw such
example, reported that the “most common re- a void” (1988, p. 456). The same study highlighted
sponse” among her subjects was “Can’t under- some problems with this approach, among them the
stand a bloody word it says.” fact that, because qualitative research generates ex-
tremely large data sets, validation is very demanding
Transferability for the auditors. That may be the major reason why
Because qualitative research typically entails in-depth, auditing has not been widely used.
intensive study of a small number of people, qualitative
findings tend to flow out of the context in which the ob- Confirmability
servations are made. As Lincoln and Guba (1985, p. 316) Objectivity is a difficult if not impossible standard to
put it, whether the findings “hold in some other con- live up to in social research. For that reason, the con-
text, or even in the same context at some other time, firmability criterion is designed to ensure that the
is an empirical issue.” Instead of trying to come up researcher has acted in good faith. In other words, it
with findings that can definitely be applied to other should be apparent that personal values or theoretical
times, places, and people, qualitative researchers are inclinations did not blatantly and unduly sway either
encouraged to produce what Geertz (1973) called the conduct of the research or the findings derived from
thick description: rich, detailed accounts of a group’s it. Lincoln and Guba suggest that establishing confirm-
culture or people’s experiences. Lincoln and Guba ability should be one of the objectives of auditors.
argued that thick description provides others with
the database they need in order to assess the possible Overview of the criteria issue
transferability of findings to other milieus. There is a general recognition that quantitative re-
searchers’ notions of reliability and validity cannot be
Dependability applied to qualitative research without some modi-
Lincoln and Guba proposed that to establish depend- fication, although critics vary in their views on how
ability, researchers should adopt an “auditing” ap- much modification is necessary. To bolster their ac-
proach. That is, they should keep complete records of counts, qualitative researchers may use some of the
all phases of the research process—problem formu- strategies advocated by Lincoln and Guba, such as
lation, selection of research participants, fieldwork thick description and respondent validation exercises.
notes, interview transcripts, data analysis decisions,
and so on—and ensure that the records are access- The main goals of
ible. Peers would then act as auditors, possibly during qualitative researchers
the course of the research and certainly at the end, to
make sure that proper procedures have been followed Seeing through the eyes
and to assess the degree to which the study’s inter- of the people being studied
pretations and theoretical inferences can be justified. Many qualitative researchers try to view the social
Auditing has not, however, become a popu- world through the eyes of the people they study.
lar approach to enhancing the dependability of Hiller and DiLuzio (2004) went so far as to analyze
qualitative research. A rare example is a study of be the interview experience itself from the perspective
haviour at an American “swap meet,” where second- of the interviewees. Studying internal migrants in
hand goods are bought and sold (Belk et al., 1988). Canada, they examined what interviewees got out of
9 | The Nature of Qualitative Research 207
the process and found, for example, that they liked • Hallgrimsdottir et al.’s (2006) study of sex-
the fact that it allowed them to meet people who trade workers, which revealed that, contrary to
valued them and validated their life experiences. media depictions, these workers often see their
Qualitative researchers maintain that “(1) . . . face- occupation in a mundane light (for example, as
to-face interaction is the fullest condition of partici- simply a way to make a living); and
pating in the mind of another human being, and • McDonough and Polzer’s (2012) examination of
(2) . . . you must participate in the mind of another the daily struggles of municipal employees fol-
human being (in sociological terms, ‘take the role of lowing the creation of the “mega-city” of Toronto,
the other’) to acquire social knowledge” (Lofland & which showed that rather than creating a leaner,
Lofland, 1995, p. 16). This tendency reveals itself in fre- more efficient workforce, the restructuring led
quent references to empathy. Here are some examples: many manual workers to question their commit-
ment to excellence on the job, while white-collar
• Armstrong (1993) carried out research on Brit- employees felt they had to do extra work just to
ish soccer hooliganism through participant maintain professional standards of service.
observation. He described his work as rooted
in “Verstehende sociology—trying to think On the other hand, sometimes qualitative research
oneself into the situations of the people one is confirms popular notions of a particular group. For
interested in . . . in this case the ‘hooligan.’ This example, Bell (2007) found that media depictions of
approach involves recognizing social phenom- the western Canadian separatist movement as decid-
ena as due not to any single or simply identifi- edly right wing were essentially correct.
able cause and attempting to make sense from The empathetic effort to see the world through the
the [multi-causal] social actors’ viewpoint” eyes of their subjects can present researchers with prac-
(Armstrong, 1993, pp. 5–6). tical problems. For example: How far should one go to
• For their research on teenaged girls and vio- develop empathy? What if the research subject’s ex-
lence, Burman et al. (2001, p. 447) “sought to perience includes participation in illegal or dangerous
ground the study in young women’s experi- activities? There is also the risk that the researcher will
ences of violence, hearing their accounts, and be able to see through the eyes of some of the people in
privileging their subjective views.” a particular social scene but not those of others, such
as people of another gender or culture. These and other
The goal of seeing through the eyes of the people practical difficulties will be addressed in later chapters.
under study is often accompanied by the closely re-
lated goal of probing beneath surface appearances. Description and the emphasis on context
After all, by taking the position of those under study, Qualitative researchers are much more inclined than
researchers may be able to see things in a way that their quantitative colleagues to include descriptive
an outsider with little direct contact would not. This detail when reporting their research. However, they
insight is revealed in: are not exclusively concerned with description. They
also try to explain, and what they endeavour to ex-
• Foster’s (1995) research on crime in a housing plain may pertain to issues of power and exploitation.
complex in East London, which found that the For example, Skeggs (1997, p. 22) tried to answer this
residents do not see crime as a serious problem question: “Why do women who are clearly not just
where they live; victims of some ideological conspiracy, consent to a
• Taylor’s (1993, p. 8) study of female intravenous system of class and gender oppression which appears
drug users, which showed that they are not “pa- to offer few rewards and little benefit?”
thetic, inadequate individuals” but “rational, Many qualitative studies do provide a detailed ac-
active people making decisions based on the count of what goes on in the setting under investiga-
contingencies of both their drug-using careers tion. On the surface, some of this detail may appear
and their roles and status in society”; irrelevant; indeed, there is a risk of becoming too
208 Part III Qualitative Research
embroiled in descriptive detail. Lofland and Lofland unfold over time. Qualitative evidence often conveys
(1995, pp. 164–165), for example, warned against a strong sense of change and flux. As Pettigrew (1997,
what they called “descriptive excess” in qualitative p. 338) put it, process is “a sequence of individual
research, whereby the amount of detail overwhelms and collective events, actions, and activities unfold-
or inhibits the analysis of the data. ing over time in context.” Qualitative research using
But one of the main reasons for providing de- participant observation is particularly well suited to
scriptive detail is that it permits a contextual the study of process. Ethnographers are typically im-
understanding of social behaviour. This implies mersed in a social setting for a long time—in many
that we cannot understand the behaviour of mem- cases, for years. Consequently, they are able to observe
bers of a social group without some knowledge how events develop over time and how the different
of the specific environment in which they oper- elements of a social system (such as values, beliefs,
ate. Behaviour that appears odd or irrational may and behaviour) interconnect. Tracing interdependent
make perfect sense when understood in the context streams of actions and events allows the researcher to
within which it takes place. The emphasis on con- present social life as a process (see Box 9.2).
text in qualitative research goes back to the clas- A sense of process can also be developed through
sic studies in social anthropology, many of which semi-structured and unstructured interviewing in
demonstrated how seemingly illogical practices which participants are asked to reflect on the activities
(e.g., a magical ritual that accompanies the sowing leading up to or following an event. McKee and Bell
of seeds) make sense when they are understood as (1985, p. 388), for example, used a “largely unstruc-
part of the society’s belief system. The provision of tured, conversational interview style” with 45 couples
descriptive detail is also a manifestation of qualita- in which all the husbands were unemployed, to
tive researchers’ desire for naturalism: that is, their show how both husbands and wives adapted to the
belief that researchers should study the social world situation. The various accommodations they made
as it actually is, rather than relying on contrived were not an immediate effect of unemployment but
settings such as formal interviews or experiments. were gradual and incremental responses over time.
The life history approach is another form of qualita-
Emphasis on process tive research that can be used to show process. One of
Qualitative research tends to view social life in terms the best-known studies of this kind is Lewis’s (1961)
of processes. This tendency reveals itself in a number classic study of a poor Mexican family, in which he
of ways, but the main one is the qualitative research- conducted extended taped interviews with family
er’s concern with showing how events and patterns members to reconstruct their life histories.
Karabanow (2002) described his experiences in a period of years provided the basis for an analysis
two Canadian shelters for homeless and runaway of the evolution of their organizational processes.
youth. As a participant observer, he was able to This example shows the development of a sense
monitor routine activities there and describe the of process in at least two ways. First, observation
shelter culture. In addition he carried out in-depth of the shelters over time made it possible to bring
interviews with three levels of shelter workers out developments and interconnections between
and used agency archival materials. The dramatic events. Second, connecting these events with his-
transformations in the shelters’ external environ- torical and other data made it possible to show how
ment and internal operations that took place over the shelters were affected by the larger society.
9 | The Nature of Qualitative Research 209
Along with some colleagues, Bryman undertook As one of the interviewees said, “It’s like going
an evaluation of new staff appraisal schemes in through the motions of it [appraisal]. It’s just get
four universities. The research entailed collecting it over with and signed and dated and filed and
both quantitative and qualitative data, the former that’s the end of it” (Bryman et al., 1994, p. 180).
derived from large numbers of interviews with ap- On the basis of these findings the researchers
praisers, appraisees, senior managers, and many suggested that the attitudes toward appraisal and
others. In the course of conducting the interviews the behaviour of those involved in it were charac-
and analyzing the data, the team became increas- terized by “procedural compliance”—a term that
ingly aware that many of those interviewed ex- the researcher coined and defined as “a response
pressed a certain cynicism about the appraisal to an organizational innovation in which the
project. For some it was a belief that nothing of technical requirements of the innovation . . . are
any significance happens as a result of an ap- broadly adhered to, but where there are substan-
praisal meeting; for others it was a feeling that tial reservations about its efficacy and only partial
the appraisal process itself is not very meaningful. commitment to it” (1994, p. 178).
210 Part III Qualitative Research
collection style of qualitative research can suggest a deep understanding of the people or groups being
alternative avenues of inquiry or ways of thinking. studied. But qualitative researchers’ goals often go
beyond simply understanding. They may also wish
Ultimate goals to do research in order to pursue social justice and
The goals of qualitative research mentioned above— bring about social change, as mentioned in our dis-
seeing through the eyes of others, bringing out a sense cussion of critical approaches to science in Chapter
of process, and having a flexible and unstructured 1. Box 9.4 illustrates how people doing participatory
method of inquiry—are essentially immediate or action research can use qualitative methods to ad-
proximate goals that are used as means of achieving dress a serious social problem. Indeed, qualitative
Type 2 diabetes among Indigenous children and researchers to marginalize Indigenous experi
youth is a growing problem, one with import- ences and ways of knowing. The goal was to make
ant implications for the quality of life among In- the members of the community equal partners in
digenous people in Canada. Sharma et al. (2011) the research, and in so doing allow them to bring
sought to do something about it, but rather than their own issues and perspectives to bear on it.
approach the issue from the perspective of con- The various participants decided collectively
ventional social science, they used the methods of that the project would take the form of a short
participatory action research (PAR) to address it. film to be created by five Indigenous youths aged
Instead of devising the methodology themselves 14–16. The film featured interviews with elders
and then seeking consent from prospective re- suffering from diabetes, a description of what dia-
search participants, they first consulted members betes is, and some tips on healthy eating. The pro-
of the community where the research would take duction of the film involved a constant negotiation
place to get their permission to do a study on the of the relationship between the professional re-
topic, and then asked for their views on what an searchers and the youths. The former tried to stay
appropriate methodology might be. After gaining in the background as much as possible, although
permission to go ahead with the project from the they did end up editing the first version of the film.
band and chief of a rural Indigenous commun- In order to enhance the collaborative nature of the
ity in British Columbia, Sharma et al. held focus project, the professional researchers then showed
groups with several community groups, includ- the edited version to the youths and removed
ing elders with diabetes, the Chief and Council, a any parts that the young people thought were
health-care services team, and some Indigenous inappropriate. This illustrates several aspects of
youths to determine how the study would be con- PAR : explicit recognition of the power differentials
ducted. This was done in an attempt to neutralize that often characterize conventional research; an
the power differential between the researchers attempt to minimize those differentials in order to
(who were non-Indigenous and had PhDs, secure address the subject matter from the perspective
middle-class incomes, and Western cultural out- of the people participating in the research; and
looks) and the members of the community. It was full participation by the group that is the focus
especially important to recognize the potential of the inquiry in order to produce knowledge and
for exploitation in this research, given the history actions that are beneficial to them and not just to
of colonialism in Canada and the tendency of past the professional researchers.
9 | The Nature of Qualitative Research 211
studies have shed light on a wide range of societal that qualitative social scientists experience when
injustices in Canada, including racism (Branker, they revisit ground previously covered by another
2017), homophobia (Abramovich, 2016), and poverty researcher (often referred to as a “restudy”) do not
(Coloma & Pino, 2016). inspire confidence in the replicability of qualitative
research (Bryman, 1994). Qualitative researchers
answer that the replicability criterion is largely irrel-
Critiques of qualitative evant to their research goals because it is not con-
research sistent with their epistemological and ontological
Just as quantitative research has been criticized by positions. For example, they maintain that it is to be
people in the qualitative tradition, the latter has been expected that different researchers will view a par-
taken to task by quantitative social scientists. Some ticular situation or event differently, and then take
of the more common criticisms of qualitative re- the study in different directions.
search are discussed below.
Issues of generalization
Too subjective It is sometimes suggested that the scope of qualita-
Quantitative researchers sometimes criticize quali- tive findings is restricted. Because research based on
tative research for being too impressionistic and participant observation or unstructured interviews
subjective. By this they usually mean that qualitative typically involves only a small number of individuals
findings depend too much on the researchers’ values in a single organization or locality, it may be diffi-
and opinions about what is significant. A related cult to know whether the findings can be applied to
contention stems from the close personal relation- other people or other settings. Can just one or two
ships that many researchers strike up with the people cases be representative of some larger population?
they study, which can lead to bias. Also, because Can Karabanow’s (2002) study of two Canadian
qualitative researchers often begin in a relatively shelters for homeless youth (Box 9.2) be considered
open-ended way and then gradually narrow down generalizable to all youth shelters in the country? In
their research questions or problems, they may not the case of research based on qualitative interviews
explain clearly at the outset why they came to focus rather than participation, are those who were inter-
on one area rather than another. viewed representative of some larger population?
Do the interview findings from sex-trade workers in
Difficult to replicate Victoria (Hallgrimsdottir et al., 2006), for example,
Some quantitative researchers argue that the prob- apply to sex workers in Toronto?
lem of subjectivity is made worse by the difficulty of The answer to these questions in most qualitative
replicating a qualitative study, even though replica- research studies is “probably not.” But a case study is
tion is uncommon in the social sciences, regardless not a sample of one drawn from a known population.
of the research methods used. Because it is unstruc- The people interviewed in qualitative research are
tured and relies on the researcher’s ingenuity, and not necessarily meant to be representative of some
because it may examine processes and situations that larger group. In fact, with some social groups, such
are in flux, qualitative work is almost always impos- as workers in the sex trade, it may be impossible to
sible to replicate. What is observed and heard often enumerate the population in any precise manner,
depends on the researcher, as does the choice of what which makes the selection of a representative sample
to concentrate on; different researchers may choose impracticable.
to observe different things and as a result they may Moreover, in most cases it is not the purpose of
come up with different findings. Also, the behaviour qualitative research to produce generalizable know-
of the people being observed or interviewed is likely ledge. Such research normally stands on its own, pro-
to be affected by the characteristics of the researcher viding in-depth analysis of a person or a small group
(personality, age, gender, and so on). The difficulties of people—something that quantitative research
212 Part III Qualitative Research
rarely does. Also, qualitative research can contribute TABLE 9.1 | Common contrasts between
to the creation of generalizable knowledge insofar quantitative and qualitative research
as it produces theories or concepts that can be as-
Quantitative Qualitative
sessed using other research methods. For instance,
Numbers Words
Williams (2000) argued that qualitative researchers
are often in a position to produce generalizations re- Point of view of Points of view of
researcher research participants
garding the people or events studied (a group of drug
users or soccer hooligans, a strike) that “can be seen Researcher distant Researcher close
to be instances of a broader set of recognisable fea- Theory testing Theory development
tures” (2000, p. 215). In addition, Williams argued Structured Unstructured
that a researcher studying the hooligans who follow a Generalizable knowledge Contextual understanding
certain soccer club will often draw comparisons with Hard, reliable data Rich, deep data
the findings of other researchers who have studied Macro Micro
comparable groups. He or she might also draw com- Behaviour Meaning
parisons with followers of other professional sports Artificial settings Natural settings
teams, or with violent groups in contexts other than
sport. Such generalizations will always be limited
and somewhat more tentative, though no less useful, • “Numbers” versus “Words.” Quantitative re-
than the statistical generalizations produced by searchers set great store by precise, numerical
quantitative research. measurements of social phenomena. They use
those measurements to calculate statistics to
Lack of transparency understand social life, whereas qualitative re-
Another criticism of qualitative research is that it may searchers use mainly words in their analyses
be less than transparent about what the researchers of society.
actually did and how they arrived at their conclusions. • “Point of view of researcher” versus “Points of
For example, qualitative research reports are some- view of participants.” In quantitative research,
times unclear about how the subjects were chosen for investigators are in the driver’s seat; their con-
observation or interview. This contrasts with the ten- cerns structure the investigation. In qualitative
dency of quantitative researchers to include detailed research, the investigation is structured by the
accounts of sampling procedures in their reports. people being investigated: their perspective—
Though, since qualitative research does not claim what they see as important—provides the
generalizability the way quantitative research does, point of orientation.
sampling procedures are less relevant to their work. • “Researcher is distant” versus “Researcher is
close.” In quantitative research, researchers
tend to be less involved with their subjects and
Some contrasts between in some cases, as in research based on online
quantitative and or mailed questionnaires, they may have no
direct contact at all. Quantitative researchers
qualitative research often consider this distance desirable, because
Several writers have used tables to contrast quantitative they want to be as objective as possible and
and qualitative research (for example, Hammersley, feel their objectivity would be compromised
1992). Table 9.1 highlights the main differences. As if they were to become too involved with the
with any summary of a large body of material, the people they study. By contrast, qualitative re-
table outlines general tendencies. The full picture is searchers seek involvement with the people
more nuanced: there are exceptions to these general they study in order to promote a more equali-
points, as we will see in Chapter 14. tarian relationship, which among other things
9 | The Nature of Qualitative Research 213
helps them see the world through the eyes of as “hard”—robust and unambiguous—because
the participants. they strive for precision when devising their
• “Theory and concepts tested in research” versus measurement techniques. Qualitative research-
“Theory and concepts developed from data.” ers, by contrast, claim that their contextual ap-
Before data collection begins, quantitative re- proach and often prolonged involvement in a
searchers typically have a theory in mind that setting mean that the information they collect is
they want to test. In qualitative research, concepts rich, nuanced, and very detailed.
and theories develop as the data are collected. • “Macro” versus “Micro.” Quantitative research-
• “Structured” versus “Unstructured.” Quantita- ers often seek to uncover large-scale social
tive research is typically highly structured to trends and connections between variables; as we
maximize validity and reliability; in qualita- noted above, their research is generally designed
tive research, the approach is less structured in to be applicable to large populations. Qualitative
order to allow the researcher to get a sense of researchers are usually concerned with in-depth
the meanings people derive from their every- analysis of small-scale aspects of social reality,
day lives, and to permit the development and such as personal interaction, although these may
elaboration of concepts and theories as the be linked to larger social and political issues.
data are collected. • “Behaviour” versus “Meaning.” It is sometimes
• “Generalizable knowledge” versus “Contextual suggested that quantitative researchers focus
understanding.” Whereas quantitative research- more on people’s behaviour, and qualitative
ers want their findings to be applicable to some researchers more on the meaning of that be-
larger population, qualitative researchers focus haviour for the actor.
on understanding the behaviour, values, and • “Artificial settings” versus “Natural settings.”
beliefs of the people in their study. Qualitative Whereas quantitative research is often con-
findings usually develop out of the context these ducted in a contrived context, such as an experi-
people find themselves in; hence the insights ment or a formal interview setting, qualitative
gained may or may not apply to other people. researchers investigate people in their natural
• “Hard, reliable data” versus “Rich, deep data.” environments, such as social gatherings or
Quantitative researchers often describe their data workplaces.
Key Points
• Qualitative research does not lend itself to a clear four conditions: credibility, transferability, depen
set of linear steps. As a research strategy it tends dability, and confirmability.
to be more open-ended than is typically the case • Most qualitative researchers try to see the
with quantitative research. social world through the eyes of their research
• In qualitative research, theories and concepts are participants.
outcomes of the research process. • Like all approaches to social research, qualitative
• There is considerable unease about the direct ap- methods have both strengths and weaknesses. Their
plication of reliability and validity criteria to quali- major strengths include the development of empa-
tative research. Many writers prefer alternative thetic understanding, rich description of context,
criteria. and emphasis on process. Weaknesses include a
• Lincoln and Guba (1985; 1994) propose two key cri- reliance on the researcher’s own perspectives when
teria to evaluate a qualitative study: trustworthi- making research decisions, difficulties in replication,
ness and authenticity. Trustworthiness requires limited generalization, and lack of transparency.
214 PART III Qualitative Research
Relevant Websites
The International Institute for Qualitative Meth- Examples of qualitative studies can be found in the
odology, based at the University of Alberta in journal Qualitative Research and in the Qualitative
Edmonton, sponsors conferences and res earch Sociology Review; you may have to access these
training programs involving qualitative re- journals through your institution’s library server.
search methods. It also publishes the International http://qrj.sagepub.com
Journal of Qualitative Methods, which is available
www.qualitativesociologyreview.org/ENG/index_
free online.
eng.php
www.iiqm.ualberta.ca
(Websites accessed 25 October 2018)
Chapter Overview
Ethnography and participant observation require extended involvement in the activities of the people
under study. This chapter explores:
• the problems of gaining access to different settings and ways of overcoming them;
• whether covert research is practicable and acceptable;
• the role of key informants;
• the different roles that ethnographers can assume in the course of their fieldwork;
• the function of field notes and the forms they can take;
• the role of visual materials in ethnography;
• bringing an ethnographic study to an end; and
• the issue of feminist ethnography.
Do you like to travel to places you’ve never been to Ethnography and participant observation involve
before? Have you ever observed a group of people placing yourself in a social environment that may be
you don’t know very well and wondered what it would foreign to you, and staying there for an extended period
be like to be a member of their group? Have you ever of time. What kinds of groups or social settings intrigue
witnessed profound human suffering and asked your- you? Non-governmental organizations? Political move-
self how things ended up that way and how the people ments? Sports teams? Criminal gangs? Hospital emer-
suffering managed to endure? Would you like to give gency rooms? Women’s shelters? All of these can be
such people a voice or expose the hardships that they subjects of ethnographic and participant observation
face? If so, doing or at least reading about ethnography research.
and participant observation should interest you.
▲ Photo by Matt McClain for The Washington Post via Getty Images
10 | Ethnography and Participant Observation 217
Overt versus covert ethnography medical school that although he was “an ‘open’
observer with regard to doctors and students,” he
One way to ease the access problem is to assume a was “a ‘disguised’ observer with regard to the pa-
covert role: in other words, to not disclose that you tients.” Also, some ethnographers move between
are a researcher. Getting access can be very prob- the two roles (see Box 10.1).
lematic, and the adoption of a covert role removes Another interesting case is provided by Glucks-
some of the difficulties. These two distinctions— mann (1994), who in the 1970s left her academic post
open versus closed setting, and overt versus covert for a job on a factory assembly line in an effort to
role—suggest a fourfold distinction in forms of eth- find out why feminism appeared not to be relevant
nography (see Figure 10.1); an example of Canadian to working-class women. In a sense, she was a covert
research is given for each of the four types. observer, but her motives for the research were pri-
Three points should be noted about Figure 10.1. marily political, not academic. At the time she was
First, the “open versus closed setting” distinction is undertaking the research she had no intention of
not hard and fast. Sometimes the process of gain- writing the book, which was eventually published
ing access in “open” settings can take on an almost under a pseudonym. Was she an overt or a covert ob-
formal quality; for instance, if you are talking to a server (or neither, or both)? Whichever description
gang leader in the street, you may have to answer applies, this is an interesting case of what might be
some pointed questions about your goals. Also, termed “retrospective ethnography”: using obser-
“closed” organizations and social movements some- vations that were gathered before the decision was
times create contexts that have a public character, made to conduct a study.
such as meetings arranged for members or pro- A third point with regard to Figure 10.1 is that
spective recruits. Bell (2007) attended many western the preferred choice is an overt role. There are sev-
Canadian separatist meetings and rallies that were eral reasons for this. As Box 10.2 reveals, these rea-
open to the public, and was able to get a lot of infor- sons have to do with practical and especially ethical
mation on the movement and its leaders that way. considerations. The ethics of covert research, which
Second, the “overt versus covert” distinction were addressed in depth in Chapter 3, involve issues
can vary from context to context even within the of deception, lack of informed consent, and invasion
same research project. Although ethnographers of privacy, since the researcher has essentially fooled
may seek access through an overt route, many the participants into believing that they are not being
of the people with whom they come into con- observed for research purposes. Because of the eth-
tact may be unaware of their status as research- ical problems (and some of the practical difficulties)
ers. A
tkinson (1981, p. 135) noted in connection that beset covert research, the bulk of the following
with his research on the training of doctors in a discussion of access issues focuses on overt research.
Access to closed settings based, it seemed like the perfect place to start.
McDonald’s had other ideas, however, and
Gaining access to most organizations requires stra- only after tenacious pestering and persuasion
tegic planning, hard work, and sometimes luck. In se- did I overcome corporate employees’ polite
lecting a case study for an ethnographic investigation, demurrals, couched in terms of protecting
the researcher may employ several criteria, depending proprietary information and the company’s
on the research area of interest. One may choose a image. (Leidner, 1993, pp. 234–235)
certain case or setting because of its “fit” with the re-
search questions, but there is no guarantee of access. This kind of determination is necessary when-
Sometimes sheer perseverance pays off. Leidner ever the target is a specific organization or group
(1993) was determined that McDonald’s would be one (e.g., a particular religious sect or social movement).
of the organizations in which she conducted ethno- Rejection can mean having to find a completely
graphic research on the routinization of service work: different research topic.
On the other hand, for many research questions
The company was a pioneer and exemplar of you are likely to find that several potential cases
routinized interaction, and since it was locally will be suitable. Organizational researchers have
220 PART III Qualitative Research
developed a range of tactics, some of which may seem you would not consider acceptable). However,
rather unsystematic but are still worth considering: this strategy carries a risk of making the re-
searcher a cheap consultant. Some writers on
• Use friends, contacts, and colleagues to help research methodology do not recommend
gain access; as long as the organization is offering something in return, although it is
relevant to the research question, the route commonplace among researchers on formal
should not matter. organizations.
• If possible, get someone in the organization to • Provide a clear explanation of your aims and
vouch for you and the value of your research. methods. Suggest a meeting to deal with the
Such people are considered “sponsors.” If per- participants’ concerns and explain what will
mission is granted at a lower level of the hier- happen in terms that people who are not fam-
archy, you may still need clearance from top iliar with social research can understand.
management or senior executives. People who • Be prepared to negotiate; almost no one gets
must grant their permission for you to gain complete access.
access are sometimes called gatekeepers. • Be frank about the amount of people’s time
• Offer something in return: for example, a you are likely to need. Time is an important
final report (if you provide a working draft, issue for commercial organizations, and for
people may ask you to make changes that many non-profit groups too.
10 | Ethnography and Participant Observation 221
Giulianotti (1995) sought access to two groups of out a sponsor who could ease his entry into the
soccer hooligans. Access to one was reasonably group. After some abortive attempts, he was
smooth in that he was a close friend of three of the finally introduced to someone at a game and
47 men caught by the police engaging in unruly this contact gave him access to more supporters.
behaviour at a soccer game. In addition he had Eventually, he was able to negotiate access to
known and socialized with many others at school, the group as a whole by striking a “research bar-
and in terms of “age, attire, and argot” he was sim- gain”: he would provide details of what rival fans
ilar to the people he was studying. Gradually his thought of them. Giulianotti described his overall
contacts multiplied and eventually he “began so- research strategy as:
cializing freely with the gang at [soccer] matches,
regularly introducing myself to new re-
travelling to and from matches within the main
search acquaintances; renegotiating asso-
grouping” (1995, p. 4).
ciation with familiar casuals; talking with
Access to the other group of supporters was
them, drinking with them, and going to
much more difficult for three reasons: absence
matches with them; generally participating
of prior acquaintanceships; his background and
with them in a variety of social situations;
accent; and a high level of negative newspaper
but disengaging myself from . . . participat-
publicity about them at the time he was seek-
ing in violence, within and outside of [soccer]
ing access, which made the hooligans wary of
match contexts. (Giulianotti, 1995, p. 3)
people writing about them. Giulianotti sought
222 PART III Qualitative Research
expressed an interest in hanging out with them. But • If the people being studied have concerns or
he tried to move too quickly in seeking information suspicions about the researcher, they may
about them and was forced to abandon his plans. appear to go along with the research but in
Eventually, Wolf’s hanging-around strategy led to fact sabotage it, engage in deception, provide
his being approached by the leader of a biker group misinformation, or not allow access to “back
(Rebels MC), who acted as his sponsor. To bring this regions” (Goffman, 1956).
off Wolf ensured that he was properly attired in biker
garb, although this was not difficult for him as he There are three ways to smooth the path for on-
had been a motorcyclist and had absorbed biker cul- going access to closed settings:
ture long before he started his research. Attention to
dress and demeanour can be a very important con- • Play up your credentials: talk about your past
sideration when seeking access to either public or work, your experience, your knowledge of the
closed settings. group, and your understanding of its problems.
As these anecdotes suggest, gaining access to • Don’t give people a reason to dislike you: be
social settings is a crucial first step in ethnographic non-judgmental when hearing about informal
research. But it is also difficult and in some cases activities or about the organization; make sure
dangerous—for example, when the group in ques- that information given to you doesn’t get back
tion is engaged in violent or criminal activities. Cau- to others, whether bosses or peers.
tion is essential when choosing an access strategy. • Help out occasionally with work, or offer
advice without being critical.
Ongoing access
Access problems do not end after initial contact and Similar considerations apply to research in public
entrée to the group. Maintaining access is an on- settings:
going activity and is likely to prove a persistent prob-
lem in closed contexts such as organizations. • Make sure you have a plan for allaying people’s
suspicions. Giulianotti (see Box 10.3) simply
• People will be suspicious, perhaps seeing the said that he was doing research on soccer fans
researcher as an instrument of top manage- for a book.
ment (it’s very common for members of or- • Be prepared for tests of either competence
ganizations to believe that researchers have or credibility. When doing research on gang
been put in place to check up on them). When members in a poor community, Horowitz
Sharpe (2000, p. 366) began research on sex wrote that she was frequently told “confiden-
work in a red-light area, she was thought to be tial” stories (which turned out to be fictional)
“anything from a social worker to a newspaper to see if she could keep a secret (Gerson &
reporter with hidden cameras and micro- Horowitz, 2002).
phones.” When conducting her research on the • Be prepared for changes in circumstances. Both
British on the Costa del Sol, O’Reilly (2000) Giulianotti (Box 10.3) and Armstrong (1993)
was suspected of being a tax inspector. found that newspaper exposés of the kinds of
• Group members will worry that what they say activities they were studying led to worries that
or do may get back to bosses or colleagues. Van they were not what they said they were.
Maanen (1991) noted that when conducting
ethnographic research among police officers, Key informants
it’s not unusual to observe activities that reflect Ethnographers rely heavily on informants, especially
badly on the police and may even be illegal. In those who develop an understanding of the research
such situations the researcher can establish and are able to identify situations, events, or people
credibility by using discretion. likely to be helpful to the investigation. Whyte (1955)
10 | Ethnography and Participant Observation 223
again provides an extreme example, quoting a key welcome a beer and a chat about “It,” or tell me
informant named Doc: who I “ought to ’ave a word wi.’” (Armstrong
1993, pp. 24–25)
You tell me what you want to see, and we’ll
arrange it. When you want some information, Unsolicited offers of information are highly at-
I’ll ask for it, and you listen. When you want tractive to ethnographers because of their apparent
to find out their philosophy of life, I’ll start an spontaneity. However, as Hammersley and Atkinson
argument and get it for you. If there’s some- (1995, pp. 130–131) observed, it’s important to rec-
thing else you want to get, I’ll stage an act for ognize that such offers may on occasion be staged for
you. (1955, p. 292) the ethnographer’s benefit.
Solicited accounts can be obtained in two ways: by
Doc was also helpful in warning Whyte that he interviewing (see Chapter 11) or by casual question-
was asking too many questions, telling him to “go ing during conversations (though in ethnographic
easy on that ‘who,’ ‘what,’ ‘why,’ ‘when,’ ‘where’ stuff” research the boundary between an interview and a
(1955, p. 303). Taylor (1993) said that in her partici- conversation is by no means clear). When the eth-
pant observation of 50 female drug users, she carried nographer needs specific information on an issue
out intensive interviews with 26 women, 8 of whom that does not lend itself to direct observation or that is
were key informants. not cropping up during “natural” conversations, so-
Key informants can also provide support that licited accounts are likely to be the only way forward.
helps with the stress of fieldwork. However, undue
reliance on them can lead researchers to see social
reality through their eyes only rather than viewing
Roles for ethnographers
things from the perspectives of several group mem- Related to the issue of ongoing access is the question
bers or the group as a whole. of the role an ethnographer adopts in relation to the
In addition, ethnographers encounter many social setting and its members. Often the various
other people who will act as informants whose ac- roles in fieldwork are arrayed along a continuum
counts may or may not have been solicited. Some from complete involvement to complete detachment
researchers prefer unsolicited information because (see Figure 10.2). It should also be noted that the full
of its spontaneity and naturalism. Very often, re- gamut of roles may be employed at different times in
search participants develop a sense of the events or the course of a single ethnographic project, and for
encounters that the ethnographer will want to see. different purposes.
Armstrong (1993), in the course of his research on For example, the participant observer can be any
soccer hooligans, would sometimes get tips from a of the following:
group of Sheffield United fans who called themselves
“Blades”: • Complete participant. The ethnographer is a
fully functioning member of a social setting
I often travelled on the same coach as Ray but one whose true identity is unknown to
[an informant]; he would then sit with me at members: in other words, a covert observer.
matches and in pubs . . . giving me background
information. Sometimes he would start con-
versations with Blades about incidents which Involvement Detachment
he knew I wanted to know about and after-
wards would ask “Did you get all that down Complete Participant-as- Observer-as- Complete
then?” . . . There was never one particular in- participant observer participant observer
formant; rather, there were many Blades . . . FIGURE 10.2 Classification of participant observer
who were part of the core and would always roles (Gold 1958)
224 PART III Qualitative Research
In such cases the ethnographer is engaged in role carries the risk of overidentification with the
regular interaction with people and partici- group being studied (see Boxes 10.5 and 10.6), but
pates in their daily lives but assumes the re- at the same time it offers an opportunity to get close
searcher’s role in private to write down notes to people. The observer-as-participant role carries
once the situation has unfolded. This is what the risk of not understanding the social setting
Humphreys (1970) did when he studied male and its people sufficiently and therefore making in-
homosexual encounters in public washrooms. correct inferences. In an interesting variant of this
• Participant-as-observer. The ethnographer par- problem, Hessler et al. (2003) cautioned observers
ticipates as described above, except that members (in a study of adolescent risk behaviour) against the
of the social setting are aware that the ethnog- temptation to act as counsellors to respondents.
rapher is studying them, such as in G iulianotti’s Gans (1968) would add that even if it were pos-
research (see Box 10.3). sible, it is probably undesirable to adhere to a single
• Observer-as-participant. The researcher is role over the entire course of a project, since this
mainly an interviewer and observer, and par- would limit the ethnographer’s flexibility in hand-
ticipates only marginally in the group’s activ- ling situations and people. In addition, the risks of
ities. This was Bell’s (2007) role in his study of either excessive involvement or excessive detach-
western Canadian separatists. Ethnographic ment would loom large.
research on the police is often similar, since
the opportunities for genuine participation Active or passive?
are limited by legal and safety restrictions. How active or passive should the ethnographer be?
Norris (1993) described how as an observer- In most studies—even when the researcher is mainly
as-participant, he concentrated on gathering an observer rather than a participant—involvement
two types of data: “naturally occurring inter- in the group’s activities will be unavoidable from
officer talk” and “detailed descriptions of how time to time. For example, Fine’s (1996) research
officers handled ‘live’ incidents” (1993, p. 126). on the work of chefs in restaurants was carried out
For other examples of situations in which largely through semi-structured interviews. In spite
minimal participation may be appropriate, see of this limited participation, he sometimes found
Boxes 10.4 and 10.6. himself in the kitchen washing dishes to help out
• Complete observer. The final possibility is no during busy periods. Similarly, in a study of boun-
interaction with the people observed. Most cers, the participant observer will not have the
writers do not include this as a form of eth- luxury of deciding whether to become involved in
nography, since by definition there is little fights, since involvement comes with the territory
or no involvement or participation. There (Winlow et al., 2001).
may be less risk of reactivity (subjects be- Sometimes ethnographers feel they have no
having unnaturally because they know they choice about getting involved because a failure to
are being observed) because the researcher participate might suggest a lack of commitment
is at a distance; but there is also greatly re- to the group. In such situations, standing back could
duced potential for understanding because lead to a loss of credibility in the eyes of the people
the researcher does not ask questions or try studied. This situation can arise especially when
in any way to get into the heads of the people the activities are illegal or dangerous. On the other
under study. hand, many writers counsel against active partici-
pation in criminal or dangerous activities. Both
Each approach has advantages and disadvantages. Armstrong (1993) and Giulianotti (1995; see Box 10.3)
In the case of covert ethnography, some of these are refused to fight while doing their research into
discussed in Box 10.2. The participant-as-observer soccer hooliganism. The latter wrote: “My own
10 | Ethnography and Participant Observation 225
Hallett and Barber (2014) argue that the study of touch with him. The student didn’t have a phone,
online spaces should be included in e thnographic and email messages often went unanswered. The
research because people’s “natural habitat” researcher soon learned that the most depend-
now includes their “online habitat.” Facebook, able means of communication was Facebook,
Twitter, email, instant messaging, Skype, You- which the student checked many times a day.
Tube, blogs, and so forth are now predominant Facebook also proved to be an invaluable source
places for participating in social networks and of ethnographic information about the organiza-
acquiring information and interpretations of the tion, how it established a sense of community, and
world. Despite that fact, most ethnography is how its members viewed the plight of undocu-
conducted offline; the authors themselves only mented students. The author states that he “had
ventured into cyberspace after their attempts considered asking participants to journal about
to use solely in-person methods proved to be how they spent time and perceived the educa-
unworkable. tional process. I soon realized that students were
In researching an organization for undocu- already ‘journaling’ on their Facebook pages, and
mented college students in California, one of the these posts were more natural than I may have re-
authors began by researching the life history of a ceived if they had used paper and pencil to write
particular student, but found it difficult to keep in at my request” (316).
rules are that I will not get involved in fighting or criminals or those involved in dangerous actions:
become a go-between for the two gangs in organ- it is much more difficult for someone in such a
izing fights” (1995, p. 10). That is an illustration of role to decline to participate in group activities
why it is not a good idea to do covert research on (see Box 10.5).
226 PART III Qualitative Research
Sometimes, as a result of prolonged immersion in A related issue for Hobbs was involvement in
the lives of the people being observed, coupled illegal activity, which may be tempting if one be-
with a commitment to seeing through their eyes, comes overly acculturated to a group in which
ethnographers become so wrapped up in the criminal actions are known to occur:
world of the people they are studying that they
A refusal, or worse still an inquiry concerning
forget their role as researchers. Under such cir-
the legal status of the “parcel,” would provoke
cumstances the ethnographer may find it diffi-
an abrupt conclusion to the relationship. Con-
cult to maintain a social scientific perspective on
sequently, I was willing to skirt the bound-
the collection and analysis of data. For example,
aries of criminality on several occasions, and
when Schulenberg (2014) conducted participant
I considered it crucial to be willingly involved
observation of Canadian police officers while
in “normal” business transactions, legal or
doing “ride alongs” with them, she had to contend
otherwise. I was pursuing an interactive, in-
with a feeling that she may have become overly
ductive study of an entrepreneurial culture,
accepting of the officers’ beliefs and values. Sim-
and in order to do so I had to display entre-
ilarly, Hobbs (1988, p. 6) wrote of his fieldwork
preneurial skills myself. . . . [My] status as an
that he “often had to remind himself that [he] was
insider meant that I was afforded a great deal
not in a pub to enjoy [himself] but to conduct an
of trust by my informants, and I was allowed
academic inquiry, and repeatedly woke up the
access to settings, detailed conversations,
following morning with an incredible hangover,
and information that might not otherwise
facing the dilemma of whether to bring it up or
have been available. (1988, pp. 7, 15)
write it up.”
Overidentification with the people one is ob- Nonetheless, she felt she “was gathering good
serving is not the only risk that participant ob- data, despite [her] feelings of being an outsider”
servers face in relation to the social situations (2000, p. 120). Similarly, although Bell (2007)
they study. Lee-Treweek (2000) carried out re- spent more than a year observing western
search on auxiliary caregivers in two homes Canadian separatists, at no point was he ever
for the elderly. In one of them she reacted in a tempted to join the group. The lesson here is
way that was the opposite of overidentification. that overidentification is not inevitable, and
She disliked the home and found the staff un- that researchers must be careful not to let either
appealing because of their uncaring approach positive or negative feelings for the people they
and lack of sympathy for the elderly residents. study distort their findings.
10 | Ethnography and Participant Observation 227
Field notes small amounts of time away from the group to write
down your observations. In that case you must take
Because of the frailties of human memory, ethnog- care not to make the participants anxious or suspi-
raphers must jot down their observations. These cious (see Box 10.1).
should be fairly detailed summaries of events and To some extent, strategies for taking field notes
behaviour, and the researcher’s initial reflections on depend on how clearly delineated the research ques-
them. The notes need to specify the key dimensions tions are. As we noted in Chapter 9, most qualitative
of whatever is observed or heard. Here are some gen- research begins with general research questions,
eral principles: but there is considerable variation in their speci-
ficity. Obviously, when there is a clearly focused
• Write down notes, however brief, as soon as research question, the ethnographer’s observations
possible after seeing or hearing something should be oriented to it. At the same time, it’s im-
interesting. portant to maintain a fairly open mind, so that
• Write up full field notes at the end of the day the study’s flexibility with respect to the themes it
at the very latest, including details such as explores—a strength of qualitative r esearch—is not
location, the people involved, what prompted impaired.
the exchange or event, and date and time In the context of her research on female drug
of day. users, Taylor (1993, p. 15) explained that in her
• You may also use an audio recorder to take early days in the field she tended to listen rather
notes from later, but you will have to allow than talk because she “did not know what ques-
plenty of time for transcription. tions [she] wanted to ask.” Armstrong (1993,
• Notes must be vivid, clear, and complete. If in p. 12) wrote in connection with his research on
doubt, write it down. soccer hooliganism that his research “began with-
out a focus” and that as a result “he decided to
It’s obviously best to take notes as soon as some- record everything.” Therefore a typical Saturday
thing significant happens. However, wandering would mean 30 pages of handwritten notes. Such
around with a notebook and continually scribbling open-endedness usually cannot last long, because
in it can make the people you are observing feel trying to record the details of absolutely every-
self-conscious. Thus it may be necessary to take thing is so tiring. Eventually the ethnographer
begins to narrow down the focus of the research, • Full field notes. These are the main data source
sometimes by relating the emerging findings to in ethnographic research. Full notes should
the social scientific literature on the topic. This be written as soon as possible, usually at the
approach is implied by the sequence suggested in end of the day. Provide as much detail as pos-
Figure 9.1. sible about events, people, conversations, and
For most ethnographers, the equipment needed so on. A particular comment from a subject
for observation consists mainly of a notepad or an account of an event that you witnessed
and pen (for example, Armstrong, 1993, p. 28). may not seem very important at the time, but
A n audio recorder may be useful but, as we have its significance may become apparent later on,
noted, transcribing what you record will take time, once you have developed an interpretive struc-
and care must be taken to ensure the audio record- ture. Without good notes, valuable data may
ings are not disseminated, as they may uninten- be lost. Write down, in brackets, initial ideas
tionally identify research participants. In addition, about your interpretations, impressions, and
recording may be more obtrusive than writing feelings.
notes. A lthough participants usually grow accus-
tomed to the presence of the researcher over time, It’s worth adding that in field notes the ethnog-
speaking into a recording device can rekindle fears rapher’s presence is frequently evident. This can be
about participating; and in some gatherings it may seen in Whyte’s account (above) of almost being
be difficult to use one because of extraneous noise. thrown down the stairs. However, in the finished
Photography can be an additional source of data, work the ethnographer is sometimes written out of
and pictures do help to stir the ethnographer’s the picture. Field notes, except for brief passages, are
memory, but in some kinds of research (especially primarily for the researcher’s own use (Coffey, 1999),
involving crime and deviance) photography may whereas the written report is for public consumption
not be feasible. and has to be presented as a credible account of the
social setting and culture in question. References to
Types of field notes the ethnographer should be limited; otherwise the
Some writers have found it useful to classify the types report will look more like an account of his or her
of field notes generated in the process of conducting activities than those of the group purportedly under
ethnography. The following classification is based on study. This issue will be addressed in further detail
categories suggested by Lofland and Lofland (1995) in Chapter 15.
and Sanjek (1990):
Analytic memos
• Mental notes. These are particularly useful Most of the previous discussion focused on the
when it is inappropriate to be seen taking writ- mechanics of note taking. But content is even more
ten notes; however, they should be put into important. Notes can be taken on the setting, the
writing at the earliest opportunity. people, the methodological problems encountered,
• Jotted notes (also called scratch or rough notes). and so on. In addition it’s possible to create analytic
The researcher makes very brief notes to jog memos recording some initial thoughts on what all
the memory about events that should be writ- those observations might mean. Analytic notes help
ten up more fully later. Lofland and Lofland to bridge the gap between the data and the concepts,
(1995, p. 90) described jotted notes as “little interpretations, and theories that researchers de-
phrases, quotes, key words, and the like.” velop to make sense of what they are observing. Such
Such notes need to be written inconspicu- memos should be dated and reviewed regularly. They
ously, preferably out of sight, since writing should also be kept separate from notes on actual ob-
in front of research participants may be off- servations because they are comments on data, not
putting for them. data themselves.
10 | Ethnography and Participant Observation 229
The rise of visual ethnography Qualitative researchers have used visual materials
in at least three ways:
The use of visual materials in social research is by
no means new. For many years cultural anthropol- • as memory aids in the course of fieldwork,
ogists have used photographs in their portrayals of where they essentially become components of
traditional societies. However, books such as Pink’s the ethnographer’s field notes
(2001) Visual Ethnography suggest that the use of • as sources of data in their own right
photos in social research has entered a new phase. • as prompts for discussion by research participants
In Chapter 12 we will discuss visual materials
(primarily photographs) that were produced some Pink (2001) drew an important distinction be-
time before the research began. Here we will con- tween two positions on visual materials. The trad-
sider visual materials that are produced more or less itional framework is a realist one in which the
exclusively for research purposes: research-driven material simply captures an event or setting that
images, mainly photographs but also video record- then becomes a “fact” for the ethnographer to in-
ings and other visual media. It’s worth noting that terpret along with other data. The image and what
although the term “visual ethnography” is increas- it represents are presented as unproblematic, as
ingly popular, it does not always refer to the kind of a window on reality. This has been the dominant
sustained immersion in a social setting that is the framework within which visual resources have
subject of this chapter. been produced and analyzed. By contrast, Pink also
230 PART III Qualitative Research
drew attention to a position that she described as bullfighter, it appealed to many connoisseurs of
reflexive. This entails an awareness of and sensi- the sport who saw in it various artistic and other
tivity to the ways in which researchers themselves meanings. For Pink, it held additional significance
determine what the visuals reveal. This sensitivity in terms of her interests in gender and the broader
requires a grasp of the ways in which researchers’ discipline of social anthropology (for more infor-
personal characteristics—age, gender, social back- mation on this photograph see Pink, 2001, p. 101,
ground, academic proclivities, and so on—may in- and http://www.brooklynsoc.org/revealingpictures/
fluence what images they choose to record, and how pink.html [accessed 26 October 2018]).
those images are presented. Photovoice is an approach to visual ethnography
This approach to the visual is frequently collab- that takes collaboration with research partici-pants
orative in the sense that research participants may even further. Using this technique, the participants
be involved in decisions about what visuals should be take photographs of their daily experiences and in-
recorded and how they should be interpreted. Fur- terpret them in terms of their own perspectives and
ther, it acknowledges that different people may inter- identities, which transfers a measure of power and
pret the images in different ways. In Pink’s research control away from the researcher to them. Since
on Spanish bullfighters, enthusiasts interpreted the photovoice may be used with people from margin-
images she took of bullfights in terms of bullfighter alized communities, those who employ it often do so
performance. Other viewers might see the same with the intention of creating social change, which
images in a different interpretive framework—for can be undertaken through an exhibition of the
example, one centred on animal rights and cruelty. photographs. When used in this way, photovoice is
Plate 10.1 presents an image called The Bullfight- a form of participatory action research (discussed
er’s Braid, from Pink’s research. Showing a female in Chapter 9).
Holtby et al. (2015) used photovoice in their re-
search on queer and trans youth in the Kitchener–
Waterloo area. The researchers used the technique
to explore how these young people confronted the
“hostile gaze” that is often directed at people in the
LGBTQ community. The 15 people in the study were
given training in photography and then asked to
take five to ten pictures that reflected their experi-
ences as queer or trans youth. The participants were
then invited to discuss their photos by responding
to questions such as, “What is seen here?” and
“How could this image be used to educate people?”
One of the themes to come out of the discussion
was invisibility, whereby participants were not rec-
ognized as having the gender or sexual identity that
they held. This experience was particularly painful
when it occurred within the LGBTQ community
itself. To illustrate invisibility, one participant took
Pink, Visual Ethnography, 2001
Radley and Taylor (2003a; 2003b) were interested month or so later at home. On each occasion, pa-
in the role that the physical setting of a hospital tients were asked about all the photographs and
ward plays in patient recovery. Nine patients in a which ones best reflected their experience in the
ward were asked to take photographs on the ward hospital.
a few days after their surgery. Each patient was Most of the images taken appear to be very
supplied with a camera and asked to take up to mundane, neither striking nor interesting. How-
12 photographs of things in the hospital that they ever, the photos took on considerable significance
found personally significant. The only constraint in the interviewees’ narratives on the positive and
was not to include people in their photographs negative aspects of their time in hospital. For pa-
(because of hospital restrictions). The research- tients, the pictures became part of the stories they
ers stayed with the patients while they took their offered about how they made the transition from
photographs. Patients were interviewed a day the hospital setting to their homes, which can be
after the photos were developed, and again a an important aspect of recovery.
in which they were generated; the potential for how they affect the personal experiences of the in-
multiple meanings among researchers, study par- dividuals involved. The approach can be explicitly
ticipants, and others; and the potential for the re- change oriented in that previously unrecognized
searcher to influence the image and its presentation. opportunities to transform institutions may come
In addition, researchers usually include non-visual to light as the research runs its course, and the re-
research methods, such as interviews, in their in- searcher may even collaborate with the people stud-
vestigations. This raises the question of the relative ied in an effort to create institutional relationships
significance of words versus images in the analy- that serve them better.
sis of data and the presentation of findings. Since For example, Restoule and colleagues (2013)
words are the traditional medium, it’s easy to slip combined Anishinaabe methodology (which
into seeing the visual as less important, but this is honours Indigenous elders’ wisdom and places
not always justified. Finally, visual research meth- a strong emphasis on representing and serving
ods raise certain ethical issues, such as invasion of the people involved) with institutional ethnog-
privacy and anonymity. raphy to examine how well Indigenous students in
Ontario made the transition from high school to
post-secondary institutions. They first established
Institutional ethnography that although the vast majority of young Indigen-
Canadian sociologist Dorothy Smith has developed ous people hoped for a post-secondary education,
an approach called institutional ethnography and that their parents were in favour of it as well,
(Smith, 2005) that explores how institutional dis- a much smaller proportion of Indigenous youth
courses (typically texts found in the workplace) relate than non-Indigenous youth actually graduated
to people’s everyday experiences with institutions, from those institutions. The researchers found that
and how examining institutional relationships may some informants declined to identify as Indigen-
reveal larger systems of social control and power in ous in their applications, citing a lack of trust in
a society (Devault, 2006). Institutional relationships government institutions, which itself indicated
are analyzed in detail, with the aim of finding out a problematic relationship between prospective
232 PART III Qualitative Research
Campbell (2000) did institutional ethnographic such a joke” (Campbell, 2000, p. 138). The lack of
research to investigate how people with disabil- assistance put the woman’s relationship with her
ities in Victoria, British Columbia, interacted with partner in jeopardy because he had to devote
local health-care providers. She was particularly what little time they could spend together to
interested in how that interaction was influenced doing heavy housework for her.
by government regulations and policies. The In another case, a woman confined to a wheel-
health-care system in the province was in a state chair suffered because she qualified for only 12
of transition at the time and had been subject to physiotherapy and massage sessions a year—the
funding reductions associated with government same number as a person without a disability. The
deficits and debt. As is typical in this kind of re- institutional justification given was that all cit-
search, Campbell’s goal was not only to describe izens had to be treated equally. Campbell makes
how the situation was perceived by the people the point that “generic” rules about health care,
experiencing it, but also to give them—in this which essentially establish “relations of ruling” in
case both people with disabilities and health-care the health-care system, are created by and for the
providers—the means to transform the situation able-bodied and thus do not serve the needs of
themselves. Both groups were trained as research- people with disabilities (Campbell, 2000, pp. 141,
ers and paid for their involvement. 143). The texts used to implement government
The study focused on Home Support services health policies (such as assessment forms), Camp-
and especially how access to those services af- bell argues, serve to bolster the government’s
fected the lives of disabled people. One person legitimacy in the face of challenges from dissent-
mentioned that because of government criteria, ing parties such as the disabled, who would bene-
she was unable to get the support she needed: “I’m fit from a more inclusive system of health-care
dog tired. And I just get the basics done. My partner delivery. Bringing such matters to light can be a
is away. I’ve been on my own since August and it’s first step in improving service provision for people
been hard because my home care was cut off two with disabilities or anyone else who is ill served by
years ago because I’m not sick enough—which is an institution.
10 | Ethnography and Participant Observation 233
setting, but over a long period of time they can forget, and thoughts, or interprets them from the stand-
especially if there has been genuine participation in point of men in the society or of the male re-
activities within that setting; therefore farewells may searcher” (Reinharz, 1992, p. 52) is rejected; and
have to be arranged. It’s also essential to remember • it understands women in context.
the ethnographer’s ethical commitments, such as
the need to ensure that persons and settings are kept Similarly, Skeggs (2001, p. 430) observed that many
anonymous—unless, of course, there has been an feminist researchers have considered ethnography
agreement that the social setting can be disclosed (as “with its emphasis on experiences and the words,
often occurs in studies of religious sects and cults). voice and lives of the participants” well suited to the
goals of feminism.
Such commitments and practices go only part of
Can there be a feminist the way. Of great significance to feminist researchers is
ethnography? whether the ethnography allows for a non-exploitative
This heading is in fact the title of a widely cited article relationship between the researcher and the researched.
in which Stacey (1988) rebuts the view that there can One of the main elements of such a strategy is for the
be a distinctively feminist ethnography. Reinharz ethnographer not to treat the relationship as a one-way
(1992) considered feminist ethnography important process of extracting information from others, but as
to feminism because: one in which the researcher can provide something in
return. Skeggs’s (1994; 1997) ethnographic research on
• it documents women’s lives and activities, young women represents one such attempt to create
which were previously regarded as marginal a non-exploitative relationship with the people being
and subsidiary to men’s; studied (see Box 10.9). Another way to do this is through
• it understands women from their perspective, so institutional ethnography (Smith, 2005), discussed
that research that “trivializes females’ activities earlier, which has its roots in feminist methodologies.
Skeggs’s (1997) longitudinal ethnographic study gender positioning) and institutions (education
of 83 working-class white women was “based on and the media) frame and inform their responses
research conducted over a total period of 12 years and how this process informs constructions of
including 3 years’ full-time, in-the-field participant their own subjectivity” (Skeggs, 1994, p. 74). This
observation. It began when the women enrolled in comment, like the previous one, reflects a com-
a ‘caring’ course at a local college and it followed mitment to documenting women’s lives and al-
their trajectories through the labour market, edu- lowing their experiences to come through, while
cation, and the family” (1997, p. 1). also pointing to the significance of context.
The elements of a distinctively feminist eth- Skeggs felt that her relationship with the women
nography can be seen in the following comments was not exploitative. For example, she wrote that
by the author: the research enhanced the women’s “sense of self-
• “This ethnography was politically motivated to worth” giving them “the opportunity to be valued,
provide space for the articulations and experi- knowledgeable, and interesting.” She claimed she
ences of the marginalized” (Skeggs, 1997, p. 23). was able to “provide a mouthpiece against injus-
• The “study was concerned to show how young tices” and listen “to disclosures of violence, child
women’s experience of structure (their class and abuse, and sexual harassment” (Skeggs, 1994, p. 81).
236 PART III Qualitative Research
Stacey (1988), however, argued on the basis of her resources for developing a sense of their self-
fieldwork experience that the various situations she worth. More importantly, the feminism of
encountered as a feminist ethnographer placed her the research provided a framework [for them
to see] . . . that their individual problems are
in situations of inauthenticity, dissimilitude, part of a wider structure and not their personal
and potential, perhaps inevitable betrayal, fault. (Skeggs, 1994, p. 88)
situations that I now believe are inherent
in fieldwork method. For no matter how Similarly, Reinharz (1992, pp. 74–75) argued that,
welcome, even enjoyable the fieldworker’s although ethnographic fieldwork relationships may
presence may appear to “natives,” fieldwork sometimes seem manipulative, they often involve
represents an intrusion and intervention into a clear undercurrent of reciprocity. The researcher
a system of relationships, a system of relation- may offer help or advice to her research partici-
ships that the researcher is far freer to leave. pants, or she may give a public airing to normally
(Stacey, 1988, p. 23) marginalized voices (although the ethnographer is
usually the mouthpiece for such voices and as such
Stacey also argued that when the research is may impose a particular “spin” on them). In some
written up, it is the feminist ethnographer’s inter- cases, such as in Campbell’s use of institutional eth-
pretations and judgments that come through and nography to research the experiences of people with
have authority, not those of the women who were disabilities (Box 10.8), the researcher may even work
studied. Skeggs responded by acknowledging that directly with study participants in an effort to help
her academic career was undoubtedly enhanced them improve their lives.
by the study, but added that Stacey’s views present The question of whether there is or can be a fem-
women research participants as victims. Skeggs inist ethnography is clearly a subject of ongoing
maintained that: debate. The fact that female researchers are in careers
that place them in a fairly desirable socioeconomic
The young women were not prepared to be position while their subjects are often situated in
exploited; just as they were able to resist less-advantaged sectors of society is an issue that
most things which did not promise economic is not easily resolved. Of course, male researchers
or cultural reward, they were able to resist who study the socially marginalized face the same
me. . . . They enjoyed the research. It provided problem.
Key Points
• The ethnographer is typically a participant ob- • Field notes are important memory aids for the
server who also uses interviews and documents. ethnographer.
• The ethnographer may adopt an overt or covert • Visual materials such as photographs and videos
role, but the latter comes with serious ethical have attracted considerable interest among eth-
difficulties. nographers in recent years, not just as adjuncts to
• The method of access to a social setting depends data collection but as objects of interest in their
in part on whether it is open (public) or closed (pri- own right.
vate or restricted). • Ethnography conducted from a feminist stand-
• Key informants frequently play an important role point has become increasingly popular, although
in ethnography, but care is needed to ensure that the possibility of a feminist ethnography as such is
their impact on the research is not excessive. open to debate.
10 | Ethnography and Participant Observation 237
• Group 1 prepares a mock funding proposal to a. Explain how its methodology would be superior
conduct a covert study of the site. to that employed by the other two groups—for
238 PART III Qualitative Research
example, Group 1 would explain how a covert class, who are to take the role of a funding
study is superior to a participant-as-observer committee that will decide whether the pro-
study and an observer-as-participant study. ject will be funded.
b. Discuss the ethical implications of the pro-
posed research. 2. As homework, each person in the class is re-
c. Explain how access to the site could be gained. quired to “hang out” at a public, safe location on
d. Explain what the researchers’ demeanour and campus (e.g., residence dining hall, campus café,
dress would be as they interact with people at weight room, library space, etc.) for 30 minutes as
the site. a complete observer. On an appointed date, each
e. Explain how they would record the data they student gives a five-minute mini-presentation on
get, for example, with field notes, photographs, his or her experience. The presentation should:
video recording, and so on.
f. Explain how the study will make a valuable a. highlight anything of significance that was ob-
contribution to knowledge. served; and
g. Upon completion of these tasks, each group b. provide suggestions as to how a more in-depth
presents the mock proposal to the rest of the ethnographic study of the site could be conducted.
Relevant Websites
For examples of ethnographic studies, take a look at http://mediatedcultures.net
the journal Ethnography and the Journal of Contem-
The ethnographic study discussed in the Research in
porary Ethnography (you may have to log on to your
the News box can be found at the Ethnography Lab
institution’s library server to access them).
website.
http://eth.sagepub.com
https://ethnographylab.ca/category/
http://jce.sagepub.com kensington-market/
(Websites accessed 26 October 2018)
At the Digital Ethnography website, explore this
topic, and in particular the use of videos in ethno-
graphic projects.
Chapter Overview
This chapter is concerned with interviews in qualitative research. Two types are discussed: unstructured
and semi-structured. This chapter is concerned mainly with interviews with individuals, but the focus
group method is also examined. The highlights of this chapter include:
• the differences between structured and qualitative interviewing;
• the main characteristics of and differences between unstructured and semi-structured interviewing;
• how to devise and use an interview guide for semi-structured interviewing;
• the kinds of questions that can be placed in an interview guide;
• the importance of recording and transcribing qualitative interviews;
• the potential uses and pitfalls of online interviewing;
• how focus groups should be conducted, including matters such as the number of participants, how
to select them, and how direct the questioning should be;
• issues concerning the interactions between participants in focus group discussions;
• some practical difficulties with focus group sessions, such as loss of control over the proceedings
and the potential for unwanted group effects;
• the significance of qualitative interviewing and focus groups in feminist research; and
• the advantages and disadvantages of qualitative interviewing relative to ethnography.
Practice makes perfect, so why not hone your quali- does qualitative research. A good way to start a discus-
tative interviewing skills in casual, everyday conver- sion is to ask other people where they grew up, or what
sations? It’s often said that the art of conversation their most memorable experiences were; then get
is dying in modern times, but if you approach every them to expand on their answers, taking care to draw
discussion you have as a qualitative interview in the out how they felt about the experiences, what the ex-
making, that may change. Everyday conversation periences meant to them, how they affected their lives,
rarely has a narrowly restricted purpose, and neither and so on. If you’re with a group of people at a party or
▲ RoosterHD/iStockphoto
240 PART III Qualitative Research
on a long trip, you can even hold an impromptu focus spark some very good research ideas. Just remember
group session, secretly appointing yourself the mod- that institutional ethics approval will be required if
erator. Besides being good practice, chance conversa- your casual conversations morph into actual qualita-
tions and seemingly inconsequential interactions can tive research.
Malbon (1999, p. 33) described his strategy for motivations and histories; and to decide how
interviewing “clubbers” (people who frequent to approach the night(s) out that I would be
nightclubs) as follows: spending with the clubber. . . . The main con-
tent of the second [more relaxed] interview
interviews were taped. The first interview was consisted of comments, discussion, and ques-
designed . . . to put the clubber at ease while tions about the club visits . . . and the nature of
also explaining fully and clearly in what ways the night out as an experience. . . . [D]iscussion
I was hoping for help; to begin to sketch in occasionally diversified . . . to cover wider as-
details of the clubbers’ clubbing preferences, pects of the clubbers’ lives.
Wright et al. (2014) used qualitative interviewing number of long-term benefits for the participants
to assess the success of an after-school arts pro- that went beyond developing their theatre skills.
gram that was offered to Canadian youth living Here are some sample interview questions, along
in low-income communities. The study indicated with sample probe questions that were used to
that the program, which involved having the get the respondents to elaborate on their initial
youth participate in theatre productions, had a replies:
included on the list may be asked in response In semi- and unstructured interviews, the process
to what the interviewee says. Sometimes lines is designed to bring out how the interviewees them-
of thought identified by earlier interviewees selves interpret and make sense of issues and events.
are taken up and presented to later ones. None- Bell (2007), for example, wanted to understand what
theless, all the questions on the list are usually western Canadian separatists thought of Canadian
asked and similar wording is used from inter- federalism and the federal party system. Tastsoglou
viewee to interviewee. Box 11.3 illustrates how and Miedema (2003) sought to understand the
these features of qualitative interviewing are meaning of community from the perspective of im-
used in life history research. migrant women, and Smith (2008) examined the
One special form of interview associated with quali- the growing popularity of “narrative interviews”
tative research is the life history interview. It is often (discussed in Chapter 13) in which the person
combined with personal documents such as diaries, studied, instead of responding to an interviewer’s
photographs, and letters. Subjects are invited to look questions, “tells a story.” In the past this approach
back in detail across their entire life course and to was associated with the study of a single life, or
report their experiences and how they understood perhaps two, but today it is increasingly applied to
their world. Valuable insights may be gained from a several lives in a single project. M. Atkinson (2002;
life history regardless of whether other people have 2004), for example, conducted narrative inter-
had the same experiences as the person described views with women tattoo enthusiasts.
in the work. It also has the advantage of illustrat- P. Atkinson (2004) observed that the duration
ing process: how events unfold and interrelate in of life story interviews varies considerably from
people’s lives over long periods of time. study to study, but that it usually takes two or
An example of the life history interview ap- three sessions lasting 60 to 90 minutes each. He
proach is provided by Lewis (1961) in his research provided a catalogue of questions that can be
on the Sánchez family and their experiences in a asked, and divided them into the following groups
Mexican slum: (1990, pp. 43–53):
I asked hundreds of questions of [the five mem-
• birth and family of origin (e.g., “How would you
bers of the Sánchez family]. . . . While I used a
describe your parents?”)
directive approach to the interviews, I encour-
• cultural traditions (e.g., “Was your family differ-
aged free association, and I was a good listener.
ent from others in town?”)
I attempted to cover systematically a wide
• social factors (“e.g., What were some of your
range of subjects: their earliest memories, their
struggles as a child?”)
dreams, their hopes, fears, joys, and sufferings;
• education (e.g., “What are your best memories
their jobs; their relationship with friends, rela-
of school?”)
tives, employers; their sex life; their concepts of
• love and work (e.g., “How did you end up in the
justice, religion, and politics; their knowledge
work you do or did?”)
of geography and history; in short, their total
• inner and spiritual life (e.g., “What are the
view of the world. (Lewis, 1961, p. xxi)
stresses of being an adult?”)
Miller (2000) points out that a recent resur- • major life themes (e.g., “What were the crucial
gence of interest in the life history method reflects decisions in your life?”)
11 | Interviewing in Qualitative Research 243
• vision of the future (e.g., “Is your life fulfilled The chief problem with the oral history inter-
yet?”) view (a problem it shares with life history in
• closure questions (e.g., “Have you given a fair terviews in general) is the possibility of bias caused
picture of yourself?”) by memory lapses and distortions. Sugiman (2004)
A variant of the life history interview is the pointed out in her work on Japanese-Canadian
oral history interview, a technique often used by women interned during the Second World War
historians. It is usually somewhat more specific in that remembering is a social and even political
tone than the interviews described above in that act, involving recalling, forgetting, transforming,
the subjects may be asked to reflect on certain his- and shaping of recollections. On the other hand,
torical events or eras they have lived through. The oral history testimonies can give a voice to groups
emphasis is often on how the individual’s life was that are typically marginalized in historical re-
affected by those things. The information gath- search (this is true of life history interviews in
ered is sometimes combined with other sources of general), either because of their lack of power or
data, such as documents. because they are regarded as unimportant.
meanings of pain for professional wrestlers. Projects • Multiple–case study research generally
like these, unlike quantitative studies, are usually needs some structure to ensure cross-case
not designed to test hypotheses or theories. In short, comparability.
qualitative research is not just quantitative research
without the numbers. Preparing an interview guide
There is a growing tendency to refer to semi- An interview guide is much shorter and less detailed
structured and unstructured interviews collectively than a structured interview schedule. In fact, it is
as in-depth interviews or qualitative interviews, al- often simply a brief list of memory prompts for areas
though the semi- and unstructured types can pro- to be covered in unstructured interviewing and
duce very different results. The choice of one type a somewhat more elaborate list of issues to be ad-
rather than the other is affected by a variety of factors: dressed or questions to be asked in semi-structured
interviewing. What is crucial is that the actual ques-
• Researchers who feel that using even the most tioning is flexible, allowing interviewers to pursue
rudimentary interview guide hinders genuine leads offered by research participants as they begin
access to the world views of members of a social to open up and reveal their view of the social world.
setting are likely to favour an unstructured ap- In preparing for qualitative interviews, Lofland
proach. But even here the interview is rarely and Lofland (1995, p. 78) suggested that researchers
completely unstructured: the researcher usually ask themselves “Just what about this thing is puz-
has at least a general topic that is to be discussed. zling me?” This question can be applied to each of
• If the researcher is beginning the investiga- the research questions generated; it can also serve
tion with a fairly clear rather than a general as a mechanism for generating research questions.
focus, it is likely that the interviews will be Lofland and Lofland suggested that puzzlement may
semi-structured, so that they can address more be stimulated in various ways: by recording random
specific issues. thoughts in different contexts (writing them down
• If more than one person is to carry out the as quickly as possible); through discussions with
fieldwork, semi-structured interviewing may colleagues, friends, and relatives; and, of course, by
be preferred, to ensure that interviewing styles reading the existing literature on the topic. The for-
are comparable. mulation of the research question(s) should not be
244 PART III Qualitative Research
so specific as to close off the alternative avenues of group, etc.); such information is useful for put-
inquiry that may arise during fieldwork. Premature ting people’s comments into context.
closure of the research focus would be inconsistent
with qualitative research’s prime purpose: to explore There are also some practical details to attend to
the world view of the people being studied, rather before the interview begins:
than to test the researcher’s own ideas about that
world. • Become familiar with the research setting,
The interview guide should: which in many cases will be the everyday sur-
roundings of the interviewees. This will help to
• establish a certain amount of order, so that put their comments in context.
questions flow reasonably well, but still allow • Get a reliable recording device. Qualitative
for changes in the order of the questions and researchers (with the possible exception of
the impromptu asking of different questions; ethnographers and people doing participant
• include questions or topics that address the re- observation research) nearly always record
search questions (without being too specific); and then transcribe their interviews. This is
• use language that is comprehensible and fam- important for the detailed analysis required
iliar to those being studied; in qualitative research, and to ensure that the
• not ask leading questions—that is, questions interviewees’ answers are captured in their
that imply a “correct” or socially acceptable own terms. Simply taking notes makes it too
answer (for example, “Have you ever done easy to lose specific words and phrases.
something really stupid, like taking crack?”); • Make sure as far as possible that the interview
and takes place in a setting that is quiet (so that the
• include prompts to remind the researcher to quality of the recording will be as good as pos-
record basic information about the participant sible) and private (so that the interviewee won’t
(name, age, gender, etc.) as well as more specific worry about being overheard).
information that is relevant to the research • Prepare for the interview by cultivating the
questions (position in company, number of traits of a quality interviewer suggested by
years employed, number of years involved in a Kvale (1996) (see Box 11.4).
BOX 11.4 Kvale’s ten traits of an effective interviewer (plus three others)
Kvale (1996) proposed that a successful inter- • Open: responds to what the interviewee
viewer has ten characteristics: considers important; is flexible
• Knowledgeable: is thoroughly familiar with the • Steering: knows what needs to be found out
topic of the interview • Critical: is prepared to challenge what is said
• Structuring: tells the interviewee the purpose of when, for example, there is an inconsistency in
the interview; asks if the interviewee has questions the interviewee’s replies
• Clear: asks simple, easy, and short questions; • Remembering: relates what is said to what has
no jargon previously been said
• Gentle: lets people finish; gives them time to • Interpreting: clarifies and extends the meaning
think; tolerates pauses of interviewees’ statements but without impos-
• Sensitive: listens attentively both to what is said ing meaning on them
and to how it is said; is empathetic
11 | Interviewing in Qualitative Research 245
To Kvale’s list of traits, please add the following • Non-judgmental: does not communicate (even
three: subtly) a moral judgment about what an inter-
• Balanced: does not talk too much, which can viewee has said or done; negative judgments
make interviewees passive, but does not talk may cause people to withhold information
too little, which can cause interviewees to feel that could be useful to the study, while positive
that their responses are not satisfactory judgments may encourage them to focus on
• Ethically sensitive: for example, assures the pleasing the interviewer rather than providing
interviewee that all answers will be treated authentic commentary
confidentially
After the interview, make notes about: General More Interview Formulate
research specific topics interview
area research questions
• how the interview went (was interviewee talk- questions
ative, cooperative, nervous, etc.?);
• where the interview took place; Review/
revise
• any other feelings about the interview (did it interview
open up new avenues of interest?); and questions
Kinds of questions
Revise
The kinds of questions asked in qualitative inter- interview
views are highly variable; Kvale (1996) suggested questions
nine. Most interviews contain virtually all of them,
although those that rely on lists of topics are likely to
Finalize guide
follow a somewhat looser format. Kvale’s nine types
are as follows: FIGURE 11.1 Formulating questions for an
i nterview guide for social research
• Introducing questions. “Please tell me about
when your interest in X first began”; “Have you launched the interviewee into a detailed de-
ever . . . ?”; “Why did you go to . . . ?” scription of shoplifting techniques.
• Follow-up questions. These get the interviewee • Probing questions. These follow up what has
to elaborate on an answer; for example: “What been said through direct questioning, for ex-
do you mean by that?”; even, “Yeesss?” Kvale ample: “Can you say some more about that?”;
suggested repeating significant words in an “You said earlier that you prefer not to do
answer to stimulate further explanation, an X. Can you say what kinds of things have
approach used by Davies (2000) in a study of put you off it?”; “In what ways do you find X
female offenders. One of Davies’s interviewees disturbing?” Box 11.2 provides more examples.
mentioned the use of foil to block detection of • Specifying questions. “What did you do then?”;
security tags on clothing, so she asked, “You “How did (name) react to what you said?”;
mentioned security tags and foil?” which “What effect did (event) have on you?”
246 PART III Qualitative Research
• Direct questions. “Do you find it easy to keep the past and factual matters come first, then ques-
smiling when serving customers?”; “Are you tions about feelings, and finally questions of process
happy with the way you and your husband and summing up.
decide how money should be spent?” Such
questions are perhaps best left until later in the • Initial open-ended questions. “What events
interview, in order not to influence the direc- led to . . . ?”; “What was your life like prior
tion of the interview too much. to . . . ?”; “Is this organization typical of others
• Indirect questions. Before asking an indirect you have worked in?”
question—for example, “What do most people • Intermediate questions. “How did you feel
around here think of the ways that manage- about . . . when you first learned about it?”;
ment treats its staff?”—be sure to ask the inter- “What immediate impact did . . . have on your
viewee for his or own view (“How do you feel life?”; “What do you like most/least about
about . . . ?”). working here?”
• Structuring questions. “Now I would like • Ending questions. “How have your views
to move on to a different topic. Could you about . . . changed?”; “What advice would
describe . . .?” you give to someone who is undergoing a
• Silence. A pause will give the interviewee an similar experience . . . ?”; “If you had to do
opportunity to reflect and amplify an answer; it over, would you choose to work for this
just don’t pause for so long that the interviewee organization?”
feels embarrassed.
• Interpreting questions. “Do you mean that your Most questions are likely to be of the intermedi-
leadership role had to change from one of en- ate kind, and some categories are likely to overlap.
couraging others to a more directive one?”; “Is Nonetheless, these are useful distinctions to bear in
it fair to say that you don’t mind being friendly mind.
toward customers most of the time, but you Remember as well that interviews include differ-
find it more difficult when they are unpleasant ent kinds of topics, such as the following:
or demanding?”
• Values: of the interviewee, of the group, of the
As this list suggests, one of the main jobs of the organization
interviewer is listening: being attentive to what the • Beliefs: of the interviewee, of individual others,
interviewee is saying or not saying. The interviewer of the group
must be active without being intrusive—a difficult • Behaviour: of the interviewee, of others
balance. This means that the interviewer can’t just • Formal and informal roles: of the interviewee,
sit back and relax, even if the interview is being re- of others
corded. In fact, the interviewer must be attuned and • Relationships: of the interviewee, of others
responsive not just to what is said and not said, but • Places and locales
also to what the interviewee is doing. This is im- • Emotions: particularly of the interviewee, but
portant because body language can indicate that also of others
the interviewee is becoming uneasy with a line of • Encounters
questioning. An ethically sensitive interviewer does • Stories
not place undue pressure on an interviewee and is
prepared to cut short any line of questioning that is Try to vary the types of questions you ask (see
clearly a source of anxiety. Kvale’s nine types) and, where appropriate, the topic
It’s likely that the kinds of questions asked will areas (see the list above). Vague or overly general
vary in the different stages of a qualitative interview. questions are usually best avoided; as Mason (2002)
Charmaz (2002) distinguished three types of ques- pointed out, they only force interviewees to puzzle
tion in this connection. Note that questions about over them or to ask for clarification.
11 | Interviewing in Qualitative Research 247
Vignette questions can be used to ground inter- they felt that drug users should engage in (such as using
viewees’ ideas and accounts of behaviour in particu- prophylactics when having sex) and then about how
lar situations (Barter & Renold, 1999). By presenting they felt the hypothetical users would behave (when,
interviewees with concrete and realistic scenarios, for example, there was an opportunity for unprotected
the researcher can elicit a sense of how certain con- sex). Hughes argued that a scenario approach is par-
texts mould behaviour. Hughes (1998) employed the ticularly valuable with sensitive topics and for eliciting
technique in a study of perceptions of HIV risk among a range of responses to different contexts.
intravenous drug users. Context is important for this
topic because drug users’ willingness to engage in Using an interview guide: An example
risky behaviour is influenced by situational factors. Box 11.5 is taken from a study of visitors to Disney
Scenarios depicting risk behaviour were presented and theme parks (Bryman, 1999). The interviewees were
respondents were asked about the kinds of behaviour a man in his sixties and his wife who was two years
Interviewer: OK . What were your views or feelings American Adventure. Now whether we were there
about the presentation of different cultures, as on an unusual day in that respect I don’t know, but
shown in, for example, Jungle Cruise or It’s a Small we saw plenty of black Americans in the Magic
World at the Magic Kingdom or in World Showcase Kingdom and other places, but very few if any in
at Epcot? that World Showcase. And there was certainly
little mention of black history in the American Ad-
Wife: Well, I thought the different countries at
venture presentation, so maybe they felt alienated
Epcot were wonderful, but I need to say more than
by that, I don’t know, but they were noticeable by
that, don’t I?
their absence.
Husband: They were very good and some were
better than others, but that was down to the host Interviewer: So did you think there were any spe-
countries themselves really, as I suppose each of cial emphases?
the countries represented would have been respon-
Husband: Well, thinking about it now, because I
sible for their own part, so that’s nothing to do with
hadn’t really given this any consideration before
Disney, I wouldn’t have thought. I mean some of the
you started asking about it, but thinking about it
landmarks were hard to recognize for what they
now, it was only really representative of the de-
were supposed to be, but some were very well done.
veloped world, you know, Britain, America, Japan,
Britain was ok, but there was only a pub and a Welsh
world leaders many of them in technology, and
shop there really, whereas some of the other pavil-
there was nothing of the Third World there. Maybe
ions, as I think they were called, were good ambas-
that’s their own fault, maybe they were asked to
sadors for the countries they represented. China, for
participate and didn’t, but now that I think about
example, had an excellent 360-degree film showing
it, that does come to me. What do you think, love?
parts of China and I found that very interesting.
Wife: Well, like you, I hadn’t thought of it like that
Interviewer: Did you think there was anything
before, but I agree with you.
lacking about the content?
Husband: Well I did notice that there weren’t many Source: Bryman, A. (1999). Global Disney. In P. Taylor & D. Slater (Eds.), The
black people at World Showcase, particularly the American century. Oxford: Blackwell.
248 PART III Qualitative Research
younger. They had visited Walt Disney World, and frequently interested not only in what people say
were very enthusiastic about the visit. but also in the way they say it. If this aspect is to be
The sequence begins with the interviewer asking fully woven into an analysis, the complete series of
what is considered a “direct question,” in terms of exchanges in an interview must be available. Also,
the nine question types suggested by Kvale (1996). because interviewers are supposed to be highly
The replies are very bland and do little more than alert to what is being said—following up on inter-
reflect the interviewees’ positive feelings about esting points, prompting and probing where ne-
their visit to Disney World. The wife acknowledges cessary, drawing attention to any inconsistencies
this when she says, “. . . but I need to say more than in the interviewee’s answers—it’s best if they don’t
that, don’t I?” Interviewees frequently know that have to concentrate on writing down what is said
they are expected to be expansive in their answers. as well. In addition, recording interviews allows
This sequence occurred about halfway through scrutiny by other researchers, who can evaluate the
the interview, so by then the interviewees were original analysis or even conduct a secondary an-
primed to recognize that more details were ex- alysis, which can help to counter accusations that
pected. There is a tinge of embarrassment that the the findings were influenced by the researcher’s
answer is so brief and not illuminating. The hus- values or biases. Recording also makes it possible
band’s answer is more expansive but not particu- for the data to be reused in ways other than those
larly enlightening. intended by the original researcher; for example, to
Then the first of two important prompts by be analyzed using new theoretical ideas or analytic
the interviewer follows. The husband’s response strategies.
after the initial prompt is more interesting in that As with just about everything in social research,
he begins to wonder whether black people are there is a cost (apart from the financial cost of
under-represented at attractions like the Amer- buying recording equipment) in that the use of a
ican A dventure. The second prompt yields fur-
ther useful reflection, this time that Third World
countries are under-represented in World Show-
Practical Tip | T
ranscribing
case. The couple are clearly aware that it was the
interviews
prompting that led them to these reflections: “I
hadn’t really given this any consideration before A student doing research for a thesis may not
you started asking about it.” This is the whole point have the resources to pay for professional tran-
of prompting: to get the interviewee to think more scription, and unless he or she is an accurate
about the topic and to provide an opportunity for a touch typist, transcription can take a lot of time.
more detailed response. It isn’t a leading question, The usual estimate is five to six hours per hour
since the interviewees were not asked whether they of speech, although using transcription soft-
thought the Disney company failed to recognize ware along with a foot pedal for stops and starts
the significance of black history or ignored the makes the task much easier. Speech-to-text
Third World. There is no doubt that the prompts software, an emerging technology that tran-
elicited the more informative replies, which is pre- scribes spoken words directly into a text file,
cisely their role. is also available, although it is not yet widely
used and at present has certain drawbacks,
Recording and transcription especially for focus groups. The important
The point has been made several times that in thing is to allow sufficient time for transcription
qualitative research the interview is usually audio- and to be realistic about how many interviews
recorded and then transcribed (not just listened can be transcribed in the time available.
to) whenever possible. Qualitative researchers are
11 | Interviewing in Qualitative Research 249
recorder can upset respondents, who may become example, one transcript contained the following
self-conscious or alarmed at the prospect that their passage:
words will be preserved in that way. Most people will
I think unless we want to become like other
allow their interviews to be recorded, though it’s not
countries, where people have, you know,
uncommon for a small number to refuse. When that
democratic freedoms . . .
happens (or when the recording device malfunc-
tions), the interview can still proceed with the inter- But the actual words on the audiotape were:
viewer taking notes. Among those who do agree to
I think unless we want to become like other
be recorded, some will still be fearful and as a result
countries, where people have no democratic
their interviews may not be as informative as they
freedoms . . . (1995, p. 294)
otherwise would be.
Another consideration, as we have seen, is the Clearly, steps need to be taken to check on the qual-
time required to transcribe a recorded interview. ity of transcription.
There is some debate over who should do the tran- In qualitative research there is often wide varia-
scribing: the interviewer (more familiar with what tion in the time that different interviews take. For
was actually said) or someone else (less time- example, in Wilson’s rave study (2002), the inter-
consuming) (Rafaeli et al., 1997, p. 14). Furthermore, views lasted between 45 minutes and four hours.
the vast amounts of text produced must then be Shorter interviews are not necessarily inferior, with
read. Beardsworth and Keil (1992, p. 262) reported the exception of those in which the interviewee was
that their 73 interviews on vegetarianism generated
“several hundred thousand words of transcript ma-
terial.” Clearly, even though transcription has the
Practical Tip | T
ranscribing sections
advantage of keeping the interviewee’s (and inter-
of an interview
viewer’s) words intact, it does so by piling up the
text to be analyzed. It’s no wonder that writers like It’s not always necessary to transcribe every
Lofland and Lofland (1995) advise researchers not word of an interview. Quite often you will find
to leave the analysis of qualitative data until all the that large portions of a particular interview
interviews have been completed and transcribed. (even all of it) won’t be useful, perhaps be-
Insights may be gained by examining the content of cause the interviewee was reticent or because
early interviews, and it may be useful to pursue those the comments made were not as relevant to
points in greater detail in later interviews. Thus there the research topic as you had expected. There
are good grounds for making analysis an ongoing seems to be little point in transcribing such
activity. Also, to wait until all interviews have been material. Before you begin transcribing, there-
completed may make the task of transcription seem fore, listen to the interviews closely once or
overwhelming. twice, and then transcribe only the portions
Transcription may seem a relatively unprob- that seem useful or relevant. The same applies
lematic matter of converting the spoken word into to focus group research, which is often more
written form. However, given the importance of difficult and time-consuming to transcribe
transcripts in qualitative research, the process because of the number of speakers involved.
should not be taken lightly. Transcribers need to be The downside is that you may miss things, or
trained in much the same way that interviewers are. certain comments may emerge as significant
Moreover, even experienced transcribers can make only later on. If that happens, you will have
mistakes. Poland (1995) provided some fascinating to go back to the recording to get the needed
examples of the errors that can result from factors material.
such as mishearing, fatigue, or carelessness. For
250 PART III Qualitative Research
non-cooperative or anxious about being recorded. construct meaning together through their verbal
Indeed, when a long interview contains very little of and non-verbal symbolic interactions. Interpreta-
significance, it may not be worth the time and cost tions emerge out of those interactions; they are not
of transcription. Thankfully, such occasions are rela- simply “out there” in the minds of the study partici-
tively rare. People who have agreed to be interviewed pants, waiting to be discovered.
are usually cooperative, and most loosen up once Ajodhia-Andrews (2016), in her research on how
they get over their initial anxiety about the micro- six Canadian children with disabilities viewed their
phone. As a result, even short interviews are often experiences at school, makes this co-construction of
quite revealing. understanding explicit and even weaves it into her
methodology:
Flexibility in the interview
After coding and searching for emerging
Flexibility in interviewing means more than being
themes and patterns, participants and I col-
responsive to what interviewees say and following
laboratively told, re-told, wrote, and re-wrote
up on the interesting points they make. A flexible
their narratives, negotiating which pieces of
interviewer is able to vary the order of questions and
stories to include, what they thought of sur
clear up inconsistencies in answers. Flexibility is also
facing themes and interpretations. . . . Through
important when audio recording equipment breaks
this dialogue and debate participants and I
down, or when an interviewee refuses permission for
co-constructed narratives together, whereby
recording to take place.
I also became part of the conversations, re-
Another common problem for interviewers is
sponding to their stories and experiences,
knowing when to switch off their recording equip-
thus shaping and re-storying their narra-
ment. Often an interviewee will resume talking
tives. These narratives become unique to the
about the topic of interest as soon as the machine
interaction between myself and participants;
is turned off. It’s usually not feasible to switch the
if participants shared their stories and under-
machine back on again, so try to keep it going as
standings with any other person/researcher,
long as possible. If that fails, take some notes either
the narratives would never be conveyed and
while the person is talking or as soon as possible
told in the same way. . . . (Ajodhia-Andrews,
after the interview. Such “unsolicited accounts” can
2016, p. 262)
often be the source of revealing information. This is
certainly what Parker found in his study of organ-
izations: many interviewees offered “unrecorded
Focus groups:
comments prefixed with a silent or explicit ‘well, if
you want to know what I really think.’ . . . Needless An introduction
to say, a visit to the toilet to write up as much as Most people think of an interview as an exchange
I could remember followed almost immediately” between one interviewer and one interviewee. The
(2000, p. 236). focus group technique, however, involves speaking
with more than one (usually at least four) inter-
Being reflexive about how qualitative viewees at the same time. Essentially it is a group
data are co-constructed interview in which the interviewees can speak to and
Finally, when transcribing and later interpreting interact with one another.
qualitative interviews, researchers should reflect on
how their own interactions with participants have • Most focus group researchers work within
affected what was said, how it was said, and what was a qualitative research tradition. They try
left unsaid. Qualitative interviewing is more than to provide a fairly unstructured setting in
simply digging out pre-existing verbal gems from which the person who runs the focus group,
interviewees. The interviewee and the interviewer usually called the moderator or facilitator,
11 | Interviewing in Qualitative Research 251
BOX 11.6 Focus group in action: AIDS in the Media research project
Focus groups were part of Kitzinger’s research on individuals drawn together simply for the pur-
the representation of AIDS in the mass media. The poses of the research, we elected to work with
focus groups were concerned with how “media pre-existing groups—people who already lived,
messages are explored by audiences and how worked or socialized together” (Kitzinger, 1993,
understandings of AIDS are constructed. We were p. 272). As a result, the people in each group were
interested not solely in what people thought but known to each other: a team of civil engineers
in how they thought and why they thought as they working on the same site, six members of a re-
did” (Kitzinger, 1994, p. 104). tirement club, five intravenous drug users, and so
Details of the groups are given in Table 11.1. on. The sessions themselves were “conducted in
Since one goal of the research was to empha- a relaxed fashion with minimal intervention from
size the role of interaction in the construction of the facilitator—at least at first” (Kitzinger, 1994,
meaning, it was important to provide a platform p. 106). Each session lasted approximately two
for doing that. Accordingly, “instead of work- hours and was tape-recorded.
ing with isolated individuals, or collections of
when the participants are likely to have a lot to say, experience, and so on. The aim is to look for any vari-
as often occurs when they are emotionally involved ation in how the different groups discuss the matter
with the topic, when the topic is controversial or at hand. For example, to examine women’s responses
complex, or when exploring personal interpreta- to scenes of violence, Schlesinger et al. (1992) showed
tions. A larger group is appropriate when the re- 14 groups (see Table 11.1 and Box 11.7) four levels of
searcher wants “to hear numerous brief suggestions” mass-media violence: incidental violence, moderate
(Morgan, 1998, p. 75). violence, marital violence, and an extremely vivid
For most topics there is no need to select a par- sexual assault scene. The authors concluded:
ticular kind of participant; however, the topic should
be relevant to those taking part. A wide range of Having a particular experience or a particu-
people is often required, but participants may be put lar background does significantly affect the
into separate groups on the basis of age, gender, edu- interpretation of a given text. The four pro-
cation, whether they have (or have not) had a certain grams screened are obviously open to various
readings. However, on the evidence, how they takers. They then sought out participants in various
are read is fundamentally affected by various social clubs. They defended this new approach on the
socio-cultural factors and by lived experience. grounds that, because their research questions con-
(Schlesinger et al., 1992, p. 168; emphasis in cerned shopping in relation to the construction of
original) identity and how it relates to people’s sense of place,
recruiting people who knew each other would be
A slight variation on this approach is Kitzinger’s highly appropriate. Recruiting people from natural
(1994) study of reactions to media representations of groups is not always feasible, however, because of the
AIDS (see Box 11.6 and Table 11.1). Her groups were difficulty of getting everyone in the group to par-
made up of people in a variety of situations. Some ticipate. Morgan (1998) revealed another downside:
were what she called “general population groups” (for because people in natural groups know one another,
example, a team of civil engineers working on the they may share certain assumptions that they feel no
same site) but others were made up of people likely to need to explain or justify. He suggested that in cases
have a special interest in AIDS, such as male sex work- where it is important that such assumptions are
ers and intravenous drug users. Increasingly, focus made explicit, a group of strangers is a better choice.
group practitioners try to discern patterns of variation
by putting together groups with particular attributes.
A further issue in designing a focus group is
Asking questions and level
whether to select people who are unknown to each of moderator involvement
other, or to use natural groups (for example, friends, There are different questioning strategies and ap-
co-workers, or students in the same class). Some re- proaches to moderating focus group sessions. Most
searchers prefer to exclude people who know each lie somewhere between a rather open-ended style
other, fearing that known status differences or and a more structured one. Macnaghten and Jacobs’s
pre-existing patterns of interaction would contamin- (1997) middle-of-the-road approach to question
ate the session. Others prefer to select natural groups structuring can be seen in the following passage in
whenever possible. Holbrook and Jackson (1996) re- which a group of working women express cynicism
ported that, for their research on shopping centres, about government and experts regarding the reality
they initially tried to secure participants who were of environmental problems (in this passage “F” is
unknown to each other, but this strategy attracted no “female” and “Mod” is “moderator”):
In this exchange, the moderator focuses on the topic If in doubt, the best advice is to err on the side of
to be addressed but is also able to pick up on what minimal intervention.
the group says.
How involved should the moderator or facilitator Recording and transcription
be? As with question structuring, above, the most Recording is even more important with focus groups
common approach is middle-of-the-road. There is a than it is in other forms of qualitative research. Writ-
tendency to use a fairly small number of very general ing down not only exactly what is said but who says
questions to guide a focus group session. Obviously, it is too difficult. In an individual interview you may
if the discussion goes completely off topic it may be be able to ask a respondent to “hold on” while you
necessary to refocus the participants’ attention, but write down a response, but this is not feasible in an
even then the moderator must be careful, because interview where several people are speaking rapidly,
apparent digressions can often reveal something of and would almost certainly break the flow of the
significance. More direction is probably needed if the discussion.
participants are not addressing the research ques- Transcribing focus group sessions is also more
tions, or if a particularly meaningful point made by complicated and hence more time-consuming than
one participant is not followed up by the others. it is with other interview forms. Sometimes voices
Both intervention and non-intervention carry are hard to distinguish, making it difficult to deter-
risks. The style of questioning and moderating de- mine who is speaking. Also, people sometimes talk
pends on the nature of the research topic; if it is over each other, which can make transcription even
embarrassing for some participants, for example, more problematic. Therefore a very high-quality
additional direction may be required from the mod- recording device, capable of picking up even faint
erator. Levels of interest and knowledge among the voices from many directions, is a necessity. Focus
participants can make a difference as well. Limited group transcripts always seem to have more missing
interest or knowledge on the part of participants bits than transcripts from other sorts of interview,
may require a somewhat more structured approach. mainly because of audibility problems.
Group interaction in focus This sequence brings out the emerging consensus
around the question of who has heart attacks and
group sessions why. However, as Kitzinger (1994) suggested, argu-
Kitzinger (1994) observed that reports of focus mentative interaction in focus groups can be equally
group research frequently fail to take group inter- revealing. In such cases moderators can play an im-
action into account. This is surprising, given that portant role in identifying differences of opinion
such interaction is a distinguishing feature of focus and exploring with participants the factors that lie
groups. Wilkinson reviewed more than 200 fo- behind them. Disagreement can give participants a
cus group studies, published between 1946 and 1996, chance to revise their opinions or to think about why
and concluded: “Focus group data is most commonly they hold them. A passage from Schlesinger et al.
presented as if it were one-to-one interview data, (1992; see Table 11.1) illustrates. The group is made
with interactions between group participants rarely up of women with no experience of violence, and the
reported, let alone analyzed” (1998, p. 112). debate concerns a rape scene in a film:
In the context of her research on the coverage
Speaker 1: I think . . . that they could’ve ex-
of AIDS in the mass media, Kitzinger (1994) drew
plained it. They could easily leave [out] that
attention to two types of interaction in focus
rape scene.
groups: complementary and argumentative. Com-
Speaker 2: But it’s like that other film we
plementary interaction allows collective interpret-
watched. You don’t realize the full impact, like,
ations and understandings to develop, with each
the one we were watching, the first one, until
participant building on the preceding remark, as
you’ve got the reconstruction.
in the following passage taken from Morgan and
Speaker 3: Yeah, but I think with that sort
Spanish’s (1985, p. 414) research on heart-attack
of film, it would cause more damage than it
victims:
would good, I mean, if someone had been
raped, would you like to have [to] sit through
No. 1: But I think maybe what we’re saying here
that again?
is that there’s no one cause of heart attacks,
there’s no one type of person, there’s probably The debate then moved on to consider the signifi-
umpteen different types of heart attacks and cance of the scene for men:
causes coming from maybe smoking, maybe
Speaker 1: Men would sit down and think,
obesity, maybe stress, maybe design fault, her-
“Well, she asked for it. She was enjoying it and
editary, overwork, change in life style. Any of
look, the men around enjoyed it.” (Schlesinger
these things in themselves could be . . .
et al., 1992, pp. 51–52)
No. 2: And when you start putting them in
combination [unclear]. It seems, then, that behind the unease some of the
No. 3: Yeah, you may be really magnifying women expressed about the rape scene was the idea
each one of these particular things. that some men might enjoy it and identify with the
No. 2: Yeah, and depending on how, and in onlookers depicted in the film. This interpretation
each person that magnification is different. came out because of an earlier disagreement within
Some people can take a little stress without the group, which allowed the women to give a fuller
doing any damage, some people can take a account of their reactions to the scene.
little smoking, a little drinking, a little obesity, In sum, as Kitzinger (1994) argued, drawing
without doing any damage. But you take a attention to patterns of interaction in focus groups
little of each of these and put them together allows a researcher to determine how group par-
and you’re starting to increase the chances ticipants view the issues at hand. The posing of
of damage. And any one of these that takes a questions by focus group members as well as the
magnitude leap increases the chances. agreement and disagreement among them help to
11 | Interviewing in Qualitative Research 257
bring out this knowledge. The resolution of disagree- researchers often see it as an advantage. However,
ments also forces participants to explain why they the question of control raises questions about
hold the views they do. the extent to which a researcher should allow the
participants to “take over” the discussion. There
is clearly a delicate balance to be struck in decid-
Limitations of focus groups ing how involved moderators should be and how
Focus groups are particularly useful in illustrating much a set of prompts or questions should be al-
how meaning is jointly constructed. What, then, are lowed to influence the conversation. What is not
their chief limitations? clear is the degree to which it is appropriate to
surrender control, especially when there is a set
• The researcher probably has less control over of research questions to be answered.
the proceedings than in individual interviews. • An unwieldy amount of data is sometimes
As we have seen, not all writers on focus groups produced. For example, Bloor et al. (2001) sug-
perceive this as a problem; indeed, feminist gested that one focus group session can take up
258 PART III Qualitative Research
to eight hours to transcribe—somewhat longer to impress each other or avoid embarrassment, and
than an equivalent personal interview, because were influenced by peer pressure. But this does not
of variations in voice pitch and the need to mean that such data are tainted: in fact, it may be
identify who says what. Also, group recordings the gulf between privately and publicly held views
often include inaudible comments from par- that is of interest.
ticipants sitting too far from the microphone,
and these also affect transcription. • Madriz (2000) proposed that there are cir-
• The data may be difficult to analyze. Develop- cumstances when individual interviews are
ing an analysis strategy that incorporates both more appropriate. One is when participants
the themes of the discussion and the patterns are likely to disagree profoundly with each
of interaction is not easy. other; another is when participants are not
• Focus groups are difficult to arrange. It may comfortable in one another’s presence (for
be hard to get people to agree to participate, example, people in hierarchical relation-
and harder still to get them to show up. Very ships). Finally, because of the potential for
small payments, such as gift cards, are some- focus groups to cause discomfort among par-
times offered to induce participation, but it is ticipants (for example, when intimate details
still common to find that some people will not of private lives need to be revealed), an indi-
turn up. vidual interview or even a questionnaire may
• Group effects may be a problem. This category be better.
includes the obvious challenge of dealing
with participants who either are too reticent
to speak up or hog the stage. Krueger (1998)
Online interviews and focus
suggested that the people running the focus groups
group should make it clear to participants that Conducting qualitative interviews and focus groups
other people’s views are definitely required; for online has become increasingly common as more
example, he suggested saying something like and more members of the general public become
“That’s one point of view. Does anyone have familiar with the Internet. At the time of writing,
another point of view?” (1998, p. 59). As for online focus groups appear to be attracting greater
those who are reluctant to speak, Krueger rec- attention than online personal interviews, perhaps
ommended that they be actively encouraged to because of the time and administrative effort that
say something. can be saved by conducting focus groups online.
There is less to be gained in that regard with online
It would be interesting to know whether agree- personal interviews, unless face-to-face interviews
ment among focus group participants is more would require a great deal of travel.
common than disagreement; if that is the case, one Doing this sort of research online has its pros
reason might be group pressure to conform. It’s also and cons. Selecting participants may be difficult,
possible that participants are more likely to express since everyone taking part must have access to the
socially acceptable views in a group setting than in necessary hardware and software. Carefully selected
individual interviews. Morgan (2002) cited a study websites, online bulletin boards, and chat rooms are
in which group interviews with boys on the topic potential sources of recruitment.
of relationships with girls were compared with in-
dividual interviews with boys on the same subject. Online qualitative interviews
Alone, the boys expressed a degree of sensitivity Markham (1998) conducted Internet interviews
that was not evident in the group context, where with people she recruited after “lurking” (reading
they tended to express more macho views. This sug- but not participating) for some time in computer-
gests that in the group setting the boys were trying mediated communication forums such as chat
11 | Interviewing in Qualitative Research 259
rooms. Questions were asked and answered in real more difficult with larger numbers, and as Adriaens-
time rather than via email, which could have meant sens and Cadman (1999) suggested, large groups can
delays of hours or days. Markham used an interview present administrative problems.
guide, and the interviews lasted between one and Before starting an online focus group session,
four hours. Such interviews are a challenge for both moderators are advised to send out a welcome
interviewer and interviewee because neither party message introducing the research and laying out
can pick up on visual cues (puzzlement, anxiety) or some of the ground rules for the discussion. There
auditory cues (sighs, groans). is evidence that participants respond more posi-
One of Markham’s research interests was the re- tively if the researchers reveal something about
ality of online experiences. This can be seen in the themselves (Curasi, 2001). This can be done in the
following brief sequence (Markham is Annette): opening message or by creating links to a personal
website.
Annette: How real are your experiences in the In online focus groups, participants respond
Internet? more or less immediately to whatever has just been
Sherie: How real are experiences off the Inter- “said,” whether by the moderator or by other par-
net? (1998, p. 115) ticipants. As all are simultaneously online, contri-
butions can be responded to as soon as they appear
Markham noted that her notion of “real” was dif- on the screen (and with some forms of software,
ferent from that of her interviewees. For her, “real” the contributions can be seen as they are being
carried connotations of genuineness and authenti- typed). As Mann and Stewart (2000) observed, be-
city, but for her interviewees it had more to do with cause several participants can type in a response
distinguishing experiences that occur offline (real) to something at the same time, the convention
from those online (not real). These distinctions be- of taking turns in regular conversation is largely
tween life online and life offline are likely to become sidelined.
less significant as everyday life is increasingly lived Online focus groups are unlikely to replace their
on the Internet. face-to-face counterparts, but they are likely to be
Hanna (2012) reports that Internet interviews used for certain kinds of research topics and/or
using webcam technologies such as Skype can be samples. As regards the latter, people who are dis-
as effective as in-person qualitative interviews, persed or inaccessible are especially suitable for
and that they have advantages similar to telephone online focus group research; slow typists are not.
interviews. No travelling is needed with Skype As Sweet (2001) pointed out, the most appropriate
interviews, they can be postponed and re-scheduled topics for online research are likely to be those in-
without difficulty, and the interviewee is allowed to volving sensitive issues or Internet use itself (as in
stay home and thus feel safe and secure. Moreover, Box 11.8).
both audio and visual aspects of the interview are Much research on the Internet treats the tech-
easily recorded. The use of technologies like Skype nology as a given, but some studies examine how
to conduct qualitative interviews is still relatively it is interpreted by users. Hine (2000, p. 9), for in-
rare, so their impact on qualitative research remains stance, presented the Internet “as a product of cul-
to be seen. ture . . . shaped by the ways in which it is marketed,
taught, and used.” Box 11.9 summarizes the main
Online focus groups advantages and disadvantages of conducting focus
Mann and Stewart (2000) suggested that an online groups and personal interviews online compared to
focus group should have between six and eight administering them in person. The two methods are
people. Larger groups make it difficult for some combined because the tally of advantages and dis-
people to participate, especially those with limited advantages applies more or less equally well to both
keyboard skills. Also, moderating the session can be of them.
260 PART III Qualitative Research
O’Connor and Madge (2001; 2003; see also Madge Kerry: I feel the same. Like the HV [health
& O’Connor, 2002) employed conferencing software visitor] is judging even though she says she isn’t
in a virtual focus group study that examined the Kerry: Although my HV has been a lifeline as
use of an online information website for parents. I suffer from PND [post-natal depression]
Initially, the researchers set up a Web survey on Amy: Also, there are some things that are so
the use of the site. When respondents sent in their little that you don’t want to feel like you’re
questionnaires, they were thanked for their par- wasting anyone’s time. Asking the HV or GP
ticipation and asked via email whether they were might get in the way of something mroe
willing to take part in an in-depth interview. Of the important, whereas sending an e-mail, the
155 respondents who returned the questionnaires, person can answer it when convenient.
16 agreed to be interviewed and were sent the soft- Amy: My HV is very good, but her voice does
ware to install on their computers. The researchers sound patronising. I’msure she doesn’t mean
tried to ensure that each group was asked more or it, but it does get to me . . .
less the same questions. For each session, the re- Kerry : Being anon means that you don’t get
searchers introduced themselves and asked par- embarrassed asking about a little point or
ticipants to do likewise. In addition, they placed something personal. (O’Connor & Madge,
descriptions and photographs of themselves on a 2001, p. 10.4)
website to which participants were directed. An im-
This extract reveals a good flow without inter-
portant part of the process of building rapport was
vention by the researchers. It contains several mis-
the fact that both of the researchers were mothers.
takes (“mroe,” “I’msure”), but they are retained to
They found that the relative anonymity afforded by
preserve the authenticity of the interaction. Another
the Internet gave participants greater confidence
plus is that there is no need to transcribe the ma-
in asking embarrassing questions, a finding that
terial because it is already in textual form. Also, the
has implications for online focus groups in general.
fact that participants appear to relish the anonym-
This can be seen in the following excerpt:
ity of the Internet has implications for online focus
Amy: I feel better asking BW [Babyworld] groups, since participants may find it easier to ask
than my health visitor as they’re not going to naive questions or make potentially embarrassing
see how bad I am at housekeeping!!! comments online than in face-to-face focus groups.
BOX 11.9 A
dvantages and disadvantages of conducting focus groups and
individual interviews using an online, text-based format rather
than face-to-face
• Interviewees and focus group participants are • Only people with access to online facilities and
able to read and reread their statements (and who are comfortable using them are likely to
in focus groups, those of the other participants participate.
as well). • It is more difficult for the interviewer to estab-
• Audio recording is not required; this elim- lish rapport with interviewees, though this is
inates both interviewee apprehension and less of a problem when the topic is of particular
the need for time-consuming and expensive interest to the participants.
transcription. • Probing is more difficult, though not impos-
• Records of online interviews are more accur- sible. Curasi (2001) reported some success in
ate because nothing has to be transcribed eliciting further information from respondents,
and problems of mishearing don’t arise. This but working online makes it easier for them to
is a particular advantage with focus group ignore or forget about such requests.
discussions. • There is less spontaneity in responding since
• Participants can use pseudonyms to conceal interviewees are able to reflect on their an-
their identity, making it easier for them to dis- swers more carefully than is possible in a
cuss embarrassing issues or to divulge poten- face-to-face situation. However, this can be
tially unpopular views. construed as an advantage in some respects
• In focus groups, shy or quiet people may find since interviewees are likely to give more con-
it easier to participate and overbearing partici- sidered replies.
pants are less likely to predominate, though • There is a tendency for non-response rates to
variations in keyboard skills may still prevent be higher in online personal interviews.
equal participation. • The researcher cannot be certain that the par-
• Focus group participants are less likely to be ticipants are who they say they are (though
influenced by factors such as the age, eth- this may sometimes be the case in face-to-face
nicity, appearance, and (if pseudonyms are interviews as well).
used) possibly even the gender of the other • Online interviews and focus groups require
participants. considerable commitment from interviewees
• Similarly, interviewees and focus group partici- and participants if they have to install software
pants are less likely to be affected by the char- and remain online for extended periods of
acteristics of interviewers or moderators, and time. This is especially problematic if they have
the latter are less likely to be affected by the to pay for the software (though reimbursement
participants’ characteristics; in this way, reac- may be possible).
tivity is reduced. • The interviewer/moderator may not be aware
• Online focus groups and interviews allow that the interviewee/participant is distracted
people to participate in an environment that by something and in such circumstances will
is “anonymous, safe and non-threatening” continue to ask questions as if they have the
(O’Connor & Madge, 2001, p. 11.2); this may be person’s full attention.
especially helpful for vulnerable groups. • Online connections may be lost and long break-
• Similarly, researchers do not need to enter downs may ensue, interrupting the flow of the
potentially unsafe environments to reach the discussion.
people they are studying. • Interviewers cannot rely on body language to tell
them when someone is puzzled or (in the case of
Disadvantages focus groups) losing interest in the discussion.
• Online focus groups and interviews take longer Sources: Adapted from Adriaenssens and Cadman (1999); Bampton and
Cowton (2002); Clapper and Massey (1996); Curasi (2001); Mann and Stew-
than their face-to-face counterparts. art (2000); O’Connor and Madge (2001); Sweet (2001); Tse (1999).
262 PART III Qualitative Research
or not—may have to decide what to do when she or However, not all writers accept the argument
he strongly disagrees with the views of the people that focus groups are more natural than in-
under study. dividual interviews. Even when pre-existing
groups are used, gathering people to discuss
The focus group as a feminist method a certain topic (such as a television program)
The use of focus groups by feminist researchers has is not inherently naturalistic, the critics claim,
grown considerably (see Box 11.10), and Wilkinson because the interaction is so unusual and con-
(1998, 1999b) has argued for its great potential in this trived. In real life, people are rarely asked to
regard. Three aspects of the method are particularly discuss, in a group, issues not of their choosing.
compatible with the ethics and politics of feminism: • Feminist researchers have expressed a pref-
erence for methods that study the individual
• Focus group research is less artificial than within a social context. The tendency for most
many other methods because it features group methods to treat the individual as an isolated
interaction, which is a normal part of social entity devoid of a social context is disliked by
life. The tendency to recruit participants from many feminist researchers, who prefer to ana-
naturally occurring groups reinforces the real- lyze “the self as relational or as socially con-
istic quality of such research, since people are structed” (Wilkinson, 1999b, pp. 229–230).
able to discuss matters in situations that are • As we noted earlier, feminist researchers are
normal for them. As a result, there is greater suspicious of methods that exploit participants
likelihood that the researcher’s findings will and put the researchers in a position of power
resonate with the “lived experience” of women. over the respondent. Wilkinson observed that
In Canada, feminist research using focus groups mothers on welfare after their support services
includes Cassano and Dunlop (2005), Wane were reduced and enhanced measures were put in
(2004), Ristock (2001), Little (2001), Wachholz and place to force them to look for work. She was able
Miedema (2000), and Carter et al. (2016). Cassano to dispel some of the myths about these women
and Dunlop examined South Asian–Canadian and drew attention to some of the ways they were
women’s perspectives on balancing work and forced to cope, from going hungry some days to
family life, which led to a recommendation that a moving in with abusive ex-partners. Wachholz
day centre for seniors and children be established. and Miedema examined how immigrant women
Wane’s focus groups revealed that black Canadian felt about the police intervention that followed
feminism encompasses a broad range of meanings official reports of spousal abuse. The “solution”
and paradigms, but is united in its efforts to bring a often brought with it the harm of extra economic
black feminist voice to both the creation of know- hardships for immigrant women who were already
ledge and the struggle against oppression. Ristock socioeconomically vulnerable. Finally, Carter and
examined how 70 feminist counsellors responded colleagues used focus group data to reveal how
to clients reporting abuse in a lesbian relationship, women living with HIV in British Columbia have
and found that even feminist models and ther- to negotiate access to treatment facilities, and in
apies for heterosexual violence may need revision particular how HIV stigma, racism, and class bias
in such instances. Little looked at Ontario single may prevent them from getting the care they need.
264 PART III Qualitative Research
the risk of exploitation is greatly reduced in more fleeting contacts, although qualitative research
focus groups because the participants are able interviews can last many hours and re-interviewing
to take control of the discussion and even sub- is not unusual.
vert the goals the researcher had for the ses-
sion. As a result, participants’ points of view Learning the local language
are much more likely to be revealed in a focus Becker and Geer (1957) argue that ethnographers are
group than in a traditional interview. like social anthropologists visiting a distant land: to
understand its culture, they must have a good grasp
Wilkinson did not argue that the focus group, or of its language. To penetrate the culture of particular
indeed any method, is inherently feminist: only that group, it’s particularly important to be familiar with
focus groups have considerable potential for feminist its “argot”: its informal slang and special uses of words.
research. A key point is that they allow the voices of Learning that informal language takes prolonged ob-
highly marginalized women to be heard. According servation. See Chapter 12 for more on language analysis.
to Madriz (2000, p. 843), for example, focus groups
constitute a relatively rare opportunity for women of Things taken for granted
colour in lower socioeconomic classes to “empower Because interviews rely primarily on the inter-
themselves by making sense of their experience of viewees’ accounts of their world, elements of that
vulnerability and subjugation.” world that the interviewees take for granted may not
be mentioned. For example, a street sex worker may
never mention her relationship with the police, and
Qualitative interviewing an interviewer may never think to ask about it. By
(without immersion in a social contrast, an ethnographer immersed in the street
scene may learn that sex workers have to deal with
setting) versus ethnography police surveillance on a regular basis, and that it is
The aim of this section is to compare the merits and an important part of their daily lives.
limitations of interviewing in qualitative research
(without immersion in a social setting) with those Deviant and hidden activities
of ethnography. They are probably the two most There are certain activities that most people are re-
prominent methods of data collection in qualitative luctant to talk about in one-on-one interviews:
research, so there is some value in assessing their drug-taking, violence, shoplifting, illegal commerce,
relative strengths. hooliganism, and so on. For that reason, much of what
is known about criminal and deviant subcultures has
Advantages of ethnography compared been gleaned from ethnography. Ethnographers have
to qualitative interviewing also uncovered information about workplace resist-
Seeing through others’ eyes ance practices (such as working-to-rule and industrial
As we noted in Chapters 1 and 9, seeing through sabotage) and groups that support deviant ideologies,
others’ eyes is one of the main purposes of qualita- such as white supremacists. Ethnographers are better
tive research. On the face of it, ethnographers would positioned than interviewers to infiltrate the social
seem to be especially well placed to gain insight into worlds of people who are wary of talking to outsiders.
social reality in this way, because of their prolonged
immersion in particular social settings. Ethnog- Sensitivity to context and flexibility
raphers are not only in much closer contact with Extensive contact with a social setting allows the
people for a longer period of time than interviewers ethnographer to map the context of people’s behav-
are: they also participate in many of the same kinds iour. Interacting with people in a variety of different
of activity as the people they are studying. Research situations makes it possible to connect behaviour
that relies on interviewing alone is likely to entail and context.
11 | Interviewing in Qualitative Research 265
Naturalistic emphasis around in other people’s homes, waiting for the violence
to unfold? Reactivity would impinge in a big way, since
Ethnography can be more naturalistic because the perpetrators would do their best to hide their violence.
researcher confronts members of a social setting in
their natural environments. Interviewing, which Reconstruction of events and future plans
tends to interrupt the normal flow of events even In-depth reconstruction of past events and plans for
when it is informal, is often less naturalistic. It’s no future behaviour is not possible through observation
surprise that when discussing naturalism as a trad- alone. Qualitative research entails reconstruction of
ition in qualitative research, Gubrium and Holstein events when interviewees are asked to think about
(1997) referred mostly to ethnography. how a series of previous activities might have created a
current situation. Beardsworth and Keil used the sym-
Advantages of qualitative interviewing bolic interactionist notion of career to understand how
in comparison to ethnography people become vegetarians. In fact, most qualitative
Issues resistant to observation studies ask about events that occurred before the study
Many issues are simply not open to observation, so began; for example, questions may be asked about the
asking people about them may be the only way to get early family experiences of gang members, recent im-
information about them. For example, consider domes- migrants, or sex workers. Some call this “retrospective
tic violence. Would it be feasible for a researcher to hang interviewing.” See Box 11.11 for a further example.
McKeganey and Barnard (1996) discussed their [Then] after it happened I thought it was
strategies for conducting research into sex workers bad, I didn’t like it but at least I was getting
and their clients. Their approach was largely that paid for it. I’d been abused by my granddad
of the observer-as-participant (see Figure 10.2), when I was 11 and it didn’t seem a million
in that their research was based primarily on inter- miles from that anyway. (1996, p. 25)
views with sex workers and their clients, supple-
One area of particular concern to McKeganey and
mented by (frequently accidental) observation of
Barnard was the spread of HIV/AIDS infection and
interactions and overheard conversations. The
its implications for sex workers and their work.
interviews they conducted were especially im-
This topic was specifically addressed in the inter-
portant in gaining information on matters such as
views. For example:
how the sex workers had moved into that line of
work; permitted and prohibited sex acts; links with I’ve got a couple of punters who’ll say “I’ll
drug use; experience of violence; and manage- give you so and so if you’ll do it without [a
ment of identity. In the following passage, a sex condom].” But never, I always use a condom
worker reconstructs her movement into the trade: for anal sex, oral sex and even for hand jobs;
there’s no way I’ll let them come anywhere
I was 14 and I’d run away from home. I ended
near me. (1996, p. 66)
up down in London where I met a pimp. . . .
He’d got me a place to stay, buying me things You still get the bam-pots [idiots] asking for
and everything and I ended up sleeping with sex without. I had one the other night—I
him as well. . . . One night we got really drunk said, “where have you been living—on a
and stoned and he brought someone in. . . . desert island?” (1996, p. 66)
266 PART III Qualitative Research
Reactive effects any sustained absence from work and other personal
commitments.
Reactive effects are by no means straightforward. As
with structured observation (see Chapter 6), it can Greater breadth of coverage
be anticipated that the presence of an ethnographer Interviewing may allow access to a wider variety of
will lead people to behave unnaturally (although people and situations. In ethnographic work, the
ethnographers, like researchers using structured ob- researcher is invariably limited to a fairly restricted
servation, typically find that their subjects become range of people, places, and incidents. An ethnog-
accustomed to their presence and over time begin rapher observing a large organization, for example,
to behave more naturally). Ethnography also suffers is unlikely to have extensive knowledge of operations
from the related problem that an observer disturbs outside the area under study.
the very situation under study, sparking conversa-
tions and interactions, both with and about the ob- Specific focus
server, that otherwise would not occur. As we noted in Chapter 9, qualitative research
sometimes begins with a specific focus; Silverman
Less intrusive in people’s lives (1993) was even critical of the notion that it should
Ethnography can be costly in that the researcher is be regarded as an open-ended form of research.
likely to take up more of the participants’ time than Qualitative interviewing seems to be better suited
would be the case with interviews alone. Interviews in for dealing with a specific issue, since an interview
qualitative research can sometimes be very long, and can address a particular matter in detail. Research
as we have noted, re-interviewing is not uncommon. by Bryman et al. (1996) had a very specific focus:
Nevertheless, the impact of undergoing an interview namely, conceptions of leadership among police of-
is probably still less than the impact of having to ficers. Because of its clear focus, it was more appro-
deal with an ethnographer on a regular basis. Eth- priate to conduct the research using semi-structured
nography can be especially intrusive in terms of time interviewing than ethnography, since issues involv-
when it is conducted in organizational settings, be- ing leadership are unlikely to emerge naturally on a
cause it disrupts the rhythm of the work day. regular basis.
Checklist
Checklist of issues to consider in qualitative interviewing
☐☐ Have you thought about how you will present point of view rather than imposing a frame of
yourself in the interview, such as how you will reference on them?
dress? ☐☐ Has the setting in which the interviews
☐☐ Is there a clear and comprehensive way of will take place been checked out? Has the
introducing the research to interviewees? recording equipment been put through
☐☐ Does the interview guide clearly relate to the a dry run? Have all aids to be used (e.g.,
research questions? visual aids, film clips, case studies) been
pre-tested?
☐☐ Has a pilot test been done with some
appropriate respondents? ☐☐ Is there a plan in place if the interviewee does
not turn up for the interview?
☐☐ Have the interviewers been fully trained?
☐☐ Does the interview guide contain a good For a focus group:
ixture of question types (e.g., probing,
m
specifying, and direct questions)?
☐☐ Have you planned what you will do if not all
participants turn up for the session?
☐☐ Do the interviews allow novel or unexpected
themes and issues to arise?
☐☐ Have the questions been designed to encour-
age group interaction and discussion?
☐☐ Is the language in the questions free of
jargon?
☐☐ Is there a strategy for dealing with silences
and for particular participants who are reluc-
☐☐ Are the questions relevant to the people tant to speak?
being interviewed?
☐☐ Is there a strategy for dealing with partici-
☐☐ Have the questions been designed to elicit pants who speak too much and “hog” the
in-depth responses so that interviewees are discussion?
not tempted to answer simply “yes” or “no”?
☐☐ Is there a strategy to follow if the discussion
☐☐ Do the questions offer a real prospect of goes off on a tangent?
seeing the world from the interviewees’
Key Points
• Interviewing in qualitative research is typically un- • The selection of focus group participants requires
structured or semi-structured. several decisions—in particular, whether to use
• Qualitative interviewing is meant to be flexible natural groups.
in order to explore the world views of research • The focus group moderator generally tries to give
participants. free rein to the discussion. However, there may be
• If an interview guide is used, it should allow some contexts in which it is necessary to ask specific
flexibility in the way questions are asked. questions.
• Qualitative interviews should be recorded and • Group interaction is an important aspect of focus
then transcribed. groups.
• Qualitative interviews can be conducted online. • The Internet offers significant opportunities for
• The focus group allows the researcher to explore gaining access to potential focus group participants.
the joint production of meaning.
268 PART III Qualitative Research
AA Online focus groups are appropriate for research Qualitative interviewing alone versus
involving sensitive issues. Identify three issues ethnography
that, because of their sensitive nature, would RR Outline the advantages and disadvantages of
be better researched with online rather than in- qualitative interviewing (without immersion in a
person focus groups, then explain why the online social setting) compared to ethnography.
technique would be more appropriate. AA Is one method more in tune with the research
needs of qualitative researchers than the other?
Feminist research and interviewing in Explain, using the topic of intimate partner vio-
q ualitative research lence to illustrate your answer.
RR Why are qualitative interviews so prominent in
feminist research?
AA Explain why focus groups may be superior to other
methods of inquiry for giving a voice to highly mar-
ginalized women.
Relevant Websites
A YouTube video of a qualitative interview uses the A YouTube video of a focus group discusses the
life history approach to examine the experiences of Trump presidency.
a Holocaust survivor. www.youtube.com/watch?v=fgSzD4S64QI
www.youtube.com/
A YouTube video demonstrates how to moderate a
watch?v=-jJ6s7ob_Dg&feature=channel
focus group.
A YouTube video of a focus group discusses the issue www.youtube.com/watch?v=xjHZsEcSqwo
of race and politics in the United States, and in par-
This YouTube video provides basic tips on qualitative
ticular Barack Obama’s political appeal.
interviewing.
www.youtube.com/watch?v=jGnePpo8JrM
www.youtube.com/watch?v=LPwO-vOVxD4
Chapter Overview
Content analysis involves the examination of various documents and texts, which may be printed, visual,
aural, or virtual. It can be quantitative, focused on coding data into predetermined categories in a sys-
tematic and easily replicable manner, or qualitative, seeking to uncover deeper meanings in the materi-
als. Although both approaches are discussed, this chapter deals primarily with the qualitative variety.
It considers:
• the kinds of research question that content analysis can answer;
• the features of documents or texts that are commonly analyzed;
• how to code (a key part of content analysis);
• semiotics and hermeneutics;
• active and passive audiences;
• conversation analysis and its roots in ethnomethodology;
• the assumptions and analytic strategies of discourse analysis, including critical
discourse analysis; and
• the advantages and disadvantages of content analysis.
Rebecca and her boyfriend, Daniel, were sitting at “You ever notice how the women look in those ads?”
home watching TV on a Saturday night. Neither was Rebecca fumed. “They’re always in bathing suits or
in a particularly good mood; both thought there shorts, even if they’re standing on a ski hill in winter. And
must be a more exciting way to spend the weekend. by some miracle they all have perfect bodies.” “Yeah,”
“Have you ever noticed that there are way more food said Daniel. “It’s a little different with the guys. A lot of
commercials—burgers, pizza, chicken—at night than them look like dorks, doing stupid things. But some guys
in the daytime?” Rebecca asked. “Are there?” Daniel re- look cool, and they’re the ones who get the girls. They’re
plied. “I know hockey games always seem to come with better-looking and taller than the dorks.” “We could
a lot of beer ads.” probably do a study comparing men and women in
beer commercials,” Rebecca said, her mood beginning active or passive, the age of the people, how attractive
to lighten. “We could even throw in a comparison with they are, overtly sexual or non-sexual behaviour—that
detergent ads. Ever see a man selling detergent?” “You kind of thing.” “That wouldn’t be too hard to do,” Reb-
could do that,” Daniel said. “You’d just have to find a way ecca answered. “Sounds like a lot more fun than watch-
to categorize the portrayals—flattering or unflattering, ing this stupid stuff.”
• animated cartoons (Thompson & Zerbinos, written to communicate with other people, diarists
1995); and presumably write for themselves. When written for
• lyrics of popular songs (e.g., to look for changes wider consumption, diaries are difficult to distin-
in the representation of women) (Marcic, 2002). guish from another kind of personal document: the
autobiography. Used with a life history or biograph-
Here we will call all such materials “documents,” ical method, diaries, letters, and autobiographies
and define them as any data source that: (whether solicited or unsolicited) can be either pri-
mary sources of data or adjuncts to other sources
• can be “read” (including visual materials such such as life story interviews. However, this chapter
as photographs); and is primarily concerned with unsolicited documents.
• was not produced specifically for the purpose The distinction between biographies and auto-
of social research. biographies can sometimes break down. Walt
Disney provides a case in point. The first biography
Documents are important because they are unobtru- of Disney, written by his daughter, Diane Disney
sive and non-reactive. The fact that the subjects don’t Miller (1956), almost certainly included information
know they are being studied removes a common from Mr Disney himself. Subsequent biographies
threat to the validity of the data. were very similar, because of the tight control exer-
In discussing the different kinds of documents cised by the Walt Disney Corporation over the pri-
used in social science, Scott (1990) distinguished be- mary materials in the Disney archive (letters, notes
tween personal and official documents and, within of meetings, and so on) out of which the biographies
the latter group, between private and state documents. were fashioned. As a consequence, even though Walt
These distinctions are used in much of the discussion Disney never wrote an autobiography in the conven-
that follows. Scott also enumerated four criteria for tional meaning of the term, he (and, later, his com-
assessing the quality of documents (1990, p. 6): pany) clearly controlled what was written about him.
In evaluating personal documents, the authentic-
• Authenticity. Is the evidence genuine and of ity criterion is obviously important. Is the purported
unquestionable origin? author of the letter or diary the real author? In the
• Credibility. Is the evidence free from error and case of autobiographies, the increasing use of “ghost”
distortion? writers has made this a standard question. But the
• Representativeness. Is the evidence typical of same kind of problem can arise with other docu-
what it is supposed to represent (for example, ments. For example, in the case of Augustus Lamb
social life at a particular time and place)? If (Box 12.1), Dickinson (1993, pp. 126–127) noted that
not, is the extent of its uniqueness known? there are “only three letters existing from Augustus
• Meaning. Is the evidence clear and com himself (which one cannot be certain were written
prehensible? in Augustus’s own hand, since the use of amanuen-
ses was not uncommon).” This raises the question of
Note that this use of “authenticity” is different whether Augustus was in fact the author of the let-
from how the term was used in Chapter 9 to assess ters, especially in the light of his apparent learning
qualitative research. difficulties.
Turning to the issue of credibility, Scott (1990)
observed that there are at least two major concerns
Personal documents with respect to personal documents: their factual ac-
curacy and whether they express the true thoughts
Diaries, letters, and autobiographies and feelings of the writer. The case of Augustus
Diaries and letters are often used by historians, but Lamb suggests that a definitive, factually accurate
they have been given less attention by other social account is at the very least problematic. Scott recom-
researchers (see Box 12.1). Whereas letters are mended a healthy skepticism regarding the sincerity
274 PART III Qualitative Research
BOX 12.1 U
sing historical personal documents:
The case of Augustus Lamb
West7megan/Dreamstime.com/GetStock
Dickinson (1993) provided an interesting account same time, she found that the people around
of the use of historical personal documents in him had difficulty coming to terms with his con-
the case of Augustus Lamb (1807–36), the only ditions, in large part because of the challenge
child of the novelist Lady Caroline Lamb and her of finding words to describe him that were con-
husband, the second Viscount Melbourne, who sistent with his high social status.
was prime minister of the UK from 1834 to 1841. The chief sources of data were “letters from
It is possible that Augustus suffered from epi- family and friends; letters to, about and (rarely)
lepsy, though he seems to have had other med- from Augustus” (1993, p. 122). Other sources in-
ical problems as well throughout his short life. cluded the record of the post-mortem examina-
Dickinson was drawn to him because of her in- tion conducted on his remains and extracts from
terest in nineteenth-century reactions to non- the diary of his resident tutor and physician for
institutionalized people with mental disabilities. the years 1817–21. Despite the many sources she
In fact, Dickinson doubted that Augustus was consulted, Dickinson still could not conclude with
mentally disabled, and suggested the somewhat certainty that she understood definitively what
milder diagnosis of “learning difficulties.” At the Augustus was like.
with which the writer reports his or her feelings. illustration is Sugiman’s (2004) suggestion that the
Famous people who are aware that their letters or reason many Japanese-Canadian women interned
diaries will be of interest to others may be very care- during the Second World War did not write down
ful about how much of themselves they reveal in their experiences was to shield their children from
their writings. Similarly, adolescent diary-keepers that painful episode in their family histories.
are often aware that their parents could “accident- Representativeness is clearly a major concern in as-
ally” read what they write. In short, what is not said sessing these materials. Since literacy rates were very
can be of great importance. A particularly poignant low in the past, letters, diaries, and autobiographies
12 | Content Analysis 275
were largely the preserve of a small class of wealthy, difficult to decipher. Also, as Scott (1990) observed,
literate people. Moreover, because boys were more letter writers tend to leave much unsaid when they
likely to receive an education than girls, the voices of take it for granted that the people they are writing to
women tended to be under-represented in documents have the same values, assumptions, and information
of that kind. Some researchers also argue that in the as they do.
past it was less socially acceptable for women than Researchers often have to search long and hard for
men to write diaries and autobiographies. Therefore new or surprising materials in diaries. However, this
such historical documents are likely to be biased in was not the case with the diaries of William Lyon
terms of their applicability to the society as a whole. Mackenzie King. When the King diaries were first
A further problem is the selective survival of released to the public in the 1970s, they caused a sen-
documents like letters. What proportion are dam- sation (see Box 12.2).
aged, lost, or thrown away? How representative are
suicide notes of the thoughts and feelings of all sui- Visual objects
cide victims? It seems that only a relatively small per- There is a growing interest in the use of visual ma-
centage of suicide victims leave notes, and some of terials, especially photographs, in social research (see
the notes that are written may be destroyed by the Box 12.3). One of the main reasons why photographs
victims’ relatives, especially if they contain accusa- are of interest to social researchers is their potential
tions against family members. to reveal important information about families.
Finally, understanding the meaning of the docu- As Scott (1990) observed, many family photos are
ments we do have is often problematic. Some pages in taken as records of ceremonial occasions such as wed-
a letter or diary may be missing or damaged, or the dings and graduations, or of recurring events such
writer may have used abbreviations or codes that are as reunions and holidays. Scott distinguished three
William Lyon Mackenzie King was prime minister in their proper historical context, the practice of
of Canada for 21 of the 27 years between 1921 and spiritualism was quite common when King was a
1948. He was an avid diarist for most of his life, young man in the late nineteenth and early twen-
recording his thoughts and feelings about both tieth centuries.) Sensationalism aside, the King
personal issues and national and international diaries are a valuable primary source of infor-
events. The contents of the diaries suggest that mation for a significant portion of Canadian and
he did not expect them to be made public. They world history.
reveal his political views in considerable detail, Nevertheless, the publication of the King diaries
but also a side of the man that shocked many does raise ethical questions. Is it a violation of pri-
people when the transcripts were released. King vacy rights to publish or quote from someone’s diary,
believed strongly in psychic phenomena and especially if the person is deceased and cannot
spiritualism—he attended séances, consulted raise objections? King’s diaries are now available
mediums, and believed that he was in communi- online at Library and Archives Canada: https://www
cation with various deceased persons, including .bac-lac.gc.ca/eng/discover/politics-government/
Franklin Delano Roosevelt and his own mother. prime-ministers/william-lyon-mackenzie-king/
He also believed that he could commune with his Pages/diaries-william-lyon-mackenzie-king.aspx
deceased dog, Pat. (To place these revelations (accessed 29 October 2018).
276 PART III Qualitative Research
Photographs can play a variety of roles in social data about the field in which the researcher is
research. They can be used in qualitative research, interested, as in the work of Blaikie (2001) and
as well as in surveys and experiments. Three Sutton (1992, see Box 12.4).
prominent roles photographs can play are the • As prompts. Photographs may be used as
following: prompts to entice people to talk about what
• As illustrations. Photographs may be used to they see in them. Both researcher-created
illustrate points and thus to enliven what might photographs and previously existing ones may
otherwise be rather dry discussions of findings. be used in this way. Sometimes research par-
This was certainly the case with some classic ticipants may volunteer their own photographs.
reports by anthropologists. For example, Riches and Dawson (1998) found
• As data. Photographs may be data in their own in their interviews with bereaved parents that
right. When produced for research purposes, many parents showed them pictures of their
they essentially become part of the research- deceased children, likely the same pictures
er’s field notes. When based on extant photo- they had shown to neighbours and friends
graphs, they may become the main source of when they were first grieving their loss.
12 | Content Analysis 277
types of home photograph: idealization (a formal issue of what is not photographed, as suggested by
portrait of a wedding party or the family in its finery); Sutton’s idea that unhappy events at Disney theme
natural portrayal (an informal snapshot capturing parks may not be photographed at all. Awareness
action as it happens, though there may be a con- of what is not photographed can reveal the “men-
trived component); and demystification (depicting tality” of the person(s) behind the camera. It’s clear
the subject in an atypical—often embarrassing— that the question of representativeness is much
situation). Scott suggested that researchers need to more fundamental than the issue of what survives,
be aware of these different types in order to avoid because it points to how the selective survival of
being deceived by the superficial appearance of photographs may be part of a reality that family
images. One must probe beneath the surface: members (or others) deliberately seek to fashion.
As in Sutton’s example, that manufactured reality
There is a great deal that photographs do not
itself may then become a focus of interest for the
tell us about their world. Hirsch [1981, p. 42]
social researcher.
argued, for example, that “The prim poses and
solemn faces which we associate with Victor-
ian photography conceal the reality of child Government documents
labour, women factory workers, whose long
The state is a source of much information of poten-
hours often brought about the neglect of their
tial significance for social researchers. It produces
infants, nannies sedating their charges with
a great deal of quantitative statistical information.
rum, and mistresses diverting middle class
For example, Bell et al. (2007) used government
fathers.” (Scott, 1990, p. 195)
voting records as well as census information to ana-
As Scott argued, this means not only that the lyze the issue of one-party dominance in Alberta.
photograph must not be taken at face value when The state is also the source of a great deal of textual
used as a research source, but that the viewer must material of potential interest, such as official reports.
have considerable knowledge of the social context For instance, in his study of the issues surrounding
in order to get its full meaning. Sutton (1992) makes synthetic bovine growth hormone, Jones (2000) used
a similar point in Box 12.4. In fact, we may wonder transcripts of a Canadian senate inquiry on the topic.
whether photographs in such situations can be of any Briefly, Monsanto, the manufacturer, lost the battle
use to a researcher. At the same time, a researcher’s to have the hormone accepted for use; health groups
interpretations should not be accepted uncritically. claimed that it was unnecessary and, worse, that it
For example, the “prim poses and solemn faces” of posed a health risk both to cows and to the humans
Victorian portraits might well “conceal” a bleak and who drank their milk. Abraham (1994) used similar
miserable reality; but is that all? Could they not also materials in his research on the role of self-interest
conceal moments of human warmth or compassion? and values in scientists’ evaluations of the safety of
Or fun? medicines, specifically the drug Opren. The author
A particular problem for the analyst of photo- described his sources as “publicly available tran-
graphs, according to Scott, is judging representa- scripts of the testimonies of scientists, including
tiveness. Photos that survive the passage of time—in many employed in the manufacture of Opren, Parlia-
archives, for example—are very unlikely to be rep- mentary debates, questions and answers in Hansard,
resentative for the simple reason that somebody and leaflets, letters, consultation papers and other
at some time decided they should be preserved. documentation disposed by the [drug regulatory
The discussion in Box 12.4 suggests that decisions authority]” (Abraham, 1994, p. 720). His research
about which photos to keep or post online may re- showed inconsistencies in the scientists’ testimonies,
flect the needs and biases of the people who make suggesting that self-interest can play an important
those decisions. Another problem relates to the role in such situations. He also used his findings to
278 PART III Qualitative Research
Sutton (1992) noted a paradox about visitors to other words, positive feelings about an experience
Disney theme parks. The Magic Kingdom is sup- may be a post hoc reconstruction, substantially
posed to be “the happiest place on Earth,” and aided by the photos that are not thrown away or
employees (“cast members”) are trained to en- deleted. Thus photographs may provide an incom-
hance visitors’ experiences. Yet it is clear that plete, somewhat distorted record.
some people do not enjoy themselves while visit- Similar observations can be made with regard
ing. Time spent waiting in lines is a particularly to pictures posted on social media. How common
common gripe for “guests” (Bryman, 1995). None- is it to see Facebook photographs of people having
theless, because people expect their visit to be unpleasant experiences on their holidays? With
momentous, they take photographs that support social media, the purpose of photo-assisted recon-
their assumption that Disney theme parks are struction seems to be expanding: now photos may
happy places. When they return home, they “dis- be used not only to create happy memories, but also
card photographs that remind them of unpleasant to present a public image of oneself as a fun-loving,
experiences and keep those that remind them of interesting person. Clearly, holiday photos repre-
pleasant experiences” (Sutton, 1992, p. 283). In sent more than an objective account of a vacation.
argue that the scientific ethos of “objectivity” has criterion requires us to consider whether the docu-
limited applicability in areas where self-interest may mentary source is biased. This is exactly the point
be a factor. of Abraham’s (1994) research: such documents can
In terms of Scott’s (1990) four criteria, such ma- be interesting precisely because of the bias they
terials can certainly be seen as authentic and as reveal. Equally, this point suggests that caution is
having meaning (in that they are clear and compre- necessary in attempting to treat them as depictions
hensible to the researcher). However, the credibility of reality.
12 | Content Analysis 279
Official documents from One of the clearest themes to emerge was the
apparently incompatible interpretations of the
private sources same events and processes among the three
Official documents available from “private sources” sub-groups within the company—senior exec-
vary widely, but a common category is company utives, HQ personnel staff, and regional per-
documents. Companies (and organizations gen- sonnel managers. . . . These documents were
erally) produce many documents, some of which not produced deliberately to distort or obscure
are in the public domain; the latter include annual events or processes being described, but their
reports, press releases, advertisements, and public effect was to do precisely this. (1994, p. 160)
relations material both in printed form and on
the Web. Other documents, such as company The perspectives that members of the different
newsletters, organizational charts, minutes of
groupings expressed in the documents reflected
meetings, memos, correspondence (internal and their positions in the organization. Consequently,
external), and manuals for new recruits, may not although the authors of the documents were able to
be accessible to the public. This kind of material confirm their authenticity, their contents could not
is often used by organizational ethnographers in be regarded as “free from error and distortion,” as
their investigations, but the difficulty of gaining Scott put it. In other words, the accounts that docu-
access to it means that many other researchers ments provide may not be completely objective. They
have to rely on public domain documents. Even have to be analyzed critically and compared with
if the researcher is an insider with the organ- other sources of data. As Forster’s study suggests, the
ization, certain private documents may remain different stances taken by the authors of documents
inaccessible. can be used to develop insights into the processes
Private documents also need to be evaluated and factors that lie behind their creation.
using Scott’s four criteria. As with the materials con- Issues of representativeness are also important.
sidered in the previous section, documents deriving Did Forster have access to a complete set of docu-
from private sources such as companies are likely to ments? It could be that some had been destroyed, or
be authentic and meaningful (in the sense of being that he was not allowed access to certain sensitive
clear and comprehensible to the researcher). How- ones. This is not to say that such documents necessar-
ever, issues of credibility and representativeness are ily exist, but a healthy skepticism is often warranted.
still likely to require scrutiny.
People who write documents generally want to
convey a particular point of view. An interesting
Mass media outputs
illustration is provided by a study of career develop- Newspapers, magazines, television programs,
ment issues in a major retail company (Forster, films, and other mass media are potential sources
1994). Forster analyzed company documentation for social scientific analysis. For example, Parnaby’s
as well as interviews and a questionnaire. Because (2003) study of how Toronto tried to deal with its
he was able to interview many of the authors of the squeegee kids examined 200 newspaper articles
documents, “both the accuracy of the documents that appeared between 1995 and 2000. Hallgrims-
and their authorship could be validated by the in- dottir et al. (2006) looked at how newspapers in
dividuals who had produced them” (1994, p. 155). Victoria, BC, characterized sex-trade workers in
In other words, the authenticity of the documents the province, while Harding (2006; 2010) analyzed
was confirmed and apparently the credibility as press coverage of Indigenous people and their insti-
well. However, Forster also said that the documents tutions. Bhuyan et al. (2017) used documents pub-
revealed divergent interpretations of key events and lished by Citizenship and Immigration Canada as
processes: well as articles from the Toronto Star, the Globe and
280 PART III Qualitative Research
Mail, and the National Post to investigate how the tone or ideological bent. Finally, although the lit-
use of language in those sources is related to the de- eral meaning of mass media outputs is often clear, it
velopment of social ideologies and the implemen- usually takes some reflection and theoretical analysis
tation of government policies on immigration. All to appreciate the broader societal impact these forms
of these studies are examples of critical discourse of communication can have.
analysis (discussed later in this chapter) in that
they examined the social and political implica-
tions of the materials they examined, an approach
Virtual outputs and the
that can be taken with virtually any medium of Internet as objects of analysis
communication. “Text” is a word that has up to this point been used
Films, television shows, and magazines have sim- sparingly in this chapter. It originally referred to a
ilar potential for research. For example, when Coté written document, but in recent years it has been
and Allahar (1994) examined magazines aimed at applied to an increasingly wide range of phenomena
adolescents, they concluded that these “teenzines” including theme parks, technologies, paintings, and
turned adolescents into uncritical consumers and buildings. Contemporary writers and researchers
diverted them from protesting against their lack of regard such “texts” as materials that can be inter-
adult privileges. preted to produce “readings.”
Authenticity is sometimes difficult to ascertain in Websites and webpages are further examples of
the case of mass media outputs. The authors are not “texts” that can be “read.” The vastness of the Inter-
always identified, as in the case of a TV news report, so net and its growing accessibility make it a valuable
it is sometimes difficult to know whether a given ac- source of documents for both quantitative and quali-
count was prepared by someone in a position to know tative data analysis. Hier (2000), for example, exam-
all the facts. Credibility is frequently an issue, and as ined a Toronto-based racial supremacy website and
the examples used in this section show, it is often the found that it allowed people access to racist ideas in
uncovering of error or distortion that is the object- a relatively anonymous way (e.g., without having to
ive of the analysis. Representativeness may not be an subscribe to a hard-copy newsletter). For that reason
issue with newspaper or magazine articles, since many he feared that it might be more successful in reaching
publications make a point of maintaining a consistent ordinary citizens than non-Internet sources would be.
12 | Content Analysis 281
Checklist
Checklist for evaluating documents
Have the following questions been answered? ☐☐ Is the document typical of its kind? If not, is it
possible to establish how atypical it is and in
☐☐ Who produced the document? what ways?
☐☐ Why was the document produced? ☐☐ Is the meaning of the document clear?
☐☐ Was the person or group who produced the ☐☐ Can the events or accounts presented in the
document in a position to write authorita-
document be corroborated?
tively about the subject?
☐☐ Are there different interpretations of the
☐☐ Is the material genuine? document from the one you offer? If so,
☐☐ Did the author have an axe to grind or a par- what are they? Have you discounted them?
ticular slant to promote? If so, why?
Jagger (1998) reported a content analysis of 1094 as the medium of choice for people whose in-
dating advertisements in four newspapers with a person activities do not provide adequate oppor-
general readership. The sample was chosen over tunities to meet a romantic partner. Since space
two four-week periods in 1996. Three research is not at a premium, such sites allow participants
questions drove the study: to give far more information than was the case
with newspaper ads, and of course the provision
• What is “the relative significance of monet-
of photographs means that verbal descriptions
ary resources and lifestyle choices as identity
of physical appearance have become far less
markers and desirable attributes for men and
important than they once were. However, many
women”?
sites still feature “dating profiles” in which people
• How do men and women vary in the ways they
have a few paragraphs to describe themselves
choose to market themselves and describe
and to outline what they would like to find in a
their preferred (or ideal) partners in terms of
partner, and the content of these profiles is some-
the body?
what similar to the older-style ad descriptions.
• To what extent are “traditional stereotypes of
There are even sites that provide tips on writing a
masculinity and femininity . . . still in operation”
good dating profile.
(1998, p. 799)?
Not surprisingly, a major concern with people
Jagger noted the tendency for considerable who engage in online dating is the misrepresenta-
percentages of both men and women to market tion of personal characteristics by potential mates
themselves in terms of their lifestyle choices (film (Hall et al., 2010). Most users want an in-person
preferences, clubs frequented, sports activities, meeting before getting involved in a romantic
etc.). She also found that women were far more relationship (Finkel et al., 2012), and not without
likely than men to stress the importance of eco- reason. Hitsch et al. (2010) compared the informa-
nomic and other resources in a preferred partner. tion provided in online dating sites with American
There was also a somewhat greater propensity for national averages, and found that both men and
women to market themselves in terms of physical women on the sites claimed higher than average
appearance. As an aside, men were just as likely heights, and that women but not men indicated
to market themselves as “slim,” suggesting that lower than average weights. Using self-reports
certain preferences in body shape were not ex- of probable misrepresentation on dating sites,
clusive to one gender. More generally, her results Hall et al. (2010) found that men were more likely
pointed to the significance of the body in iden- than women to be deceptive about their personal
tity construction for both men and women. In a assets (e.g., income), but that there was no gender
later publication, Jagger (2001) reported the find- difference in misrepresentation of relationship
ings from a qualitative content analysis of a sub- goals (e.g., interest in pursuing a serious rela-
sample of the 1094 advertisements. tionship). Women were found to be more likely to
Much has changed in the years since Jagger misrepresent their weight, while men, contrary to
did her research. The Internet dating site has expectation, were more likely to mislead about
eclipsed the classified newspaper advertisement their age.
284 PART III Qualitative Research
he argued, the “kids” were constructed “as a social meaningful relationships with family members. “I
problem requiring a law and order resolution” (2003, just keep going because I think someday my son will
p. 281). The search for pairings of keywords can be a need me,” said one farmer (Sturgeon & Morrissette,
starting point for a more in-depth analysis. 2010, p. 199). Surprisingly, only a small proportion of
farmers mentioned their spouses or life partners as
Subjects and themes the people who made their lives bearable; others such
A more interpretative approach is required to code as siblings, children, and extended family members
text in terms of subjects and themes. At this point, were more likely to be cited.
the analyst is searching not just for the obvious or
manifest content, but also for some of the underlying
or latent content as well. This occurs if the researcher
Coding
wants to probe beneath the surface to ask deeper As the foregoing has implied, coding is a crucial part
questions about what is happening. In reports on of content analysis. There are two main elements to a
crime, for example, is the victim blamed along with content analysis coding scheme: designing a coding
the accused? Is the occupation of the accused or the schedule and creating a coding manual. To illus-
victim stated? If not, is it implied in an address or a trate, imagine that you are interested in crime re-
picture? Why is it that news stories about men who ports in a local newspaper. To simplify, assume that
have been mugged sometimes mention their marital your study is limited to crimes where the victim is a
status? In seeking to answer these sorts of questions, person rather than an organization, and that it con-
the quantitative aspects of content analysis can serve siders these variables:
as a starting point for a study in which qualitative
research becomes the central focus of the endeavour. 1. nature of the offence(s)
2. gender of suspect
Value positions 3. occupation of suspect
A further level of interpretation is likely when the re- 4. age of suspect
searcher is seeking to demonstrate that the writer of a 5. gender of victim
particular text has taken a certain value position. For 6. occupation of victim
example, is a journalist who writes about crime sym- 7. age of victim
pathetic or hostile to the accused? Is all the blame put 8. depiction of victim
on the accused, with the implication that punish- 9. position of news item in the paper
ment is appropriate? Or is the focus on social condi-
tions, with the assumption that less blame should be Content analysts are normally interested in a
placed on the criminal? If there is no manifest indi- much larger number of variables, but this simple
cation of the writer’s value positions, can inferences illustration can help to get across how a coding
be made from the latent content? schedule and coding manual operate.
Another way in which content analysis can reveal
value positions is through the coding of ideolo- Coding schedule
gies, beliefs, and principles. For instance, Sturgeon The coding schedule is a form onto which the data
and Morrissette (2010) examined suicide ideation are entered (see Figure 12.1). Each of the columns in
among Manitoba farmers who called a rural crisis Figure 12.1 is a dimension (indicated by the column
line. The data were gathered from forms that were heading) to be coded. The blank cells on the coding
filled out by the crisis line counsellors (which are form are the places in which codes are to be writ-
similar in design to the coding manual depicted ten. One row is used for each media item coded. The
in Figure 12.2). One of the themes that emerged as codes can then be transferred to a computer data
codes were developed was “family salvation,” which file for analysis with a software package such as IBM
involved the belief that life is worth living if one has SPSS Statistics (SPSS).
12 | Content Analysis 285
Case Day Month Nature of Gender of Occupation Age of Gender Occupation Age of Depiction Nature of Position
number offence I suspect of suspect suspect of victim of victim victim of victim offence II of news
item
Case Day Month Nature of Gender of Occupation Age of Gender Occupation Age of Depiction Nature of Position
number offence I suspect of suspect suspect of victim of victim victim of victim offence II of news
item
001 24 11 01 1 10 26 1 16 68 2 00 2
Case Day Month Nature of Gender of Occupation Age of Gender Occupation Age of Depiction Nature Position
number offence I suspect of suspect suspect of victim of victim victim of victim of of- of news
fence II item
002 25 12 06 1 13 32 2 16 86 3 01 2
assigning codes. Sometimes these have to be qualitative in nature. Kennedy (2006), for example,
very elaborate. In the sort of content analysis in his study of the Canadian fathers’ rights move-
described in the previous section, coders gen- ment called “Fathers For Justice,” used content an-
erally have little or no discretion in deciding alysis in conjunction with participant observation
how to code the units of analysis. and semi-structured interviews to examine how
• A clear unit of analysis. For example, in the the movement developed a collective identity. The
imaginary study of the reporting of crime in content analysis was conducted on documents such
the local press, there is both a media item (for as position papers submitted to government agen-
example, one newspaper article) and a topic to cies, newsletters, pamphlets, and minutes from
be coded (one of two offences). In practice, a meetings, and was used mainly to make sense of the
researcher is interested in both but needs to data gathered through participant observation and
keep the distinction in mind. interviews. Similarly, Bell (2007) used pamphlets,
speeches, newsletters, and other texts to arrive at
To enhance the quality of a coding scheme, it an overall assessment of western Canadian separa-
is advisable to conduct a pilot study to identify tism as a neo-liberal movement. The processes by
difficulties, such as finding there is no code to which the themes are extracted in this kind of con-
cover a particular case (not exhaustive). A pilot tent analysis are often left implicit, although they
test will also help to reveal if one category of a are usually illustrated with quotations from the text
dimension includes an extremely large percentage in question.
of items. When this occurs, it may be necessary Lynch and Bogen (1997) examined core socio-
to break that category down to allow for greater logical textbooks and found that they contained
specificity. recurring themes that presented an upbeat and
As we have mentioned, the reliability of coding is scientific view of the discipline—one that, in the
a further concern. An important part of pre-testing researchers’ view, was biased and value laden. Seale
the coding scheme is examining consistency between (2002) examined newspaper reports about people
coders (inter-coder reliability) and, if time permits, with cancer. One of the phases of his analysis en-
intra-coder reliability. The process of gauging reli tailed an “ NVivo coding exercise, in which sec-
ability is more or less the same as in structured tions of text concerning themes of interest were
observation, discussed in Chapter 6. identified and retrieved” (Seale 2002, p. 109). He
was especially interested in gender differences in
Content analysis without a the representation of sufferers, and demonstrated
that stories about men were much more likely than
pre-existing coding scheme stories about women to discuss the role that the
Content analysis may be undertaken before any individual’s character played in dealing with the
specific decisions have been made regarding how disease.
the information available for analysis will be coded. Altheide (1996) outlined an approach he called
In such cases the data are searched in order to find ethnographic content analysis (ECA). He described
underlying themes, making the study primarily his approach as differing from quantitative content
288 PART III Qualitative Research
analysis in that the researcher is constantly revising • The signifier is the thing (here the yellow light)
the themes or categories distilled from the examina- that points to an underlying meaning.
tion of documents. As he put it: • The signified is the meaning to which the signi-
fier points (“caution: stop if possible”).
ECA follows a recursive and reflexive move- • The denotative meaning is the manifest or ob-
ment between concept development- vious meaning of a signifier and as such indi-
sampling-data, collection-data, coding-data, and cates its function (here to regulate traffic).
analysis-interpretation. The aim is to be sys- • A connotative meaning is one that can arise in
tematic and analytic but not rigid. Categor- conjunction with the denotative meaning (for ex-
ies and variables initially guide the study, but ample, “speed up to beat the coming red light”).
others are allowed and expected to emerge • Polysemy refers to the notion that signs can be
during the study, including an orientation to interpreted in many different ways.
constant discovery and constant comparison
of relevant situations, settings, styles, images, Semiotics seeks to uncover the hidden meanings
meanings, and nuances (1996, p. 16; emphases that reside in texts, broadly defined. Consider, by way
in original). of illustration, the curriculum vitae (CV) in academic
life. The typical academic CV contains information on
While more quantitatively oriented analysis matters such as degrees earned; previous and current
typically entails applying predefined categories to posts; administrative responsibilities and experience;
the sources, ECA allows for greater refinement of teaching experience; research experience; research
those categories as well as generation of new ones. grants acquired; and publications. It can be seen as a
For example, Parnaby’s (2003) study of squeegee system of interlocking signifiers denotatively provid-
kids included a reflexive examination of the docu- ing a summary of the individual’s experience (its sign
ments used (newspapers, magazines, government function). At the connotative level, it is an indication
documents, official reports), which was done while of an individual’s value, particularly in connec-
the process of forming and confirming theoretical tion with employability. Every CV is capable of being
concepts was still under way; it was a process of con- interpreted in different ways and is therefore poly-
tinuing discovery and comparison. In addition, ECA semic, but there is a shared code among academics
emphasizes the context within which documents are whereby certain attributes of CVs (such as lists of
generated. For instance, a study of newspaper re- publications) are seen as especially desirable and are
ports of crime would require some understanding of therefore less contentious in terms of their mean-
news organizations and the work that journalists do ing. Applicants for posts know this and design their
(Altheide, 2004). CVs to highlight the desired qualities and downplay
others, so that the CV becomes a presentation of self,
Semiotics as Miller and Morgan (1993) suggested.
Another form of predominantly qualitative content Box 12.6 outlines a semiotic study (Gottdiener,
analysis is semiotics, the science of signs. In social 1997) in which the “text” was Disneyland. The chief
research, semiotics involves analysis of the signs and strength of semiotics lies in its invitation to see beyond
symbols encountered in everyday life. It can be ap- and beneath the apparent ordinariness of everyday life
plied to documentary sources as well as many other and its manifestations. The main difficulty with semi-
kinds of data. The main terms employed in semiotics otic analysis is that, although it may offer a compelling
are as follows: exposition of some aspect of everyday life, the inter-
pretation may be somewhat arbitrary. For example,
• The sign is something that stands for some- Gottdiener’s association of “Adventureland” with col-
thing else, such as a yellow traffic light; it has onialism/imperialism might not be shared by other
two components, a signifier and a signified. analysts. However, the results of a semiotic analysis
12 | Content Analysis 289
Gottdiener (1997, pp. 108–115) proposed that the Disneyland visitor as pedestrian (walking with
Disneyland can be fruitfully explored through others in a group) and the Los Angeles resident as
semiotic analysis. He concluded that its meaning passenger (a car is necessary; danger on the con-
is based on the contrast between the alienated gested freeways). A further component of his re-
daily lives of nearby Los Angeles residents and the search entailed analysis of the connotations of the
life they experience at the theme park. He iden- different “lands” that make up the park. He sug-
tified, through this principle, several sign systems gested that each land was associated with one or
contrasting the park with the surrounding en- more signifiers of capitalism, for example:
vironment: transportation, food, clothing, shelter,
entertainment, social control, economics, politics, • Frontierland: predatory capital, conquering
and family. The first of these sign systems— • Adventureland: colonialism/imperialism
transportation—revealed a difference between • Tomorrowland: state capital for space exploration
may be no more arbitrary than those of any other in- influential in the formation of interpretivism as
terpretive approach. Indeed, it would be surprising an epistemology (see Chapter 1) and has much in
not to be struck by a sense of arbitrariness in semi- common with Weber’s notion of Verstehen. The cen-
otics, given its emphasis on the principle of polysemy. tral idea behind hermeneutics is that the analyst of a
text must seek its meanings from the perspective of
Hermeneutics its author, considering the social and historical con-
The hermeneutic approach has been used in under- text within which the text was produced.
standing and interpreting the Bible, but it has also Phillips and Brown’s (1993) hermeneutic study of
been employed in the analysis of other texts. It was the corporate image advertisements of a Canadian
290 PART III Qualitative Research
synthetic crude oil company was a “formal analysis Two approaches to the study
of the structural and conventional aspects of the text”
(1993, p. 1563). For them this meant examining the of language
texts of the advertisements in terms of their constitu- In this section two approaches to the study of lan-
ent parts and the writing conventions they reflected. guage are examined: conversation analysis (CA)
They also relied on a large database of magazine and and discourse analysis (DA). While CA and DA do
newspaper articles relating to the company for addi- not exhaust the range of possibilities for study-
tional documentary materials. They showed how ing language, they do represent two of the most
corporate image advertisements were used to mobil- prominent approaches, and each includes both
ize support for company activities from government quantitative and qualitative aspects. Both have de-
and the public, and to ward off the environmental veloped a technical vocabulary and a set of tech-
legislation that threatened them. niques. This section outlines some of their basic
elements and draws attention to their contrasting
features.
Readers and audiences—
active or passive? Conversation analysis
Audience reception is a prominent area of inquiry The roots of CA lie in ethnomethodology, a socio-
in media and cultural studies. The key question is logical approach to communication focusing on
whether audiences/readers are active interpreters of the “practical, common-sense reasoning” that
what they see or hear. Do they passively accept the people use in their everyday lives. It includes no-
meanings that authors or designers infuse into their tions of cause and effect (“if I do this, then that
texts, as in the oil company advertisements just de- will happen”) and the generalizations that allow
scribed, or do they resist those meanings and strive people to perform everyday tasks. This process of
for independent readings? Do they arrive at a middle reasoning and communication is presented as a way
point that incorporates both passive and active ele- in which social order is created. Social order is not
ments? Much of the research on this issue suggests seen as a pre-existing force constraining individ-
that audiences frequently come up with readings ual action, but as something that is worked at and
different from those intended by authors (see Fenton achieved by people through interaction. Garfin-
et al., 1998, for a summary of some of this research). kel, one of the pioneers of conversation analysis,
Although the idea of the “active audience” has claimed that the role of sociology is not to uncover
not gone unchallenged (see, for example, McGuigan, “objective” social facts as Durkheim suggested, but
1992), it is so well supported that many of the inter- to see them as an accomplishment, as the eventually
pretations of texts offered by social scientists have taken-for-granted patterns established through the
been questioned. This suggests caution in reading activities of ordinary people going about their daily
Giulianotti’s (1997) study of “fanzines” or Hier’s lives (Garfinkel, 1967, p. vii).
(2002) examination of city council minutes to Two ideas are particularly central to ethnometh-
understand the rave issue. Would other social sci- odology and find clear expression in CA: indexi-
entists arrive at the same interpretations? Do their cality and reflexivity. The former suggests that the
interpretations match those of the original readers meanings of words or utterances, including pauses
or audiences? The social researcher is always putting and sounds, depend on the context in which they
a personal “spin” on the texts analyzed. The same is are used. Reflexivity means that talk is not a “mere”
true of all social science data: the conclusions derived representation of the social world, standing for
from questionnaire or ethnographic data reflect the something else, but is itself a reality. In these ways,
author’s interpretations, not the complete set of pos- ethnomethodology fits squarely with two aspects
sible interpretations. The point is that close scrutiny of qualitative research: a preference for a context-
and critical thinking are always required. ual understanding of action (see Chapter 9) and an
12 | Content Analysis 291
ontological position associated with construction- naturalistic analysis without prior theoretical com-
ism (see Chapter 1). mitments) are married to traits associated with
In the years following its introduction, ethnometh- quantitative research.
odological research sought to conduct fine-grained However, the emphasis on context in CA is some-
analyses of the sequences of interaction revealed in what at variance with contextual understanding as it
conversations recorded in naturally occurring con- is normally thought of in qualitative research. For CA
texts. As such, CA is a multifaceted approach, part practitioners, context refers to the specific here-and-
theory, part method of data acquisition, part method now context of immediately preceding talk, whereas
of analysis. The predilection for the analysis of talk for most qualitative researchers it has a much wider
gleaned from naturally occurring situations sug- focus, encompassing phenomena such as the broader
gests that CA fits well with another preoccupation culture of the group within which the social action
among qualitative researchers: a commitment to occurs, including their values, beliefs, and typical
naturalism (see Chapter 9). modes of behaviour. This is precisely the kind of
Conversation analysts have developed a variety attribution that CA practitioners seek to avoid. To
of ways to study talk. Psathas (1995, p. 1) described import elements that are not specifically grounded
them as “rigorous, systematic procedures” that can in the here and now of what has just been said during
“provide reproducible results.” Such a framework a conversation is to risk imposing an understanding
brings to mind the procedures used to generate that is not grounded in the participants’ immediate
valid, reliable, and replicable findings in quanti- concerns. It is no wonder, therefore, that writers like
tative research. It is not surprising, therefore, that Gubrium and Holstein (1997) treated CA as a sep-
CA is sometimes described as having a positivist arate tradition within qualitative research, whereas
orientation. Thus some features that are broadly Silverman (1993) found it difficult to fit CA into
in tune with qualitative research (contextual and broad descriptions of qualitative research.
Analysts often begin practising conversational an- The transcript in Box 12.7 includes some basic sym-
alysis when they notice something significant about bols used by conversation analysts:
the way a speaker says something, and this recogni-
We:ll A colon indicates that the sound that
tion generates interest in the function that the turn
occurs directly before the colon is prolonged.
of phrase serves. To illustrate, Clayman and Gill
More than one colon means further prolonga-
(2004) referred to the way a child will often begin
tion (e.g., Ye: : : :s).
a conversation with an adult by saying “You know
what, Daddy [or whoever]?” The question generally .hh A row of h’s preceded by a dot indicate
produces a “What?” reply that allows the child to an intake of breath. If no dot is present, it
find a slot in a sequence of communications, or to means breathing out. The more h’s, the longer
inaugurate such a sequence. The use of a question, the breath.
rather than a declarative statement, may reflect the
(0.8) A figure in parentheses indicates a
child’s lesser power.
period of silence, usually measured in tenths
Once such a focus has been identified, conversa-
of a second. Thus (0.8) indicates eight-tenths of
tion analysts typically follow certain basic assump-
a second of silence.
tions. Heritage (1984; 1987) proposed three:
you and knowing An underline indicates an
• Talk is structured. Talk comprises patterns, emphasis in the speaker’s talk.
and participants are implicitly aware of the
you- you A hyphen means the speaker was
rules that underpin these patterns. As a result,
interrupted or stopped speaking.
conversation analysts do not attempt to infer
the motivations of speakers from what they (.) Indicates a very slight pause.
say, or to ascribe their talk to purely personal
characteristics. Such information is unneces- The attention to detail in the sequence of talk
sary, since the conversation analyst is oriented in Box 12.7 is striking, but what is significant in
to the underlying structures that are revealed it? Silverman (1994) drew two main inferences.
in pauses, emphases, questions followed by an- First, P initially tries to deflect any suggestion that
swers, and so on. there may be a special reason why she needs an
• Talk is forged contextually. Talk must be ana- HIV test. As a result, the disclosure that she has
lyzed in terms of its context and understood in been engaging in potentially risky behaviour is de-
terms of the talk that has preceded it. layed. Second, P’s use of “you” depersonalizes her
• Analysis is grounded in data. Conversation behaviour. Silverman argued that sequences like
analysts shun prior theoretical schemes and these show how “people receiving HIV counselling
argue that the characteristics of talk and social skilfully manage their talk about delicate topics”
order must be derived from the data. (1994, p. 75). The hesitations are designed by pa-
tients to establish that issues like these are not the
In doing a project based on CA, it is important not subject of normal conversation. The rather general
to collect too much data. The real work of CA is the replies to questions indicate that the speaker is not
painstaking analysis that its underlying theoretical the kind of person who immediately launches into
stance requires. It may be that just one or two por- a discussion about difficult sexual matters with a
tions of transcribed text will be sufficient to answer stranger. As an aside, Silverman suggested that
the research questions. P’s hesitancy and depersonalization are minimal.
12 | Content Analysis 293
regimens they prescribe. This discourse is much variety of actions are performed during the first few
more than language: it is part of the social world of moments:
mental illness.
Several different approaches fall under the DA • The callers spell out the details of their
heading (Potter, 1997). The version discussed here concerns.
is one associated with writers such as Potter (1997), • The callers seek to establish, in collabora-
Potter and Wetherell (1994), and Billig (1991). This tion with the “experts” on the line (the child
version of DA exhibits two distinctive features at the protection officers) that the incidents they
level of epistemology and ontology (Potter, 1997): are reporting do in fact require some kind of
intervention.
• It is generally anti-realist in that it denies the • The callers convey the idea that the activity
existence of an external reality awaiting a de- they are describing is serious and that they are
finitive portrayal by the researcher. It therefore concerned about it, but not so concerned that
disavows the notion that researchers can arrive the police should be contacted immediately.
at accurate and objective accounts of the social • The child protection officers show a willing-
world. ness to treat the reports as serious, without
• It is constructionist in that it gives priority to making judgments as to their actual truth or
the version of reality propounded by mem- seriousness.
bers of the social setting being investigated.
More specifically, it recognizes that discourse Analysis of these brief moments of conversa-
analysis entails a selection from many pos- tion shows that the flow of discourse achieves a
sible interpretations of a given situation. For number of objectives for both parties and in that
example, is a person who speaks to herself sense constitutes action. On the other hand, as
dangerous or harmless? Is she more in need of Gill (1996) suggested, what is said is always a way
treatment or social support? In the process of of not saying something else. Either way, discourse
answering such questions, a particular reality can be seen as providing a solution to a problem
is constructed. (Widdicombe, 1993).
DA shares with CA a preference for locating con-
Discourse is not simply a neutral device for im- textual understanding in terms of the situational
parting meaning. After all, there are many things that specifics of talk. As Potter (1997, p. 158) put it, dis-
people seek to accomplish when they talk or write. course analysts prefer “to see things as things that
DA is concerned with the strategies they employ in are worked up, attended to, and made relevant in
order to create different kinds of effect. This is illus- interaction rather than being external determin-
trated in three basic questions that DA asks: ants.” However, DA practitioners are less wedded to
this principle than are conversation analysts.
• What is this discourse doing? Discourse analysts resist the idea of codifying
• How is it constructed to make this happen? their procedures and argue that such codification
• What resources are available to perform this is probably impossible. Potter (1997, pp. 147–148)
activity? (Potter, 2004, p. 609) described his work as “a craft skill, more like bike
riding or chicken sexing than following [a] recipe.”
This action orientation is revealed in a study of Gill (2000) suggested adopting a posture of “skep-
the first few moments of telephone calls to the help- tical reading.” This means searching for a purpose
line of the UK’s National Society for the Prevention lurking behind the way things are said or presented.
of Cruelty to Children. Through analysis of these call Gill proposed that DA can be usefully thought of in
openings, Potter and Hepburn (2004) showed that a terms of four main themes, outlined in Box 12.9.
296 PART III Qualitative Research
Gill (2000) drew attention to four prominent 3. Discourse is a form of action. As Gill put it,
themes in DA : language is considered “a practice in its own
right” (2000, p. 175). Language is a way of ac-
1. Discourse is a topic. In other words, discourse
complishing acts, such as attributing blame,
is a focus of inquiry itself and not just a means
presenting oneself in a particular way, or get-
of gaining access to the aspects of social re-
ting an argument across. Moreover, a person’s
ality that lie behind it. This idea contrasts
discourse is affected by the situation he or she
with the view traditionally taken by research
is in. For instance, the reasons given for want-
interviewers, in which language simply reveals
ing a job may vary depending on whether the
what people think, or serves as a mechanism
job seeker is talking to a prospective employer,
for learning about their behaviour and the rea-
a family member, or a friend.
sons for it.
4. Discourse is rhetorically organized. DA practi-
2. Language is constructive. Discourse is a way of
tioners recognize that discourse is concerned
constituting a particular view of social reality.
with “establishing one version of the world in
Moreover, in order to create that view, choices
the face of competing versions” (Gill, 2000,
have to be made regarding the most appropri-
p. 176). In other words, the version of events that
ate way of presenting it, and these choices re-
a person presents is intended to persuade others.
flect the disposition of the person concerned.
Producing facts with the very large number of cases diagnosed each
year. Instead, they could have pointed to the abso-
In this section, the emphasis is on the strategies lute number of people who are cured (2500), but the
employed to convey allegedly factual knowledge. impact would have been less dramatic. It’s also worth
Among them is quantification rhetoric: the various noting that the phrase “a quarter of a million” not
ways in which statements involving numbers or only exaggerates the number of new cases (adding
quantities can be made to either support or refute 7000 to the actual 243,000) but also sounds signifi-
arguments. The interest in this issue lies partly in cantly larger.
the importance of quantification in everyday life, but
also in the tendency of many social scientists to use Reading the detail
this strategy themselves (John, 1992). In their analy- Discourse analysts, like conversation analysts, attend
sis of data such as the portions of transcript cited in to the details of discourse. For example, Potter and
Box 12.10, Potter et al. (1991) and Potter and Weth- Wetherell (1994) suggested that the description of
erell (1994) show how these sorts of devices are used. the three “curable cancers” as “amongst the rarest
cancers” was deployed to make the point that these
Using variation as a lever are atypical cancers, and that it would be unwise to
The authors drew attention to the phrase “1 per cent generalize from them to all cancers.
of a quarter of a million” because it incorporates two
different types of quantitative expression: a relative Looking for rhetorical detail
expression (a percentage) and an absolute frequency Attention to rhetorical detail entails sensitivity to the
(quarter of a million). The difference is important, way arguments are constructed. During the editing
because it allowed the program makers to empha- of the film, the program makers’ discourse suggested
size the very low cure rate (just 1 per cent) compared they were looking for ways to present a convincing
12 | Content Analysis 297
BOX 12.10 D
iscourse analysis in action: Producing facts through
quantification rhetoric
The study of the television program Cancer: Your had cast doubt on whether their research, much
Money or Your Life (Potter et al., 1991; Potter of it funded by charities, resulted in successful
& Wetherell, 1994) used a variety of sources, treatment:
including:
Commentary: The message from these sci-
• a video recording of the program; entists is clear—exactly like the public—
• the observations of one of the people they hope their basic research will lead to
making the program (who acted as a par- cures in the future—although at the moment
ticipant observer while it was being made); they can’t say how this will happen. In the
• drafts of the script; meantime, their aim is to increase scientific
• recordings of editing sessions; knowledge on a broad front and they’re cer-
• entire interviews (of people such as cancer tainly achieving this. But do their results jus-
research specialists and heads of charities) tify them getting so much of the money that
conducted for the program; and has been given to help fight cancer? When
• research interviews with some of the latter faced with this challenge the first thing the
people as well as some of the people in- charities point to is the small number of can-
volved in making the program. cers which are now effectively curable.
With regard to the coding process, the authors
reported that they [on screen: Dr. Nigel Kemp Cancer Research
Campaign]
made a list of about a dozen keywords and
phrases that related to the sequence— Kemp: The outlook for individuals suffering
percentage, cure rates, death rates, 1 per from a number of types of cancer has been
cent, etc.—and then ran through each of totally revolutionized. I mean for example—
the interview and interaction files, looking children suffering from acute leukemia—in
for them with a standard word-processor. . . . the old days if they lived six months they
Whenever we got a “hit” we would read the were lucky—now more than half the chil-
surrounding text to see if it had relevance to dren with it are cured. And the same applies
our target sequence. When it did we would to a number of other cancers—Hodgkin’s
copy it across to an already opened coding Disease in young people and testicular tu-
file . . . noting the transcript page numbers mours in young men. (Potter & Wetherell,
at the same time. If we were not sure if the 1994, pp. 52–53)
sequence was relevant we copied it anyway,
At this point a table showing the annual inci-
for, unlike the sorts of coding that take place
dence of 34 types of cancer begins to scroll on
in traditional content analysis, the coding
the screen. The total incidence is 243,000, and the
is not the analysis itself but a preliminary
individual incidences range from placenta (20)
to make the task of analysis manageable.
to lung (41,400). The three forms of cancer men-
(Potter & Wetherell, 1994, p. 52)
tioned by Kemp and their levels of incidence are
Following is a sequence used in the research. highlighted in yellow: childhood leukemia (350),
It occurred roughly halfway through the program, testis (1000), and Hodgkin’s disease (1400). The
following interviews with cancer scientists who program continues while the table is scrolling.
continued
298 PART III Qualitative Research
Commentary: But those three curable types sometimes—the outlook is very bad and ob-
are amongst the rarest cancers—they rep- viously one is frustrated by the s[low], rela-
resent around 1 per cent of a quarter of a tively slow rate of progress on the one hand
million cases of cancers diagnosed each but equally I think there are a lot of real
year. Most deaths are caused by a small opportunities and positive signs that advan-
number of very common cancers. ces can be made—even in the more intract-
able cancers. (Potter & Wetherell, 1994, p. 53)
Kemp: We are well aware of the fact that
[pause] once people develop lung cancer
Source: Excerpts from Analyzing Qualitative Data © 1994 Routledge.
or stomach cancer or cancer of the bowel Reproduced by permission of Taylor & Francis Books UK.
argument to show that cancer remains largely in- historical political struggles are among the issues
curable in spite of the money spent on it. The pro- raised by these researchers (Blommaert & Bulcaen,
gram makers very consciously devised the strategy 2000). They also bring to their work a commit-
outlined in the discussion of “using variation as a ment to social change and the empowerment of the
lever” of playing down the numerical significance of oppressed.
the cancers that are amenable to treatment. Another Harding, for instance, has examined how dis-
element of their strategy was to employ a tactic course has been used to suppress the aspirations
that Potter et al. (1991) called a “preformulation,” of Indigenous peoples in Canada since the 1860s.
the discounting of a possible counterargument, as In one study (2006), he compared Canadian news-
when the commentary says: “When faced with this paper coverage of Indigenous peoples’ activities in
challenge the first thing the charities point to is the the mid-nineteenth century with coverage from the
small number of cancers which are now effectively late twentieth century. He concluded that although
curable.” instances of overt racism in Canadian papers de-
clined in this period, the press continued to serve
Critical discourse analysis dominant interests by portraying Indigenous people
There may be important political conflicts and as a threat to white power and dominance. One way
social justice issues underlying the texts examined in which this occurred was through the framing
by discourse analysts. For example, the makers of of the text with a binary opposition of “us” versus
the television program analyzed by Potter and col- “them.” An article published in the 1860s stated,
leagues in Box 12.10 infer that despite lavish fund- “To allow these [Indigenous] people all the privil-
ing, cancer researchers have had limited success eges of others [with regard to homesteading] would
in producing cures, which one could take to mean be to throw the whole Colony [of British Columbia]
that scarce societal resources are being misallo- into confusion. Just imagine our 80,000 Indians . . .
cated. While misallocation of funds is certainly a being allowed to locate land wherever they please”
political issue, notions of power and exploitation (Harding, 2006, p. 213). Another binary opposition
are not confronted directly or in any depth in their evident in press coverage was “reason” versus “emo-
research. tion.” Press coverage in the more recent era depicted
Those who conduct critical discourse analysis, a non-Indigenous judge presiding over an Indigen-
by contrast, are much more explicit in exposing the ous land claims case as someone who would “rule”
political nature of the texts they examine. The effect and offer “insight” by making his “decision” using
of power hierarchies, structural inequalities, and “common sense” and a “firm” interpretation of “the
12 | Content Analysis 299
rule of law” (217). Indigenous individuals, however, caseloads, rather than personal incompetence, were
on hearing of his ruling, are described as dealing offered as explanations.
with the death of their “dream,” and as people who
“argue,” cry “tears,” “vow to fight,” and use tactics of Overview
“confrontation” (217). DA draws on insights from CA, particularly when
Similarly, Harding (2010) has shown how in analyzing strings of talk. The CA injunction to focus
recent media accounts of the death of a young girl on the talk itself and the ways in which intersubject-
in the care of an Indigenous child welfare agency, ive meaning is arrived at in sequences of talk is also
the organization’s Indigeneity was made “hyper- incorporated into DA . However, DA practitioners
visible” and its social workers and officials were por- break with their CA counterparts when they specu-
trayed as incompetent. Yet when a child in the care late on motives, which is quite evident in the critical
of a non-Indigenous agency died, the ethnicity of its discourse analysis discussed above. It is precisely this
social workers and administrators was largely invis- practice of speculating on things that are not directly
ible, and systemic factors such as poverty and high discernible in the sequences of talk that conversation
300 PART III Qualitative Research
analysts reject, as when Schegloff (1997, p. 183) wrote have been at least partly influenced by antici-
of DA: “Discourse is too often made subservient to pation of such scrutiny and thus may contain
contexts not of its participants’ making, but of its some reactive error.
analysts’ insistence.” For their part, discourse ana- • Content analysis is a highly flexible method,
lysts object to the restrictions that CA imposes, be- applicable to several different kinds of un-
cause those restrictions cause conversation analysts structured information. Although it is pri-
to “rarely raise their eyes from the next turn in the marily applied to mass media outputs, it has a
conversation, [which] is not an entire conversation much broader applicability.
or sizeable slice of social life but usually a tiny frag- • Content analysis can permit the study of social
ment” (Wetherell, 1998, p. 402). groups that are difficult to access. For example,
Writing from a critical realist position, Reed much of what social scientists know about the
(2000) argued that discourses should be exam- social backgrounds of elite groups, such as
ined in relation to social structures such as the company directors and top military personnel,
power relationships that create the discourses. comes from content analyses of publications
This approach, which is exemplified in the work of such as Who’s Who and the business pages of
critical discourse analysts, shows how discourses newspapers.
work through existing structures, and expands
discourse from a self-referential concept in which
nothing of significance exists outside itself into a
Disadvantages of content
“generative mechanism.” That position is perhaps analysis
closer to the classic concerns of the social sciences Like all research techniques, content analysis suffers
than the anti-realist stance taken by some CA and from certain limitations:
DA researchers.
• A content analysis can only be as good as the
documents it explores. Recall that Scott (1990)
Advantages of content recommended assessing documents in terms
analysis of criteria such as authenticity (the document
Content analysis has several advantages: is what it purports to be); credibility (there are
no grounds for thinking that its contents have
• In its most quantitative form, it is a very trans- been distorted in any way); and representative-
parent research method, making replication ness (the specific documents examined are rep-
relatively easy. It is because of this transpar- resentative of all possible relevant documents).
ency that content analysis is often said to be an • Even in quantitative content analysis, it’s
“objective” method of analysis. almost impossible to devise coding manuals
• It allows for longitudinal analysis. For ex- that do not require some interpretation by the
ample, a crime study can be expanded to coder. It seems unlikely that there is ever a per-
examine changes in newspaper crime report- fect correspondence of interpretation between
ing over two different time periods. different coders.
• Content analysis is an unobtrusive, non- • Problems of interpretation are especially likely
reactive method. Newspaper articles and tele- to arise when the aim is to identify latent
vision scripts are generally not written with meanings (as opposed to the more readily ap-
the expectation that a content analysis might parent manifest meanings). For example, in
one day be carried out on them. On the other searching for traditional markers of masculin-
hand, if the object of analysis is an interview ity and femininity (see Box 12.5), the potential
transcript or ethnography, its content may for conflicting interpretations is magnified.
12 | Content Analysis 301
A related distinction is sometimes made be- media, but it was by far the most frequently
tween mechanical analysis (in particular, inferred discipline (not directly mentioned,
counting certain words) and analysis that but apparently the discipline under discus-
emphasizes themes in the text, which entails sion). Again, however, the reasons behind
a higher level of abstraction and a correspond- this phenomenon could only be a matter for
ingly greater risk of measurement invalidity. speculation (Fenton et al. 1998).
• Content analysis may not be helpful in an- • Some content analysis studies are accused of
swering “why?” questions. For example, being atheoretical, and it’s easy to see why. The
Jagger’s analysis of dating advertisements
emphasis on measurement in content analysis
clearly showed that both men and women put can easily lead researchers to focus on what is
less emphasis on the “attractiveness, shape measurable rather than what is theoretically
and size” they desired in a partner than they significant or important. However, content
did on their own appearance (1998, p. 807; analysis is not necessarily atheoretical. Jagger
see Box 12.5). But further research would be (1998), for instance, placed her findings on
needed to determine why this was the case. dating advertisements in the context of cur-
Similarly, Fenton et al. (1998) found that rent ideas about consumerism and the body.
sociology was only the fourth-most common And Harding’s (2006; 2010) content analysis
discipline explicitly referred to (after psychol- of media portrayals of Canadian Indigenous
ogy, economics, and social policy) when social people was underpinned by theoretical ideas
science research was reported in the mass pertaining to racism and ethnic conflict.
Key Points
• Depending on how it is done, content analysis can • The design of a coding schedule and preparation
be mainly quantitative or qualitative in nature. of a coding manual are crucial steps in content an-
• It’s important to have clear research questions and alysis. In quantitative content analysis, this is done
to be explicit about what is to be analyzed. before the materials are examined.
• Documents (as the term is used here) provide • Content analysis can be particularly challenging
many different kinds of information and can take when it is used to search for latent meanings and
the form of personal documents, official docu- themes.
ments from state or private sources, and output • Semiotics and hermeneutics are qualitative ap-
from the mass media. proaches to content analysis.
• Such materials can be used for both quantitative • There is disagreement over whether readers of
and qualitative research. documents are active or passive consumers of the
• Documents may be in printed, visual, digital, or messages they receive.
any other retrievable format. • Both conversation analysis (CA) and discourse
• Many researchers believe that almost anything analysis (DA) approaches make language itself a
can be “read” as a “text.” focus of research.
• Criteria for evaluating the quality of documents • CA is a systematic approach based on the principle
include authenticity, credibility, representative- that talk is structured by certain rules.
ness, and meaning; some may be more relevant • Practitioners of CA avoid making inferences
than others, depending on the nature of the about talk that are not grounded in its immediate
document. context.
302 PART III Qualitative Research
• DA shares many features with CA but comes in • Critical discourse analysis situates texts in the con-
several different versions and can be applied to a text of larger social structures such as power hier-
wider variety of phenomena. archies and systems of inequality.
• DA sees discourse as a means of conveying mean- • As with all research methods, there are advan-
ing and generally relates meaning in talk to con- tages and disadvantages to content analysis.
textual factors.
a. searches the Internet for newspaper accounts a. decides on at least three themes in the clip
of the event, and selects three for analysis; that are noteworthy, as well as other aspects
b. decides on at least three themes in the ac- of it that seem relevant—for example, under-
counts that are noteworthy, as well as other lying racist assumptions, gender stereotypes,
aspects of the stories that seem relevant—for jingoism;
example, number of victims, type of weapons b. produces both a coding schedule and a coding
used; manual to do a content analysis of the clip; the
c. produces both a coding schedule and a coding former should be created in such a way that the
manual to do a content analysis of the three number of instances of a particular theme can
accounts; be recorded;
d. creates a numerical summary of the three arti- c. creates a numerical summary of the clip based
cles based on the data gathered in the coding on the data gathered in the coding schedule;
schedule; and and
e. discusses salient interpretations that arose out d. discusses salient interpretations that arose out
of the content analysis, and if possible, relates of the content analysis, and if possible, relates
them to hypotheses or theories (either original them to hypotheses or theories (either original
or already established in the literature) that or already established in the literature) that
may follow from the content analysis. may follow from the content analysis.
The class as a whole then reconvenes. Each The class as a whole then reconvenes. Each
group presents the results of its analysis to the group presents the results of its analysis to the
class, which is encouraged to ask questions and class, which is encouraged to ask questions and
provide critical commentary. provide critical commentary.
Relevant Websites
Information on ethnomethodology and conversation
analysis can be found at the EMCA wiki page.
http://emcawiki.net/Main_Page
304 PART III Qualitative Research
The journal Discourse & Society is an international An overview of the software available to do content an-
journal that features articles on discourse analysis alysis can be found at the Audience Dialogue website.
(you may have to go through your institution’s library www.audiencedialogue.net/soft-cont.html
sever to access the material).
http://das.sagepub.com The Discourse and Rhetoric Group at Loughborough
University provides a tutorial on conversation analysis.
A YouTube video describes how to do qualitative http://ca-tutorials.lboro.ac.uk/sitemenu.htm
content analysis.
www.youtube.com/watch?v=BhQX-zKultw (Websites accessed on 30 October 2018)
Chapter Overview
Analyzing qualitative data from interviews, participant observation, content analysis, or any other qual-
itative source is never straightforward. For one thing, such data typically take the form of a large body
of unstructured textual material. Moreover, whereas quantitative research has clear-cut rules for data
analysis, this is not the case in qualitative research. This chapter will examine some general approaches
to qualitative data management. The most significant recent development in this area is the use of com-
puter software for coding: the main feature of qualitative data analysis. This software, often referred to
as computer-assisted qualitative data analysis software (CAQDAS), eliminates many, if not most, of the
clerical tasks associated with the manual coding and retrieving of data. Although there is no industry
leader among the different programs available for qualitative data analysis, this chapter introduces a
relatively new entrant that is having a big impact: QSR NVivo (discussed in detail in the Appendix).
This chapter will explore:
• grounded theory as a general strategy of qualitative data analysis; its main features, processes,
and outcomes are presented along with some criticisms;
• coding as a key process in grounded theory and in qualitative data analysis more generally; there
is an extended discussion of what it entails and some of its limitations;
• some of the debates about the desirability of using CAQDAS;
• the criticism that coding tends to fragment data; and
• the idea of narrative analysis, an approach that is gaining a following because it reduces that
fragmentation.
Einstein once famously said that the most incompre- science, is all about finding those patterns and then ex-
hensible thing about the universe is that it is com plaining them.
prehensible. What is it that allows us to understand how Qualitative social research may seem a long way
the universe works? The simple fact that it shows cer- away from Einsteinian physics, but in essence the “con-
tain regularities or patterns. Science, including social cepts” and “categories” that qualitative researchers
look for as they navigate through vast seas of data are we use the right methodology. This chapter explores
simply patterns. This suggests that the social world of how qualitative researchers make sense of the data
meanings, perceptions, and feelings is not chaotic, but they gather, including how some of them use it to gen-
has a coherence to it that we can know and describe if erate theory.
One further problem with analytic induction is concept or category. Glaser and Strauss (1967)
that (unlike grounded theory) it does not provide advised writing a memo (see p. 311) on each
useful guidelines on the number of cases required concept/category after a few phenomena have
before the absence of negative cases can be assumed been coded. Comparison also entails sensitiv-
and the validity of the hypothetical explanation ity to differences between emerging concepts/
(whether reformulated or not) can be confirmed. categories.
• Theoretical saturation: the point at which
Grounded theory there is nothing to be gained by further re-
Grounded theory is defined as “theory that was viewing of old data or collection of new in-
derived from data, systematically gathered and formation to see how it fits with emerging
analyzed through the research process” (Strauss & concepts or categories; new data are no longer
Corbin, 1998, p. 12). Its two central features are its illuminating.
development of theory out of data, and an iterative or
recursive approach in which data collection and an- Coding in grounded theory
alysis proceed in tandem, repeatedly referring back Coding in grounded theory entails reviewing tran-
to each other. scripts and/or field notes and giving labels (names)
Grounded theory is by far the most widely used to items that seem to share a similar theme, to be of
framework for analyzing qualitative data. Yet there potential theoretical significance, and/or to be par-
is considerable controversy about what it entails ticularly salient within the social worlds under inves-
(Charmaz, 2000). For example, it is vague on the tigation. As Charmaz (1983, p. 186) put it, “Codes . . .
difference between concepts and categories. As we serve as shorthand devices to label, separate, compile,
will see later in this chapter, it seems that the former and organize data” (emphases in original). This coding
term is increasingly being replaced by the latter, and process differs from the coding used in the analysis of
this inconsistent use of key terms is not helpful either quantitative data (e.g., social survey data) in two ways.
to practitioners of the craft or to people trying to First, whereas quantitative analysis often involves test-
understand the overall process. ing a pre-existing theory, grounded theory uses coding
Against such a background, defining grounded as an important first step in the generation of theory.
theory is not easy. As well, grounded theory cannot Second, whereas in quantitative analysis coding is
be described here in all its facets; we will simply out- more or less a way of managing data that have already
line its main features. been at least broadly categorized, in grounded theory
(and other approaches to qualitative data analysis),
The basic features of grounded theory coding is more tentative and fluid. The data are treated
Some of the basic features of grounded theory have as potential indicators of concepts/categories, and the
been introduced in earlier chapters: indicators are repeatedly compared to see which con-
cepts/categories they fit with best. Ad hoc compromises
• Coding: the key process in grounded theory, may have to be made when more than one researcher is
whereby data are broken down into component doing the coding (Tastsoglou & Miedema, 2003).
parts and given names. It begins soon after the Strauss and Corbin distinguished three types of
initial collection of data. According to Char- coding:
maz: “Unlike quantitative research that re-
quires data to fit into preconceived standardized • Open coding: “the process of breaking down,
codes, the researcher’s interpretations of data examining, comparing, conceptualizing, and
shape his or her emergent codes in grounded categorizing data” (Strauss & Corbin, 1990,
theory” (2000, p. 515, emphasis in original). p. 61); this process stays very close to the data
• Constant comparison: continuous comparison and yields the concepts that are later grouped
of new and existing data within a particular together to form categories. Thus concepts
308 PART III Qualitative Research
such as anger, jealousy, or affection could form frames the account. In the case of hardship and
the category “emotion.” loss, the core category might be “adaptation.”
• Axial coding: “a set of procedures whereby data
are put back together in new ways after open The three types of coding are really different
coding, by making connections between cat- levels of the same process, each relating to a differ-
egories” (Strauss & Corbin, 1990, p. 96). This is ent point in the process of elaborating concepts and
done by linking codes to contexts, consequences, categories. Not all grounded theory practitioners
patterns of interaction, and apparent causes. operate with this threefold distinction; the notion
Thus the category of emotion, above, could be of axial coding has been especially controversial
linked to the contexts in which it is expressed: because it is sometimes perceived as closing off the
for example, contexts of hardship or loss. coding process too quickly. Charmaz (2004) pre-
• Selective coding: “the procedure of selecting the ferred to distinguish between open or initial coding
core category, systematically relating it to other and selective or focused coding. The former tends to
categories, validating those relationships, and be very detailed and may produce as much as one
filling in categories that need further refine- code per line of text. It is crucial at this stage of the
ment and development” (Strauss & Corbin, research to be open-minded and to generate as many
1990, p. 116). A core category is the focus around new ideas, and hence codes, as necessary to organize
which other categories are integrated—what the data. Selective or focused coding emphasizes the
Strauss and Corbin called the storyline that codes that (a) appear most frequently and (b) seem
13 | Qualitative Data Analysis 309
most revealing. Combining the initial codes gener- Overall, then, the process of coding begins with
ates new codes. The data are then re-explored and the data themselves and then gradually moves to
re-evaluated in terms of the selected codes. Pidgeon more selective and abstract ways of conceptualizing
and Henwood (2004) provided a useful example in the phenomenon of interest.
their study of 60 mother–daughter relationships:
Outcomes of grounded theory
The initial coding led to the development of a The following are the products of different phases of
long and varied, but highly unwieldy, list of in grounded theory.
stances under the label “Relational Closeness.” . . .
[A] closer reading and comparison of the individ- • Concepts: the “building blocks of theory”
ual instances indicated a much more mixed view (Strauss & Corbin, 1998, p. 101), the discrete
of the emotional intensity of the relationships, phenomena produced through open coding
ranging from a welcome but painful sense of • Categories: at a higher level of abstraction, sub-
gratitude and debt to a stance of hypersensitivity sume two or more concepts. An especially cru-
and a desire to flee from a relationship which in- cial category may become a core category (see
volved “confinement” or “smothering.” . . . [T]his Box 13.1).
subdivision was retained and coded through • Properties: attributes or aspects of a category
their respective labels “Closeness” and “Over- • Hypotheses: initial hunches about relationships
closeness.” (Pidgeon & Henwood, 2004, p. 638) between concepts
Whiting et al. (2014) used grounded theory to ex- behaviour, one of the categories that emerged was
plore how male perpetrators of intimate partner “remorse.” One interviewee stated, “I have always
violence (IPV) perceived their actions. The re- been upset by the abusive behaviours,” while an-
searchers suggest that examining the perpetra- other said, “I was in the wrong forever [sic] putting
tor’s point of view is important in order to fully my hands on her” (Whiting et al., 2014, p. 282). The
understand IPV and develop effective therapies authors mention that a finding of perpetrator re-
and interventions. morse is rare in studies of IPV, possibly because
Qualitative interviews with 13 men were con- researchers do not want to appear to be softening
ducted, with three research questions guiding or excusing violence. They take the position that
the analysis: How do men who have been violent violence is inexcusable and make their views on
describe their relationship in terms of abuse? that issue quite clear in the article, and also point
What do men believe contributes to the violence? out that the offenders’ remorse may simply be
How do they feel about the violence in their re- an attempt at manipulation. Nonetheless, the
lationship? Prior to open coding, the data were researchers make the case that counsellors and
examined to identify “significant statements” or other professionals should take the remorse ex-
“meaning units.” Open coding was then performed pressed by perpetrators as a starting point to help
by labelling the selected statements with descrip- them develop empathy for their victims and to
tors. Following that, axial coding was done to reflect on how IPV goes against their values and
group the data into “identifiable categories that self-image. Doing so may reduce the tendency to
tied together” (Whiting et al., 2014, p. 279). rationalize IPV and could ultimately result in less
Although it was not uncommon for perpetra- physical abuse.
tors to try to justify or rationalize their violent
310 PART III Qualitative Research
Charmaz (1997) used grounded theory to examine idea of “identity dilemmas”: the ways in which
the identity dilemmas of men with chronic but not men approach and possibly resolve the assault
terminal illnesses. She outlined clearly the chief on their traditionally masculine self-images.
steps in her analysis, which were: She showed that men often use strategies to
• interviews with men and a small number of re-establish earlier selves in order to preserve
women; their identities, at least in their own eyes;
• exploring the interview transcripts for gender • further interviews designed to refine the
differences; categories;
• searching for themes in the men’s interviews • rereading personal accounts of chronic illness
and published personal accounts such as auto- with a particular focus on gender;
biographies. An example was the theme of • reading a new group of personal accounts; and
“accommodation to uncertainty,” as the men • making “comparisons with women on selected
found ways of dealing with the unpredictable key points” (Charmaz, 1997, p. 39).
paths of their illnesses; Charmaz proposed a substantive theory that
• building “analytic categories from men’s def- helped to explain the importance of notions of
initions of and taken-for-granted assumptions masculinity in maintaining an identity for chron-
about their situations” (Charmaz, 1997, p. 39). ically ill men.
Of particular significance in her work is the
One of the key insights of qualitative research get information on the demographic characteris-
is that people often interpret their situations in tics of the participants and she makes no mention
ways that an outsider might not expect. Take the of their nationality. Like many other virtual inter-
example of women with eating disorders. You actions, the ones she observed were probably not
might think that they would definitely want to limited to people from a single country, although
change their eating habits, but Whitehead (2010) much of what she describes is applicable to Can-
found that that is not always the case. Instead, she adian society.
found that some women suffering from anorexia Taking a grounded theory approach to the
nervosa and bulimia do not want to change that issue, Whitehead identified five practices that con-
aspect of their lives, and in fact may seek to de- tributed to the development of a collective iden-
velop a collective identity with women in similar tity: encouraging surreptitious eating behaviours
situations by interacting with them on pro–eating by sharing stories about how they had hidden their
disorder (Pro-ED) websites. These sites are contro- bulimia from their partners or family members;
versial in that they do not present eating disorders focusing on domestic tasks such as cooking (for
in a negative light or recommend that sufferers example, sharing recipes for dishes that are deli-
seek treatment, but instead foster a sense of soli- cious but not calorific and that are easily purged);
darity and companionship for people who already encouraging the idea that beauty, in particular
have the condition and who do not want to change. thinness, is essential to self-worth; promoting
Some openly revel in their situation, such as one friendships among people with eating disorders
woman who wrote that “Ana [anorexia] loves me, and even casting the disorder as a friend; and
She may use me, But she won’t forget me. I am not drawing inspiration from celebrity women who
her victim, I am one of her lovers. Every day I pray to have battled eating disorders or who are simply
be loved above the rest” (Whitehead, 2010, p. 617). thin and beautiful. One of the author’s main con-
The research involved the author regularly clusions is that these practices are “gendered” in
visiting pro-ED websites and closely monitoring the the sense that they reinforce and reproduce dom-
postings made there. Since the research was not inant norms and stereotypes about what it means
conducted in person, Whitehead was not able to to be a woman.
it seems unlikely that this awareness can be can make it difficult for a researcher with a tight
put aside. In fact, few people today believe deadline to carry out a genuine grounded theory
that theory-neutral observation is possible. It analysis, given the constant interplay of data col-
is generally agreed that what is “seen,” even lection and conceptualization that it requires.
in research, is conditioned by what is already • It isn’t clear that the grounded theory method
known about the social world under study (in necessarily results in theory. It offers a rigor-
terms of both social scientific concepts and ous approach to the generation of concepts
general, everyday ones). Many writers con- and categories, but it is often difficult to see
sider it an advantage to build on the work of what theory, in the sense of an explanation of
others. something, is put forward. Moreover, although
• In practical terms, the time required to tran- lip-service is frequently paid to the idea of gen-
scribe recordings of interviews, for example, erating formal theory, most grounded theories
13 | Qualitative Data Analysis 313
In their research into the bus industry, Bryman, • the survival of attitudes and behaviour pat-
Gillingwater, and McGuinness (1996) noticed that terns, particularly among bus drivers, that
managers frequently mentioned that their compan- are seen as inappropriate to the new en-
ies still followed officially discontinued rules and vironment (for example, lack of concern for
practices. Managers often suggested that “inherited” customer service) and that hinder service
characteristics were holding them back in their innovation; and
efforts to adapt to a new competitive environment. • high wages from the earlier era, which make
“Inheritance” is what Strauss (1987) called an in vivo it possible for competitors who pay their
code: one derived from the language of people in the drivers lower wages to enter the market and
social context under study, as opposed to a sociologi- charge less for their services.
cally constructed code, which is a label created by the
Sample comments:
researcher. The following memo outlines the concept
of inheritance, provides some illustrative quotations, “I suppose another major weakness is that
and suggests some properties of the concept. we are very tied by conditions and practi-
ces we’ve inherited.” (Commercial Director,
Memo for inheritance Company G)
Inheritance: Many of our interviewees “We have what we’ve inherited and we now
suggest that they have inherited certain have a massive surplus of double decks [i.e.,
company traits and traditions that they too many buses of the wrong size]. . . . We
themselves would not have chosen. The key have to go on operating those.” (Managing
point about this inheritance is that our inter- Director, C
ompany B)
viewees see it as hindering their ability to
respond to a changing environment. Managing Director of Company E said the
company had inherited staff steeped in old
Inherited features might include: attitudes: “We don’t have a staff where the
• expensive and often inappropriate fleets of message is ‘the customer is number one.’”
vehicles and depots;
pertain only to the specific social phenomenon from coding and memos to the very idea of al-
under investigation—not to a broader range of lowing theory to emerge from data, have also been
phenomena. very influential. It is striking that one of the main
• Grounded theory tends to invite researchers to developments in qualitative data analysis in recent
code their data into discrete chunks. However, years—the use of computer software—has implicitly
some commentators believe that this kind of promoted grounded theory, because the software
fragmentation results in a loss of context and programs have in many cases been written with its
narrative flow (Coffey & Atkinson, 1996); we processes in mind (Lonkila, 1995).
will return to this point later in this chapter.
Basic operations in qualitative
That said, and even though many studies do not data analysis
use it consistently, grounded theory is probably the Coding (or indexing) is the starting point for most
most influential overall strategy in use today for forms of qualitative data analysis. The princi-
qualitative data analysis. Many of its core elements, ples involved have been well developed by writers
314 PART III Qualitative Research
concerned with grounded theory and other perspec- in more than one way. An outburst of anger,
tives. Among the issues they consider in developing for example, can be seen as an emotion, a
codes (cf. Lofland & Lofland, 1995) are the following: cause of stress, or the beginning of a new level
of integration. As a first stage, Charmaz (2004)
• Of what general category is this datum an recommended “line by line coding,” in which
instance? virtually every line in the transcript (or other
• What does this datum represent? data source) has a code attached to it: this way,
• What is this datum about? the qualitative researcher does not lose contact
• What question does this datum suggest? with the data and the perspectives and inter-
• What sort of answer to that question does this pretations of those being studied. Although
datum imply? the line-by-line approach almost always results
• What is happening here? in a proliferation of codes, this should not be
• What are people doing? alarming. What qualitative researchers need to
• What do people say they are doing? do is ask what these codes have in common so
• What kind of event is going on? that they can be combined into higher-order
and more abstract codes.
Steps and considerations in coding • Review the codes, possibly in relation to the
Following are some guidelines for coding: transcripts. Are two or more words or phrases
being used to describe the same phenomenon?
• Code as soon as possible (that is, as data are If so, remove one of them. Do some of the
being collected), as grounded theory suggests. codes relate to concepts and categories in the
This may sharpen an understanding of the data existing literature? If so, would it make sense
and help with theoretical sampling. It can also to use the existing terms? Are there connec-
help to alleviate the feeling of being swamped tions between the codes? Is there evidence that
by the data. At the very least, begin transcribing respondents believe one thing tends to be asso-
recorded interviews at a relatively early stage. ciated with or caused by something else?
• Read through the initial set of transcripts, field • Consider more general theoretical ideas in rela-
notes, documents, and so on, without taking tion to codes and data. Now you can start gen-
any notes or considering any interpretations. erating some general theoretical ideas about
Only when you have finished should you con- the data. Try to outline connections between
sider jotting down a few general notes about concepts and any developing categories. Con-
what seems especially interesting, important, sider in more detail how they relate to the ex-
or significant. isting literature. Develop hypotheses about the
• Do it again. Read through the data again, but linkages you perceive and go back to the data
this time make marginal notes about signifi- to see if they can be confirmed.
cant remarks or observations—as many as • Finally, keep coding in perspective. It’s only one
possible. Initially, they will be very basic: per- part (albeit an important part) of the analysis:
haps keywords used by respondents or themes a mechanism for thinking about the meaning
in the data. Now you are coding: generating of the data and for reducing the data to a man-
terms that will help in interpreting the data. ageable size. The larger task of interpretation
• Don’t worry about generating too many codes— awaits, including forging interconnections be-
at least in the early stages of analysis. Some will tween codes, reflecting on the overall import-
be fruitful, others not. The important thing is ance of the findings for the research literature,
to be as inventive and imaginative as possible; and pondering the significance of the coded
tidying up can be done later. Remember that material for the lives of the people who are the
any one datum can, and often should, be coded subjects of the research.
13 | Qualitative Data Analysis 315
Turning data into fragments some of the main packages can be downloaded from
the distributors’ Internet sites.
The coding of materials such as interview transcripts There is no one correct approach to coding data.
typically entails writing marginal notes on them and As we suggested earlier, grounded theory conceives
gradually refining those notes into codes and then of different types of code. Coffey and Atkinson
cutting and pasting (sometimes in the literal sense, (1996) pointed to three different levels of coding, ap-
with scissors and glue). It entails cutting transcripts plied below to a passage from an interview that first
into chunks of data (and of course carefully identi- appeared in Box 11.5 (p. 247).
fying the origins of each chunk with, for example,
name, position in organization, date) for later data • First there is a very basic coding, which, in the
retrieval. Although word-processing programs can case of the Box 11.5 material could be liking
be used for this task, CAQDAS is becoming increas- or disliking the visit to a Disney theme park.
ingly popular. However, such a coding scheme is unlikely
CAQDAS has been a growth area in terms of both to provide anything more than a superficial
the proliferation of programs and the numbers of analysis.
people using them. Most of the best-known programs • A second level involves a deeper awareness of
allow analysts to code text and later retrieve it, tasks the content of what is said and is organized
that were once done manually. For example, the soft- around the focus of the research. An example
ware can search for all chunks of text relating to a might be countries “well-represented” and
code, and then cut and paste them together. Human “missing” at Disney World.
input is still crucial, however. CAQDAS cannot help • A third level moves slightly away from what
with decisions about the codes, or the coding of text- the respondent says to explore broader analytic
ual materials, or the interpretation of findings. As themes. The responses in the Box 11.5 tran-
with quantitative data analysis software such as IBM script were coded in terms of characteristics
SPSS Statistics (SPSS), someone must still choose the such as enthusiasm (“uncritical enthusiasm”),
variables to be analyzed and the analytic techniques attitudes toward the Disney Corporation (“not
to be used, and then make sense of the results. Each critical of Disney”); comments made about
form of software requires creativity. CAQDAS differs typical visitors (“visitors’ ethnicity”); and
from SPSS largely in terms of the type of data that can the nature of critical comments (“aesthetic
be used with it. critique,” “ethnicity critique,” “nationality cri-
There is no CAQDAS industry leader, but NVivo is tique”). Interestingly, the passage also reveals
a package that most researchers would at least know the potential for a code employed by Coffey
by name. Earlier versions of the software became and Atkinson (1996, pp. 43–45): the use of a
very popular in the 1990s. The Appendix outlines “contrastive rhetoric” (making a point about
how to use NVivo for qualitative data analysis. one thing by comparing it to something else).
To use or not to use CAQDAS? With a very small data This feature occurred when the husband made
set, it is probably not worth the time and trouble re- a point about the representation of British cul-
quired to become familiar with new software. Catterall ture, which he regarded as poor compared to
and Maclaran (1997) have argued that CAQDAS is not that of China.
very suitable for focus group data because the code and
retrieve function tends to hide the communication As Coffey and Atkinson (1996) observed, follow-
process typical of focus groups. On the other hand, ing Strauss and Corbin’s account (1990) of grounded
learning new software will give you skills that you may theory, codes should not be thought of purely as
need in the future. CAQDAS is likely to be too expen- mechanisms for the fragmentation and retrieval
sive for personal purchase, though there are student of text. They can do more than simply manage the
and educational discounts. Demonstration copies of data gathered. For example, examination of the
interconnections between codes may reveal that said. (For a general critique of qualitative data an-
some are dimensions of a broader phenomenon. For alysis packages see Box 13.5.) By plucking chunks of
example, “ethnicity critique” came to be seen as a text out of the context in which they appear, such
dimension of “ideology critique,” along with “class as a particular interview transcript, the social set-
critique” and “gender critique.” In this way, a map ting can be lost. A second criticism of coding is
of the more general or formal properties of the con- that it results in fragmentation of the data, causing
cepts under development can be started. the narrative flow to be lost (Coffey & Atkinson,
1996). Sensitivity to this issue has been heightened
Problems with coding since the late 1980s by a growing interest in nar-
One of the most common criticisms of coding quali- rative analysis (see later in this chapter). Riessman
tative data is the risk of losing the context of what is (1993) became concerned about fragmentation when
Whereas the use of computer software is almost retrieving text—that owes a great deal to
universal in quantitative data analysis, among grounded theory. In their view, the emergence
qualitative data analysts it is often avoided, for of grounded theory as a new standard is incon-
several reasons. sistent with a key strength of qualitative re-
• Some writers worry that because qualitative search: its flexibility.
data analysis packages make it so easy to • On the other hand, several writers have ex-
quantify coded text, qualitative research will tolled the virtues of such packages:
become subject to the reliability and validity • CAQDAS , like NVivo, invites thought about
criteria of quantitative research (Hesse-Biber, “trees” of interrelated ideas—a useful feature
1995). in that it encourages the analyst to consider
• Others feel that CAQDAS reinforces and even possible connections between concepts.
exaggerates the tendency toward a fragmenta- • Quantitative researchers often criticize the
tion of the textual materials (Weaver & Atkinson, tendency toward “anecdotalism” found in much
1995), destroying the natural flow of interview qualitative research—that is, the tendency to
transcripts and field notes. Awareness of con- use quotations from interview transcripts or
text is crucial to many qualitative researchers, field notes with little sense of how represent-
and they are concerned by the prospect that ative they might be. CAQDAS makes it possible
this awareness may be pushed to the sidelines. to count the frequency with which a particular
Stanley and Temple (1995) suggested that most form of behaviour occurs or a particular view-
of the coding and retrieval features needed for point is expressed in interviews.
qualitative data analysis, such as search, cut, • CAQDAS enhances the transparency of qualita-
and paste, already exist in word-processing tive data analysis. How qualitative data were
software. Using Microsoft Word, for example, analyzed is often unclear in published reports
would save not only money but also the time (Bryman & Burgess, 1994). CAQDAS may force
required to learn new software. researchers to be more explicit and reflect-
• Coffey et al. (1996) argued that the style of ive about the process of analysis, and may in-
qualitative data analysis in most CAQDAS pre- directly encourage replication, a feature often
sumes a methodology—based on coding and lacking in qualitative analysis.
318 PART III Qualitative Research
As part of a three-year participant observation feminist theories about bodies, he found that these
study of tattooing, Atkinson (2002, 2004) elicited women are not misfits, and that they use tattoos to
tattoo narratives from the women he interviewed. signify their “established” or “outsider” construc-
The great increase in the numbers of women getting tions of femininity. Ideas about femininity, including
tattoos has clearly challenged the standard “mascu- conformity to and resistance against cultural norms,
linity” explanations of male tattooing. Drawing on are crucial in explaining women’s tattooing.
of what happened?” Proponents of narrative an- how she applied narrative analysis to conventional
alysis argue that most approaches to the collection interview transcript material, and then began to un-
and analysis of data ignore the fact that people per- cover the stories her interviewees were telling her.
ceive their lives in terms of continuity and process, In this case, Riessman was applying a narrative ap-
and that efforts to understand social life that do not proach to materials gathered in a conventional way
take that fact into account may misunderstand the for conventional purposes. Other researchers start
perspective of those being studied. Narrative a nalysis out with the intention of conducting a narrative an-
is obviously suitable for life history research, but its alysis and deliberately ask people to recount stories
applications can be much broader. For example, it (for example, Miller, 2000). While stories can emerge
can also be applied to accounts of shorter episodes in response to questions not designed to elicit a nar-
and their interconnections. rative, certain kinds of question are especially likely
Some researchers apply narrative analysis to to produce them. Riessman (2004a) suggested that
interviews. In her account of her “click moment” as prompts such as “Tell me what happened” followed
a narrative researcher, Riessman (1993) described by “And then what happened?” are much more
320 PART III Qualitative Research
likely to elicit a narrative account than “When did observation), but it has also become an interviewing
X happen?” While some narrative researchers prefer approach in its own right. Also, as we will see in con-
to start people off by asking them to tell their story nection with the writing of ethnographic research
of an event, Riessman argued that it is usually neces- (see “Writing up ethnography” in Chapter 15), there
sary to keep asking follow-up questions to stimulate is growing recognition of ethnography as a narrative
the flow of details and impressions. For example, designed to tell a story about a way of life.
in her study of divorce, she often asked: “Can you Narrative analysis has been criticized. Bury (2001),
remember a time when . . . ?” and then followed it while noting the increased interest in illness narra-
up with “What happened that makes you remember tives (accounts of illnesses people have experienced),
that particular moment in your marriage?” argued that there has been a tendency for narrative
Coffey and Atkinson (1996) argued that a narra- researchers to treat such stories uncritically. For ex-
tive should be examined in terms of the functions ample, he suggested that the frequent references in
that it serves for the teller. The aim of narrative illness narratives to coping with and “normalizing”
interviews is to elicit interviewees’ reconstructed illness may often represent attempts to convince the
accounts of connections between events and links audience (whether the interviewer or the reader of
between events and contexts. Miller (2000) proposed the narrative) of the subject’s competence; in other
that narrative interviews in life story or biographical words, such accounts may be motivated more by the
research are far less concerned with eliciting facts desire to be seen as a fully functioning member of
than with drawing out interviewees’ reflections and society than by the desire to give a realistic account
interpretations, as revealed in their accounts of their of coming to terms with a medical condition. How-
lives or families. ever, as Bury recognized, the social conditions that
Narrative analysis, then, is an approach to the prompt such narratives and the forms the narratives
analysis of qualitative data that examines the stories take are themselves revealing. In drawing attention
that people use to explain events. It can be applied to the motives behind illness narratives, he was not
to data that have been acquired through a var- seeking to undermine narrative analysis but to show
iety of research methods (notably semi-structured how the range of issues addressed in this form of re-
and unstructured interviewing and participant search can be expanded.
Key Points
• The collection of qualitative data frequently results • Coding is a key process in most qualitative data
in the accumulation of a large volume of information. analysis strategies, but it may have the effect of
• Qualitative data analysis is not governed by codi- fragmenting and decontextualizing information.
fied rules to the same extent that quantitative data • Narrative analysis emphasizes the stories that
analysis is. people tell in the course of interactions with a
• There are different approaches to qualitative data qualitative researcher; it has become a strategy
analysis; grounded theory is probably the most for producing and analyzing qualitative data in its
prominent. own right.
qualitative interviews. What concepts might Basic operations in qualitative data analysis
emerge when you begin open coding? RR What are the main steps in coding?
AA Is narrative analysis the answer to the problem
RR What is theoretical sampling? of data fragmentation that results from coding?
AA Charmaz wrote that theoretical sampling “rep- Explain.
resents a defining property of grounded theory”
(2000, p. 519). Why do you think she feels this way? Narrative analysis
RR Explain how the emphasis on stories in narrative
RR What are some of the main criticisms of grounded analysis represents a distinctive approach to the
theory? production and analysis of qualitative data.
AA How realistic is the assumption in grounded AA Come up with a research question pertaining to
theory that a researcher can begin to collect data gender roles that can be answered using narrative
without being influenced by existing theories and analysis.
concepts? Explain.
Relevant Websites
A wealth of information is available at the website of If you’d like to experiment with NVivo software, trial
the Grounded Theory Institute. downloads are available at the NVivo website.
www.groundedtheory.com www.qsrinternational.com/products_free-trial-
software.aspx
These YouTube videos discuss grounded theory,
including coding and constant comparison. (Websites accessed 1 November 2018)
www.youtube.com/watch?v=4SZDTp3_New
www.youtube.com/watch?NR=1&v=gn7Pr8M_Gu8
www.youtube.com/watch?v=vi5B7Zo0_OE&NR=1
PART IV
Transcending the Quantitative/Qualitative
Divide and Some Practical Advice
© kendo_OK/Shutterstock.com
14
Revisioning Quantitative and Qualitative
Chapter Overview
This chapter is concerned with how far the quantitative/qualitative distinction should be taken. Although
there are many differences between the two research strategies, there is also much that unites them, as
we will see when we discuss how they can be combined. This chapter explores:
• aspects of qualitative research that can contain elements of the natural science model;
• aspects of quantitative research that can contain elements of interpretivism;
• the idea that research methods are more independent of epistemological and ontological assump-
tions than is sometimes supposed;
• ways in which the quantitative/qualitative contrast may break down;
• studies in which qualitative research is used to analyze quantitative research and vice versa;
• the use of quantification in qualitative research;
• arguments against the combination of quantitative and qualitative research;
• two versions of the debate on combining quantitative and qualitative research, one concentrating
on methods of research and the other on epistemological issues;
• different ways in which multi-strategy research has been carried out; and
• the claim that multi-strategy research is not inherently superior to research employing just one
research strategy.
Philosophers of science sometimes use the term have a unity of form and purpose that will allow for a
“consilience” to refer to the notion that there is a unity more thorough and profound understanding of natural
or compatibility between all branches of knowledge. and social phenomena. This chapter is written in a sim-
The word came into wide usage after the biologist E.O. ilar spirit, seeking to show that quantitative and quali-
Wilson published Consilience: The Unity of Knowledge tative research strategies have a great deal in common
(1998), in which he argued that the natural sciences, the and can complement each other. It makes the case that
social sciences, and even the humanities may one day both are required to gain knowledge of the social world.
As you progress through your academic career, you researchers and can ultimately be quite destructive.
may be tempted to take a dismissive attitude toward Drawing on the insights of multiple perspectives is
approaches and methodologies that are different from much more conducive to good research and produces a
your own. Try to resist that temptation: giving in to it deeper level of understanding.
can lead to needless tensions and rivalries between
Critical realism (Bhaskar, 1975; 1979; 1986) pre- spent three months as a staff nurse. He found that
sents a view of reality that in many ways is in racism was expressed in the form of remarks made
keeping with the ontological approaches taken behind the backs of members of racial minorities.
in the natural sciences, despite important differ- However, it did not intrude into work relationships,
ences (Gorski, 2013). Critical realist ethnography, because more weight was given to achievements
for example, maintains that the social world has and performance (such as qualifications and
a form and structure that exists independently of medical skills) than to “race” when judging mem-
our perception of it and which affects human con- bers of professions. In terms of critical realism,
sciousness and behaviours in complex ways. This one possible structural mechanism (racism) was
sort of ethnography is also based on the idea that countered by the operation of another structural
the descriptions and explanations presented by mechanism (professional ideology). Only on rare
social researchers are inherently provisional and occasions did racial tension surface.
fallible, and as such should be subject to empir- Similarly, Barron (2013) used the critical realist
ical scrutiny, and on the notion that researchers paradigm in his ethnographic study of ethnic iden-
should seek to uncover causal tendencies to ex- tity among white British and Pakistani-British chil-
plain their subject matter. dren. He concluded that his previous ethnographic
These sorts of positivist-friendly assumptions work on the topic, although insightful, could have
are evident in the work of Porter (1993; 2002), who been improved by acquiring a broader range of
took a critical realist stance in his ethnographic empirical data, a practice that has important par-
study of racism in a large Irish hospital where he allels in the natural sciences.
method. Another illustration is Charmaz’s (2000) structured interviews and questionnaires. For ex-
suggestion that in spite of the differences that de- ample, there is a huge literature on political attitudes,
veloped between Glaser (1992) and Strauss (for ex- and many indexes and scales have been designed to
ample, Strauss & Corbin, 1998)—who were major measure them (see Robinson et al., 1999). The wide-
proponents of grounded theory—both maintained spread inclusion of questions about attitudes in
that an objective, external reality exists (the position social surveys suggests that quantitative researchers
taken in the natural sciences). Each posited a social are interested in matters of meaning too.
world beyond the researcher and maintained that It may be objected that survey questions do not
it was the job of the social investigator to reveal its really tap issues of meaning because they are based
nature and functioning. on categories devised by the designers of the inter-
view schedule or questionnaire and not by the par-
ticipants themselves. Two points are relevant here.
Quantitative research First, the notion that qualitative research is better
and interpretivism at gaining access to the point of view of those being
Qualitative research would seem to have a studied is generally assumed rather than demon-
near-monopoly on the ability to study meaning.
strated. Qualitative researchers frequently claim
But is it really the case that only qualitative research to have tapped into participants’ world views be-
makes it possible to see the social world through the cause of their extensive participation in the daily
eyes of the people studied? That contention seems at life of their subjects, the length of time they spend
odds with the widespread study of attitudes using in the setting under investigation, or the lengthy
328 Part IV Transcending the Quantitative/Qualitative Divide and Some Practical Advice
and intensive interviews they conduct. However, the cancer were in women under 50, and the mean age at
explicit demonstration that interpretative under- diagnosis was 65. This inconsistency allowed Lantz
standing has been accomplished—for example, and Booth to uncover how the media drew a con-
through respondent validation (see Chapter 9)—is nection between youthful lifestyles (working outside
rarely undertaken. Second, if the design of attitude the home, postponing parenthood, greater sexuality)
questions is based on prior questioning that seeks to and breast cancer—a connection consistent with the
bring out the range of possible attitudinal positions articles’ “blame the victim” theme. The quantitative
on an issue, attitudinal questions can certainly pro- content analysis of articles on breast cancer played
vide access to meaning. an important part in revealing the social construc-
Also, the practice in much survey research of tion of beliefs about the disease. More generally, this
asking respondents the reasons for their actions example shows how quantitative research can con-
implies that quantitative researchers are frequently tribute to constructionist research.
concerned with uncovering meaning. For example,
in the research on delinquency in Box 1.2, the boys
were asked to give the reasons for their actions Research methods and
in their own words and then to choose the socio- epistemological and
logical theories that came closest to explaining
them. E xamples such as these point to the possibil-
ontological considerations
ity that the gulf between quantitative and qualitative That each research method does not carry with it a
research is not as wide as is sometimes supposed. full cluster of epistemological and ontological com-
mitments can be seen in studies on social research
itself. For example, Platt’s (1986) historical research
Quantitative research on American sociology suggested that the alleged
and constructionism connection between conservative functionalist
We noted in Chapter 1 that a keynote in qualita- theory (which is often associated with positivism)
tive analysis is constructionism: how people build and survey research is greatly exaggerated. Her re-
up images or representations of the social world. search indicated that “the two originated independ-
Qualitative content analysis plays an important ently, and that leading functionalists had no special
role in developing an understanding of how people propensity to use surveys and leading surveyors no
construct their visions of reality, in the same way special propensity for functionalism” (Platt, 1986,
that discourse analysis (Chapter 12) does using p. 527). Moreover, Platt’s general conclusion on the
materials such as newspaper reports and television use of research methods in American sociology
programs. between 1920 and 1960 is very revealing:
Lantz and Booth’s (1998) research on the social
construction of breast cancer provides an example. Research methods may on the level of theory,
As Box 1.6 makes clear, much of their understand- when theory is consciously involved at all,
ing of the representation of breast cancer came reflect intellectual . . . post hoc justifications
from a qualitative content analysis, but they con- rather than . . . carefully chosen fundamen-
ducted a quantitative content analysis as well. The tal assumptions. Frequently methodological
latter showed that 80 per cent of the women in the choices are steered by considerations of a prac-
photographs attached to breast cancer articles were tical nature . . . and are just slogans and aspira-
under age 50, and that 85 per cent of the anecdotes tions. . . (1996, p. 275)
and case stories related to women in this age group.
This emphasis on younger women helped to create Again, even when there are discernible links be-
the impression that they were the most at risk. But in tween research methods and assumptions about
reality, fewer than 20 per cent of new cases of breast knowledge, the connections are not absolute.
14 | Revisioning Quantitative and Qualitative 329
doctor–patient interactions in public and private of their replies. For example, Phoenix reported one
oncology clinics, for example, Silverman (1985) young black woman as saying that she had liked the
quantified some of his data to show that patients in interview, adding: “I had the chance to explain how
private clinics have more influence over what goes I feel about certain things and I don’t really get the
on in the consultations. Still, Silverman warned that opportunity to do that much” (Phoenix, 1994, p. 61).
such quantification should reflect the research par- Another interviewee said it was a “good interview”
ticipants’ own ways of understanding their social and added: “I have never talked so much about
world. myself for a long time, too busy talking about kids
It has often been noted that qualitative research- and their problems” (Phoenix, 1994, p. 61). While
ers engage in “quasi-quantification” when they such qualitative interviews are clearly valuable in al-
use terms such as “many,” “often,” and “some” (see lowing the perspectives of people whose voices are
p. 332). The only difference in studies like Silver- normally silent to surface, they don’t fit the defin-
man’s is that the researcher makes such estimates of ition of “naturalistic” used by critics of quantitative
frequency more precise. social science. Thus it would be incorrect to assume
that artificiality is a problem only in quantitative
Artificial versus natural research.
The distinction between artificial and natural is sim- As we noted in Chapter 11, focus group research
ilarly open to criticism. It is often assumed that be- is often described as more natural than qualitative
cause much quantitative research employs research interviewing because it resembles the way people
instruments (such as questionnaires) that offer only discuss issues in real life. Natural groups (people al-
limited and indirect indicators of people’s lives, its ready known to each other) are often used to empha-
account of the social world is artificial. Qualitative size this element. Whether focus group participants
research tends to be considered more naturalistic in see their interaction as “natural” is unclear, however.
that it focuses on observing people in their natural In particular, since the participants are often paid
settings, behaving as they normally do. This would strangers who have travelled some distance to dis-
seem to be particularly true of ethnographic re- cuss topics they rarely if ever talk about, it’s clear that
search, since the participant observer studies people this sort of research may not be as natural as some
in their normal social worlds and contexts, as they have assumed.
go about their everyday activities. However, when In participant observation, the researcher can be
qualitative research is based on interviews, the “nat- a source of interference, making the research situ-
ural” label is less applicable. Interviews still have to ation less natural than it might appear. It’s difficult
be arranged and interviewees have to be taken away to estimate how much impact reactivity has on the
from activities they would otherwise be engaged in, research findings in such cases, but once again the
even when the interviewing style is of the more con- naturalism of such studies is questionable. When
versational kind. Furthermore, little is known about the ethnographer also engages in interviewing (as
interviewees’ reactions to and feelings about being opposed to casual conversation), the naturalistic
interviewed. quality is definitely compromised, although the re-
Phoenix (1994) reported on the responses of inter- search will probably be somewhat less artificial than
viewees to in-depth interviews in connection with it would be if quantitative methods were used.
two studies: one focusing on mothers under the age These observations cast doubt on the rigidity of
of 20 and the other on the social identities of young the quantitative/qualitative distinction. Once again,
people. Although many of her subjects seemed to this is not to suggest that the usual contrasts are un-
enjoy being interviewed, they were clearly conscious helpful: only that they should not be exaggerated.
of the fact that they had been engaged in interviews Quantitative and qualitative research strategies are
rather than conversations. This was revealed in some not absolutely divergent.
14 | Revisioning Quantitative and Qualitative 331
of quantitative researchers” (Hodson, 1999, p. 68). interview transcripts, and accounts of encounters
In other words, the application of quantitative meth- between people say little or nothing about the
ods to qualitative research may provide a meeting prevalence of the phenomena that such evidence
ground for the two research strategies. is supposed to indicate. There is also the related
risk that the researcher may attribute undue sig-
nificance to a particularly striking statement or
Quantification in unexpected activity.
qualitative research At least partly in response to these problems,
As we noted in Chapter 9, perhaps the most basic qualitative researchers sometimes undertake a lim-
difference between quantitative and qualitative re- ited quantification of their data. Gabriel (1998) col-
search is the “numbers versus words” distinction. Yet lected 377 stories about organizational culture in
most qualitative research does contain some quan- the course of 126 interviews in five organizations.
tification. Three observations are worth making He identified different types of stories and noted
about quantification in the analysis and writing up the frequency of each type. For instance, there were
of qualitative data. 108 comic stories (usually a mechanism for the
disparagement of others); 82 epic stories (survival
Thematic analysis against the odds); 53 tragic stories (undeserved mis-
Chapter 13 noted that qualitative data analysis often fortune); and 40 gripes (personal injustices). The
takes the form of a search for themes in transcripts stories could have been presented as anecdotes, but
or field notes. The choice of themes to look for is simple counting conveyed a clear sense of their rela-
often determined by how frequently certain inci- tive prevalence.
dents, words, phrases, and so on recur. This process Using actual numbers can counter the criticism
may also account for the prominence given to some that qualitative data are too anecdotal and do noth-
themes over others. In other words, a kind of implicit ing to indicate the extent to which certain beliefs
quantification probably influences both the identifi- or behaviours occur. The precision of numbers also
cation of themes and the elevation of some themes makes them superior to the estimates of frequency
over others. that must be inferred from quasi-quantification
terms such as “some” or “many.”
Quasi-quantification in qualitative We may see greater use of quantification in
research qualitative research in the future as a result of the
As we have noted, qualitative researchers engage in growing use of software programs for qualitative
“quasi-quantification” when they use terms such as data analysis. Most of the major software programs
“many,” “frequently,” “rarely,” and “some,” which by make it possible to produce simple counts for the
definition are based on the relative frequency of the use of a particular word or the incidence of a coded
phenomenon of interest. As expressions of quantity, theme. In many cases, they can also produce simple
however, these terms are imprecise, and it’s often cross-tabulations—for example, showing which sex
difficult to tell why they are being used at all. When uses the passive voice more often.
quantification is needed to support a point, it may be
better to use actual numbers.
Multi-strategy research
Combatting anecdotalism through The ways in which quantitative and qualitative re-
limited quantification search can complement each other can be seen most
Qualitative research is sometimes criticized for clearly in multi-strategy research, which by defin-
the anecdotal nature of its data, which leaves the ition involves combining the two modes of investiga-
reader with no way of assessing their generalizabil- tion. However, there are two basic arguments against
ity. Brief sequences of conversation, snippets from this sort of research.
14 | Revisioning Quantitative and Qualitative 333
beliefs and assumptions about how the world works • An epistemological argument, as in the em-
and how knowledge of it is to be gained. Kuhn (1970) bedded methods and paradigm positions dis-
popularized the term in his portrayal of natural sci- cussed above, sees quantitative and qualitative
ence as going through periods of revolution, whereby research as grounded in incompatible epis-
“normal” science (science carried out in terms of the temological principles (and ontological ones
prevailing paradigm) is increasingly challenged by too, though these tend to be given less atten-
new findings inconsistent with the assumptions and tion). According to this version, multi-strategy
established findings of the discipline. The increas- research is not possible in principle.
ing frequency of such anomalies eventually sparks a • A technical argument perceives research meth-
crisis, which in turn sparks a revolution. The period ods as independent of any specific epistemo-
of revolution comes to an end when a new paradigm logical position. According to this view, a
gains acceptance and a period of the new “normal” research method from one research strategy
science sets in. An important feature of paradigms can be pressed into the service of another:
is that they are incommensurable: inconsistent with quantitative and qualitative methods can be
each other because of their divergent assumptions fused. This is the position taken by most re-
and methods. Disciplines in which no paradigm has searchers whose work is mentioned in the next
emerged as pre-eminent, such as sociology and other section.
social sciences, are deemed “pre-paradigmatic.”
The paradigm argument maintains that quantita- The technical argument views the two research
tive and qualitative research approaches are based on strategies as compatible, and multi-strategy research
incompatible paradigms in which epistemological as both feasible and desirable. It is in that spirit that
assumptions, values, and methods are inextricably we will now consider the ways in which quantitative
intertwined (Morgan, 1998). Therefore, when re- and qualitative research can be combined.
searchers combine participant observation with a
questionnaire, they are not really combining quan-
titative and qualitative research, since the paradigms
Approaches to multi-strategy
are incompatible: the integration is only superficial research
and within a single paradigm. Hammersley (1996) proposed three approaches to
The problem with the paradigm argument, as with multi-strategy research:
the embedded methods claim, is that it rests on a con-
nection between method and epistemology that has • Triangulation: the use of quantitative research
not been demonstrated. Moreover, although Kuhn to corroborate qualitative research findings or
(1970) argued that paradigms are incommensurable, vice versa
it is by no means clear that quantitative research and • Facilitation: the use of one research strategy to
qualitative research are fully separate paradigms. As aid research using another
we have suggested throughout this chapter, there are • Complementarity: the use of two different re-
areas of overlap and commonality between them. search strategies so that diverse aspects of an
investigation can be combined
on methods of investigation and sources of data. One appearing mature and establishing relationships
of the reasons Webb et al. (1966) called for greater with the other sex, and tended to think of designer
use of unobtrusive methods is their potential for tri- drinks as targeted at immature younger drinkers.
angulation (see Box 6.5). Triangulation can operate These connections with age were confirmed by the
within and across research strategies. It was origin- quantitative evidence, which also corroborated the
ally conceptualized by Webb et al. (1966) as a way of suggestion from the qualitative evidence that the
developing additional measures, resulting in greater consumption of designer drinks was largely associ-
confidence in findings, and was strongly associated ated with a desire to get drunk.
with a quantitative research strategy. However, tri- In this study the use of triangulation appears
angulation can also take place within qualitative to have been planned, but this need not always be
research. In fact, ethnographers often check their ob- the case. Researchers may start out using different
servations with interview questions to look for pos- research strategies for different purposes but in the
sible misunderstandings. Bloor (1997) reported that process discover that they have generated quantita-
he tackled the process of death certification in two tive and qualitative findings on related issues, and
ways: interviewing clinicians responsible for certi- that they can treat their overlapping results as a tri-
fying causes of deaths, and asking the same people angulation exercise.
to complete dummy death certificates based on case Whether triangulation is planned or unplanned,
summaries he had prepared. Increasingly, triangu- there is always the possibility that the two sets of
lation refers to the cross-checking of findings from findings will not corroborate one another. In the
both quantitative and qualitative research (Deacon event that the results are inconsistent, the solution is
et al., 1998). not to arbitrarily decide which set of findings is cor-
An illustration of a triangulation approach is rect. Usually, further research is required to resolve
Hughes et al.’s (1997) study of the consumption of the matter.
“designer drinks” (fortified wines and strong white Deacon and colleagues (1998) used several quan-
ciders) by young people. The authors used two main titative and qualitative research methods to gather
research methods: information, without intending to conduct a triangu-
lation exercise. But analysis of their data revealed
• Qualitative method: eight focus groups with 56 an inconsistency: the quantitative data suggested a
children and young adults, with each discus- broad consensus between journalists and social sci-
sion lasting around two hours entists on the reporting of social scientific research
• Quantitative method: a survey administered in the media, but the qualitative findings suggested
in two parts to a multi-stage cluster sample of some conflict of approaches and values between the
824 12- to 17-year-olds. The first part was an two groups. Instead of favouring one set of findings
interview and the second was a questionnaire over the other, the researchers r e-examined the data
designed to elicit more sensitive information. and found that social scientists tended to answer
questions about media coverage of their work differ-
The results achieved with the two research ently, depending on the research instrument. In re-
strategies were mutually reinforcing. The qualita- sponse to survey questions they typically expressed
tive findings showed age differences in attitudes relief that the coverage was not as bad as they had ex-
toward designer drinks and other types of alcohol: pected, whereas in interviews they tended to empha-
the youngest people (12- and 13-year-olds) tended size “war stories” of wounding encounters with the
to adopt a generally experimental approach; the media. In general, then, the questionnaires showed
14-and 15-year-olds thought of drinking as a way of social scientists to be relatively pleased with the re-
having fun and losing inhibitions, and felt that de- porting of their research, but their replies were more
signer drinks met their needs well; the oldest group negative when they were encouraged to reflect on
(16- and 17-year-olds) were mainly concerned with specific problems in the past.
336 Part IV Transcending the Quantitative/Qualitative Divide and Some Practical Advice
Another example of triangulation can be found in family counsellors that supported her hypoth-
Fenton et al.’s (1998) study of the same topic. It em- esis that “the process of incest in structurally
ployed both quantitative and qualitative methods: different families may vary” (Phelan, 1987,
p. 39). Similarly, Bell’s (2007) study of western
1. content analysis of news and current affairs Canadian separatism started out as a quali-
coverage (local and national newspapers, TV, tative investigation involving attendance at
and radio) separatist events and hanging out with separa-
2. mailed questionnaires soliciting social scien- tists. A
nalysis of these data led to the hypoth-
tists’ views about media coverage and their esis that support for separatism was associated
own practices (674 replies) with a neo-liberal ideology and a partisan dis-
3. mailed questionnaires soliciting the views of like for the ruling federal Liberal Party, a view
social scientists identified in the content an- that was later substantiated in a multiple re-
alysis as having received media coverage (123 gression analysis using survey data.
replies) • By aiding measurement. The in-depth know-
4. semi-structured interviews with 20 of the ledge of social contexts acquired through
latter qualitative research can be used to inform
5. semi-structured interviews with 34 journalists the design of survey questions for structured
identified in the content analysis interviewing and self-completion question-
6. semi-structured interviews with 27 represent- naires. Johnson and colleagues (2003) used
atives of funding bodies and government data from unstructured interviews to develop
7. observing journalists’ movements and activ- highly structured questions for later interviews
ities at three conferences with adolescent smokers. Walklate (2000) ex-
8. focus group analysis of audience reception of plained that for her research on fear of crime
media coverage (13 focus groups) and safety issues in high-crime areas, a survey
using traditional questions about victimiza-
Qualitative research facilitates tion was used. However, the questions were
quantitative research amended to reflect the local context following
Here are two ways in which qualitative research can six months of interviews, ethnography, and
serve as a guide for quantitative research: examination of local newspapers. Similarly,
the survey questions that Bell (2007) used in
• By suggesting hypotheses. Because of its ten- his study of western Canadian separatism were
dency toward unstructured, open-ended prompted in part by his experiences in gather-
methods of data collection, qualitative re- ing qualitative data.
search is often a source of hypotheses that
can be tested using a quantitative strategy. An Quantitative research facilitates
example is Phelan’s (1987) research in which qualitative research
she conducted qualitative interviews and con- One way in which quantitative research can prepare
versations with people attending a treatment the ground for qualitative work is through the se-
program for families dealing with incest. lection of people to be interviewed. This can occur
After a considerable amount of qualitative in several ways. In the case of Fenton et al.’s (1998)
data had been collected, Phelan became aware research on the reporting of social science research
of differences between biological fathers and in the mass media, discussed above, a media con-
stepfathers in how they interpreted their in- tent analysis (method 1) was used as a source of
cestuous acts, and in the frequency with which data. However, it also served as a means of identify-
certain behaviours occurred. Q uantifiable ing journalists who had reported relevant research
data were collected through interviews with (method 5). In addition, replies to questions in the
14 | Revisioning Quantitative and Qualitative 337
general survey of social scientists (method 2) were or qualitative research had priority, and which one
used to help identify two groups of social scientists, came first in the sequence. Sometimes research uses
those with high and those with low levels of media both strategies and several methods, but no single
coverage of their research, who were then inter- one is dominant.
viewed with a semi-structured approach (method
4). Similarly, Jamieson (2000) administered a ques- Static and process features
tionnaire to a sample of young men on their crim- Together, the characteristics listed in Table 9.1 sug-
inal activity. On the basis of their replies, qualitative gest that quantitative research presents a static pic-
interviews were conducted with equal numbers of ture of social life, while qualitative research presents
young men in each of three categories: those who had a more dynamic picture, emphasizing process. In
never committed a criminal offence; those who some contexts “static” may be considered a negative
had offended but not recently; and persistent offenders. term, but in this case it’s positive. In fact, it is the
regularities revealed by quantitative research that
Filling in the gaps allow for the analysis of process. A multi-strategy
When the researcher cannot rely on either a quanti- research approach makes it possible to combine the
tative or a qualitative method alone, a multi-strategy two. An illustration is provided by MacKinnon and
approach can be used to fill in the gaps. For example, Luke’s (2002) study of cultural change from 1981
ethnographers may employ structured interviewing to 1995. Their main method of data collection was
or self-completion questionnaires if not everything survey research, but to complete the picture they
they need to know is accessible through direct ob- also examined census and public opinion data, as
servation, or if they have difficulty gaining access to well as newspaper articles on historical events. They
certain groups of people. found that levels of homophobia and anti-Semitism
Morgan (1998) suggested a two-step approach to decreased over that period, but that sympathy for
planning multi-strategy research: Indigenous people also declined. They cautioned
that their small sample (70) and what were essen-
• The priority decision. Will the princi- tially “good guesses” meant that their conclusions,
pal data-gathering tool be qualitative or “like the interpretations of an ethnographer [were]
quantitative? subject to alternative interpretations by others”
• The sequence decision. Should the “comple- (MacKinnon & Luke, 2002, p. 332).
mentary” method be used first, as a prelim-
inary to the principal method, or second, as a Researcher and participant
follow-up? perspectives
Sometimes researchers want to gather two kinds of
These decisions yield four possible types of re- data: qualitative data that show the general perspec-
search design (see Figure 14.1). tives of the people they are studying, and quantita-
One difficulty with using Morgan’s typology is tive data on specific issues. For example, Milkman
that it may be difficult to know whether quantitative (1997) was interested in the meaning of industrial
work: in particular, whether factory conditions
Priority had changed since the 1950s, when they were por-
Quantitative Qualitative
trayed very negatively. To find out, she employed
semi-structured interviews and focus groups with
Preliminary M1 M2 General Motors production workers. She was also
Sequence
Follow-up M3 M4 interested in a “buyout” plan that the company’s
management introduced in the mid-1980s after it
FIGURE 14.1 Morgan’s classification of approaches had initiated a variety of changes to work practices.
to multi-strategy research The plan offered workers the opportunity to give up
338 Part IV Transcending the Quantitative/Qualitative Divide and Some Practical Advice
their jobs for a substantial cash payment. In 1988 Another illustration is provided by Barnard and
and again in 1991, Milkman carried out a question- Frischer’s (1995) research on HIV-related risk behav-
naire survey of workers who had accepted the com- iour among drug injectors. Structured interviews
pany’s buyout offer. The surveys inquired about the with 503 injectors revealed that “females report[ed]
reasons for taking the buyout, how those taking it significantly higher levels of needle sharing, sexual
had fared since leaving General Motors, how they activity, and AIDS awareness than their male
felt about their current employment, and differences counterparts,” and that “women . . . co-habiting
among social groups (in particular ethnic groups) in with sexual partners who [were] themselves inject-
current earnings relative to those at General Motors. ors, [were] particularly likely to report high levels of
risk behaviour and also AIDS awareness” (Barnard &
The problem of generality Frischer, 1995, p. 357). What produced this relation-
As we noted above, there is a tendency in qualita- ship between gender, risk behaviour, and cohabita-
tive research to present findings without evidence tion? Semi-structured interviews with 73 injectors
indicating how typical they are, or how the group suggested that the relationship between these vari-
under study compares with some larger popula- ables could be explained “by the tendency for women
tion. Partly to counter this problem, Tastsoglou and to be in sexual relationships with men who them-
Miedema (2003) used quantitative national data on selves inject and with whom they are unlikely to use
women’s volunteer work to compare their admittedly condoms” (Barnard & Frischer, 1995, p. 360). Here
non-random sample of immigrant women with the qualitative data were used to shed light on relation-
Canadian population; they found that the immi- ships among variables derived from quantitative
grant women actually volunteered more than the research.
national average.
Studying different aspects
Qualitative research may help to of a phenomenon
interpret the relationship There is a tendency to think of quantitative re-
between variables search as best suited to the investigation of “macro”
A frequent problem for quantitative researchers is phenomena (such as social mobility and social
how to explain relationships between variables. One stratification) and qualitative research as more ap-
strategy is to look for an intervening variable: one propriate for “micro” issues (such as interactions
that is influenced by the independent variable but between wait staff and their customers). The macro/
in turn has an effect on the dependent variable. For micro distinction can also be discerned in Table 9.1.
example, in a relationship between ethnicity and oc- In the example outlined in Box 14.2, Wajcman and
cupation, education may be an intervening variable: Martin (2002) used a quantitative method in the
form of a questionnaire survey to explore the career
ethnicity → education → occupation
patterns of male and female m anagers. However,
This sequence implies that the variable ethnicity they also carried out qualitative, semi-structured
has an impact on education (for example, ethnic interviews to explore how managers made sense of
groups differ in their levels of education), which in their career patterns in terms of their identity. Thus
turn has implications for the jobs that people in dif- the choice of methods was determined by the re-
ferent ethnic groups attain. Qualitative work could search question.
examine how different ethnic groups perceive edu- The use of multi-strategy research to study differ-
cation, which could lead to a fuller understanding ent aspects of a phenomenon can also be seen in the
of education as an intervening variable. It could family obligations studies conducted by Finch (1985)
also provide leads to a different theoretical perspec- and Mason (1994). Focusing on the distribution
tive on the general topic of ethnicity, education, and within families of the obligation to care for relatives,
occupation. this research had two main data collection elements:
14 | Revisioning Quantitative and Qualitative 339
BOX 14.2 C
ombining survey research and qualitative interviewing
in a study of managers
a survey of a sample of nearly 1000 people by struc- social processes they expressed. (Mason, 1994,
tured interview, and semi-structured interviewing pp. 90–91)
with 88 people. A major component of the survey
interviewing was the use of the vignette technique What were the two sets of research questions to
described in Box 5.4. Mason described the purpose which Mason referred? The survey was designed to
of integrating quantitative and qualitative research elicit information about “‘the proper thing to do’
as follows: for relatives in a variety of circumstances” (Mason,
1994, p. 90). Through the semi-structured inter-
From the beginning . . . we were using the views, Mason tried “to discover what people actually
two parts of our study to ask different sets of did in practice for their own relatives, and also the
questions about family obligations. Not only processes by which they came to do it and make
were we employing different methods to gen- sense of it: did a sense of obligation or responsibility
erate different types of data, but we also antici- have a role in the process? How did people in prac-
pated that these would tell us about different tice work out what to do for their kin, or ask of their
aspects of family obligations. . . . [O]ur view kin?” (Mason, 1994, p. 90).
was that an understanding of kin obligations In multi-strategy research, then, the different
in practice would require an analysis of the re- methods may be geared to addressing different kinds
lationship between the two data sets and the of research questions. The study on the reporting of
340 Part IV Transcending the Quantitative/Qualitative Divide and Some Practical Advice
social science research in the British mass media by reached through method 3 to be elaborated
Fenton et al. (1998), discussed earlier, is a further ex- and more fully understood.
ample of a project that used quantitative and quali- • Questions about reception. For example: How
tative research to answer different types of questions: do readers/viewers interpret media reporting
of social science research? (method 8).
• Questions about coverage. For example: How • Questions about the communication environ-
much coverage is there of social science re- ment. For example: What are the policies of
search? What gets covered? Where? (method 1). universities, government departments, and
• Questions about the production of media cov- funding bodies concerning the media report-
erage. For example: What kinds of attributes ing of research? (method 6).
do journalists look for when deciding whether
to write an item on social science research? This form of multi-strategy research entails
(methods 5 and 7). making decisions about which kinds of research
• Questions about social scientists’ attitudes to question are best answered using a quantitative
media reports of research. This includes re- method and which by a qualitative method. (Box 14.3
porting of research in general (method 2) provides another example.) It also requires that re-
and of their own research (methods 3 and 4). searchers think about how best to interweave the dif-
Method 4 was designed to allow the findings ferent elements, especially since, as was suggested in
BOX 14.3 Q
ualitative and quantitative methods combined to study
children’s knowledge of learning difficulties
Nowicki et al. (2014) combined qualitative and quan- understanding that often cannot be verbalized but
titative approaches in a study that examined how may manifest itself in recognition tasks. To do this
children aged 9–11 viewed the causes of classroom the students were given cards on which were writ-
learning difficulties. The study was conducted at five ten each of the 42 statements. They were asked
schools in a medium-sized Ontario city where stu- to place the cards in groups based on similarity,
dents with learning issues were fully integrated with and to produce labels for each group they had cre-
other students in the classroom. The researchers ated. The cluster analysis showed that five distinct
employed the technique of concept mapping, which clusters of statements could be identified, and that
combines qualitative interviewing with cluster an- the clusters were meaningful, coherent categories
alysis, a statistical procedure. The qualitative inter- that indicated a surprising level of understand-
viewing of individual students formed the first phase ing of learning difficulties. The clusters had the
of the study, with the focal question being “Why do themes fate, family stress, neurological problems,
some children have learning difficulties?” After edit information processing difficulties, and issues re-
ing for redundancy and clarity, this produced 42 lated to motivation and instruction. It turned out
unique statements, or a mean of 1.2 responses per that the students were more astute in identifying
student. This result was in keeping with previous a diverse range of possible causes of learning dif-
qualitative research that indicated that children ficulties when given a list of statements produced
at this age have a very limited store of “explicit” by their peers than they were in verbalizing the
knowledge of learning difficulties, which refers to causes on their own. This study illustrates how
knowledge that can be verbalized on one’s own. combining qualitative and quantitative methods
But the researchers also tested the chil- may produce more informative results than em-
dren’s “implicit” knowledge of the topic, a type of ploying just one of those approaches.
14 | Revisioning Quantitative and Qualitative 341
the context of the discussion about triangulation, the However, multi-strategy research is not necessar-
outcomes of mixed-method research are not always ily superior to one-method or one-strategy research.
predictable. Four points must be borne in mind:
Like unplanned triangulation, this category of
multi-strategy research is more or less impossible • Multi-strategy research must be competently
to plan in advance. Essentially, it offers quantitative designed and conducted: otherwise its findings
researchers an alternative to either revising their hy- will be suspect, no matter how many methods
potheses or filing the inconsistent results away (and are employed.
probably never looking at them again). • Multi-strategy research must be appropriate
for the research questions asked. There is no
point in collecting additional kinds of data on
Reflections on multi-strategy the assumption that “more is better.”
research • Multiple methods are likely to take consider-
There can be little doubt that multi-strategy research ably more time and financial resources than
is becoming more common. Two particularly signifi- research using just one approach, and spread-
cant factors in this development are: ing resources too thinly can dilute the effect-
iveness of the research.
• a growing willingness to see research meth- • Not all researchers have the skills and train-
ods simply as techniques of data collection, ing required to carry out both quantitative and
unencumbered by epistemological and onto- qualitative research, and their “trained incapa-
logical baggage; and cities” may prevent them from integrating the
• a softening in the attitude toward quantitative different forms of research.
research among feminist researchers who in
the past had resisted its use (see Chapter 5 for a Multi-strategy research is not a panacea. It may
discussion of this point). provide a better understanding of a phenomenon
than if just one method is used. It may enhance
Other factors are relevant, but these two are espe- confidence in the findings, and it may even im-
cially important. An example of their operation can prove the chances of access to research settings.
be found in research on audience reception of media Milkman (1997, p. 192), for example, suggested in
and cultural texts. In the past, researchers in the area her research on a General Motors factory that her
relied mainly on qualitative methods (in particular, promise of “hard,” quantitative data facilitated her
focus groups), but lingering unease regarding the entry to the plant even though she had no experience
reliability and generalizability of their findings has with this method. But the general point remains:
led some to recommend using quantitative research multi-strategy research, while offering great poten-
methods in tandem with qualitative methods (for tial in many instances, is subject to constraints sim-
example, Schrøder, 1999). ilar to those in research relying on a single strategy.
mental health among a sample of 302 adult bisex- results, and to explore possible theoretical con-
uals living in Ontario. nections between poverty, bisexuality, and mental
The research grew out of an awareness that health.
previous studies had shown that low socio- In the quantitative part of the study, Ross and
economic status was associated with higher risk her associates found that a sizeable proportion
of negative mental health outcomes, and that (25.7 per cent) of all study participants lived below
people with a minority sexual orientation (gays, the LICO poverty line, and that poverty was asso-
lesbians, and bisexuals) tended to have lower in- ciated with depression and PTSD. Bisexuals living
comes than heterosexuals, although some data in poverty also experienced more discrimination,
indicated that non-heterosexual women earned and after demographic controls were introduced
more on average than heterosexual women. Pre- they were more likely to be open about their
vious research also found that bisexuals were sexual orientation.
more likely to live in poverty than lesbians or The qualitative data analysis gave rise to four
gays, and had more health problems than people models of the pathways by which poverty and
with other sexual orientations. Earlier studies also bisexuality may intersect with mental health.
indicated that bisexuals experienced discrimina- One male participant described being gang
tion in the form of homophobia, heterosexism, raped at age 15, which led to a “lost decade” of
biphobia, and monosexism, with the latter two substance abuse and risky sexual behaviours at
forms of prejudice coming not only from hetero- a stage in his life that might otherwise have been
sexual individuals but also from lesbian and gay devoted to advancing his education and develop-
people. ing his job credentials. This and other accounts of
Ross et al. gathered quantitative data with early life experiences were used by the research-
an online survey that measured whether partici- ers to produce a model in which bisexuality and
pants had incomes below the Low Income Cut- poverty impact adolescent development, which
Off (LICO), a widely used indicator of poverty. The in turn affects both mental health and subse-
survey also included measures of mental health quent poverty. Other participants discussed how
issues such as depression, anxiety, post-traumatic biphobic or homophobic discrimination resulted
stress disorder (PTSD), and suicidality. Variables in job loss, promotion denials, or lack of hiring,
measuring “outness” (how open a person with a which led the authors to develop a theoretical
minority sexual orientation is about their sexual model in which bisexuality affects employment
identity) and perceived discrimination were experiences and earning potential, which then in-
included on the survey as well. In addition, in- fluences mental health in a reciprocal way. Other
formation was collected on the participants’ models explored how bisexuality and poverty
demographic characteristics such as their age, may impede access to social support and mental
sex assigned at birth, gender identity, and racial/ health services, which can contribute to poor
ethnic/cultural identity. mental health.
After the quantitative data were collected, This study provides a good example of how
Ross and her co-researchers acquired a quali- quantitative and qualitative approaches can
tative perspective on the matter by conducting complement each other, and how a combination
semi-structured interviews with a subset of the of the two may be used to generate empirically
survey participants. The interviews were designed based theories that can be the subject of further
to provide a social context for the quantitative research.
14 | Revisioning Quantitative and Qualitative 343
Key Points
• It’s important not to exaggerate the differences methods can be used to analyze the rhetoric of
between quantitative and qualitative research. quantitative researchers.
• Connections between epistemology and ontology • Some qualitative researchers employ quantifica-
on the one hand, and research methods on the tion in their work.
other, are not fixed or absolute. • Although the practice of multi-strategy research
• Qualitative research can exhibit features normally has increased, not all writers support it.
associated with a natural science model. • The view that there are epistemological and onto-
• Quantitative research can incorporate an interpre- logical impediments to the combination of quan-
tivist stance. titative and qualitative research is a barrier to
• The artificial/natural contrast used to distinguish multi-strategy research.
quantitative and qualitative research is often • There are several different ways of combining
exaggerated. quantitative and qualitative research; some can
• A quantitative research approach can be used to be planned in advance, others cannot.
analyze qualitative data, and qualitative research
class, which is invited to ask questions and provide certain psychological traits or demographic
critical commentary on each group’s presentation. characteristics; or generating hypotheses that
could be tested using quantitative methods; and
2. The instructor divides the class into small groups. d. considers the following question: “Does quali-
Each group: tative research differ from quantitative re-
a. selects a general topic that could be re- search in fundamental, irreconcilable ways?
searched using qualitative methods—for ex- Explain.”
ample, how people with eating disorders view
themselves and their relationships with food; The class as a whole then reconvenes. Each
b. produces an outline of how the research could group presents a summary of its discussions to the
be conducted, including the recruitment of in- class, which is invited to ask questions and provide
formants; questions that could be explored in critical commentary on each group’s presentation.
qualitative interviews (if applicable); commen-
tary on whether ethnographic research would 3. A class debate is held to evaluate the following
be appropriate; a consideration of how narra- statement: “Quantitative research differs from
tive analysis might be used; and so on; qualitative research in fundamental, irreconcil-
c. discusses how quantitative methods could be able ways.” The instructor divides the class into
incorporated into the research at each step in three groups. The first group must present argu-
the process—for example, by coding qualita- ments supporting the statement, the second must
tive data; addressing the issue of how prevalent oppose it, and the third can chose to either sup-
certain eating disorders are among particular port or oppose the statement. At the conclusion of
social groups or in the larger society; examin- the debate a class vote is held to determine which
ing whether people with eating disorders share group presented the most convincing case.
Relevant Websites
In this YouTube video, one of this book’s authors (the www.youtube.com/watch?v=65ZlCgGYwXc
late Professor Alan Bryman) discusses multi-strategy
For examples of multi-strategy research, see the
research in the social sciences.
Journal of Mixed Methods Research (to access the
www.youtube.com/watch?v=L8Usq_
journal you may have to go through your institution’s
TPfko&list=PLoem9zOwhTaD3dpk1aCp4e
library server).
OBzFgmOQ6DC
http://mmr.sagepub.com
Professor Ashutosh Varshney speaks about the use
(Websites accessed on 1 November 2018)
of quantitative and qualitative methods in the study
of conflict in this YouTube video.
Chapter Overview
One of the main tasks of any researcher, regardless of the project’s size, is to provide a written account of
the study’s methods, findings, and conclusions. Writing effectively is crucial, because an audience must
be persuaded that the research is credible and important. This chapter explores:
• why writing, and especially good writing, is important to social research;
• how quantitative and qualitative research are typically written up;
• the influence of postmodernism and its implications for writing; and
• key issues in the writing of ethnography, an area in which discussions about writing have been
especially prominent.
As with other skills, the best way to become proficient you will understand the subject matter and the more
at writing up social research is to do it as often as familiar you will become with the structure and flow
you can. And be patient with yourself—even the best of social science writing. In the process, you will come
writers edit and revise their work many times over to recognize that some works are better written than
before they consider it finished. It’s also helpful to get a others and you will begin to absorb the techniques of
knowledgeable person to read what you’ve written and good writing as if by osmosis. Becoming a collector of
offer advice. words also helps. Keep a good dictionary on hand (or
There are structural and technical requirements that online) while you read, and be sure to look up any word
vary depending on the type of research you’ve done, but that you aren’t sure you understand, even if it isn’t a
it’s equally important to pay attention to the quality of technical term. Finally, and perhaps most important, try
your writing as writing. Writing, even the formal variety, to bring your enthusiasm for your topic to the page. If
shouldn’t be thought of as drudgery: consider it an art your exuberance has waned by the time you start writ-
to be cultivated. ing, think about how your project relates to some larger
One of the best ways to become a good writer is to societal or theoretical issue, and keep that in mind as
become a great reader. The more you read, the better you write.
▲ tunart/iStockphoto
15 | Writing Up Social Research 347
Methods and models The final paragraph spells out inferences about
Methods and Models is a very technical section, the impact of social change on a nation’s level of re-
which outlines both the different ways in which ligious belief. The authors suggest that factors such
the relationships between the variables can be con- as modernization and the growth of education de-
ceptualized and the implications of using different press levels of religious belief, resulting in a precipi-
multivariate data analysis methods. tous rather than gradual fall in levels of religiosity.
In their final three sentences, they discuss societies
Results undergoing such change:
In the Results section the authors provide a gen-
eral description of their findings and then consider The offspring of devout families mostly remain
whether their hypotheses are supported; it turns out devout, but the offspring of more secular fam-
they are. The significance of other contextual char- ilies now strongly tend to be secular. A self-
acteristics of nations and individual differences are reinforcing spiral of secularization then sets
discussed separately. in, shifting the nation’s average religiosity ever
further away from orthodoxy. So after gen-
Conclusions erations of stability, religious belief declines
In the final section, Conclusions, Kelley and abruptly in the course of a few generations to
De Graaf return to the issues that have been driving the modest levels seen in many Western na-
their investigation, namely those presented in the tions. (Kelley & De Graaf, 1997, p. 656)
Introduction and Theory sections. They begin with
a strong statement of their findings: “The religious It may be argued that these reflections are some-
environment of a nation has a major impact on the what risky, because the data from which the auth-
beliefs of its citizens: people living in religious na- ors derived their findings are cross-sectional rather
tions acquire, in proportion to the orthodoxy of their than longitudinal. They were clearly extrapolating
fellow citizens, more orthodox beliefs than those from their scoring of the 15 nations in terms of levels
living in secular nations” (Kelley & De Graaf, 1997, of modernization to the impact of social change on
p. 654). They then reflect on the implications of the national levels of religiosity. However, these final
confirmation of their hypotheses for understanding sentences make for a strong conclusion, which may
the process of religious socialization and religious inspire further research.
350 PART IV Transcending the Quantitative/Qualitative Divide and Some Practical Advice
Just as meat tended to imply strongly nega- (Beardsworth & Keil, 1992, p. 290). The sentence
tive connotations for respondents, concepts neatly encapsulates one of the article’s main
like “fruit” and “vegetable” tended to elicit themes—vegetarianism as a response to food
positive reactions, although less frequently ambivalence—and through the reference to “the
and in a more muted form than might have carefully arranged message” alludes to semiotic
been anticipated on the basis of the an- analyses of meat and food.
alysis of the ideological underpinnings of
“wholefoods” consumption put forward by Lessons
Atkinson (1980, 1983), or on the basis of the What lessons can be learned from Beardsworth and
analysis of vegetarian food symbolism ad- Keil’s article?
vanced by Twigg (1983: 28). (Beardsworth &
Keil, 1992, p. 276) • As in Kelley and De Graaf’s paper, strong
opening sentences attract attention and give a
Explaining contemporary good indication of the nature and content of
vegetarianism the article.
The authors then discuss the meaning and signifi- • The rationale for the research is clearly iden-
cance of the findings in light of the study’s research tified, a key point being the paucity of socio-
questions on food selection and avoidance. The re- logical investigations of vegetarianism.
sults are also related to many of the ideas encoun- • Research questions are specified but—in keep-
tered in the two literature sections. The authors ing with the general orientation of qualitative
develop an idea emerging from their research which research—they are somewhat more open-
they call “food ambivalence.” This concept encapsu- ended than in Kelley and De Graaf’s article.
lates the anxieties and paradoxes concerning food The research questions revolve around the
that can be discerned in the interview transcripts issue of vegetarianism as a dietary choice and
(for example, the idea that food can be construed the motivations for that choice.
simultaneously as a source of strength and energy • The research design and methods are outlined.
and as a source of illness). Vegetarianism is in many In fact they are spelled out in greater detail
respects a response to the dilemmas associated with than is usually the case with qualitative re-
food ambivalence. search articles.
• The presentation and discussion of the find-
Conclusions ings are geared to the broad research ques-
In the Conclusions section the authors return to tions that sparked the researchers’ interest in
many of the ideas and themes that drove their re- vegetarianism. The “explaining contemporary
search. They spell out the significance of the idea vegetarianism” section provides an oppor-
of food ambivalence, which is probably the arti- tunity to explore the idea of food ambivalence
cle’s main contribution to research in this area. and its dimensions. The inductive nature of
The final paragraph outlines the importance of qualitative research means that the concepts
food ambivalence for vegetarians, but the authors and theories generated by an investigation
are careful not to imply that it is the sole reason must be clearly identified and discussed, as in
for the adoption of vegetarianism. In the final this case.
sentence they write that “for a significant seg- • The conclusion elucidates in a more specific
ment of the population [vegetarianism] appears way the significance of the results for the re-
to represent a viable device for re-establishing search questions. It also explores the impli-
some degree of peace of mind when contem- cations of food ambivalence for vegetarians,
plating some of the darker implications of the emphasizing one of the article’s major theor-
carefully arranged message on the dinner plate” etical contributions.
15 | Writing Up Social Research 353
Reflexivity has a number of meanings in social sci- ethnicity and ethnic conflict, and her power rela-
ence. To ethnomethodologists it refers to the fact tionship with the Kenyan grandmother she inter-
that speech and action do more than merely act as viewed distorted her research conclusions and
indicators of deeper phenomena (see Chapter 12). interpretations. She made an earnest effort to
Another meaning of the term carries the connota- “peel off [her] own subjectivities” and went back
tion that social researchers should reflect on the five years later to re-interview the woman, which
implications that their methods, values, biases, produced more textured and nuanced insights
and decisions have for the knowledge of the social than her original work. “Self-reflexivity,” she
world that they generate. Reflexivity entails sensi- writes, “is a fitting instrument for researchers
tivity to the researcher’s cultural, political, and to unpack how our own subjectivities impinge
social context. A person conducting a study is not on our framing of a narrative. It also enables
just someone who extracts knowledge from ob- us to examine and debunk our reproduction of
servations and conversations with others and then self-evident ‘truths’ and thus make room for al-
transmits that knowledge to an audience. Rather, ternative readings and . . . polyvocality” (van Sta-
the researcher is implicated in the construction of pele, 2014, p. 16).
knowledge both through the interactions with the There has been a growing trend toward re-
people observed and through the ways in which flexivity in social research, documented in books
an account of those interactions is constructed in that collect details on the inner workings of the
the form of a written text. actual research process as distinct from the often
Van Stapele (2014), for example, came to sanitized versions presented in research articles.
realize that her background as a white, Euro- The confessional tales referred to in Box 15.2 are
pean researcher, her pre-existing notions of manifestations of this development.
In the end, postmodernism produces an acute production of ethnographic texts has become a focus
sense of uncertainty. In questioning how one can of interest in its own right, particularly the rhetorical
ever know or capture a social reality that belongs to conventions employed in their production.
another, it points to an unresolvable tension that af- Ethnographic texts are designed to convince read-
fects ethnographic work in particular. To quote Lin- ers of the reality of the events and situations described,
coln and Denzin (1994, p. 582) again: “On the one and the plausibility of the analyst’s explanations. The
hand there is the concern for validity, or certainty ethnographic text does not simply present a set of
in the text as a form of . . . authenticity. On the other findings: it endeavours to provide an “ authoritative”
hand there is the sure and certain knowledge that all account, with strong claims to truth, of the group or
texts are socially, historically, politically, and cultur- culture in question. Stylistic and rhetorical devices
ally located. Researchers, like the texts they write, are used to persuade the reader to enter into a shared
can never be transcendent.” framework of facts, interpretations, observations,
and reflections. Ethnographic writing, like the writ-
ing found in reports of q uantitative social research,
Writing up ethnography is typically realist in orientation: the researcher tries
The term “ethnography,” as we noted in Chapter 10, to present an authoritative, dispassionate account of
refers to both a method of social research and the an external, objective reality. In this respect, there
finished product of that research. In recent years, the is very little difference between the writing styles
15 | Writing Up Social Research 355
of quantitative and qualitative researchers. Van from the personal subjectivity of the author/ethnog-
Maanen (1988) described ethnography texts that rapher (for example, suspicion that, as a fieldworker,
conform to this model (one of three he identified; see he or she might have become too involved with the
Box 15.2) as realist tales. This is the most common people under study). To this end, when writing up
form of ethnographic writing, though its realism can their ethnographic work, authors play up their aca-
take different forms. Van Maanen distinguished four demic credentials and qualifications, their previous
characteristics of realist tales: experiential authority, experience, and so on. All this enhances the impres-
typical forms, the subject’s point of view, and inter- sion that they are reliable witnesses.
pretive omnipotence. Another way in which authors often seek to
underline their experiential authority is by empha-
Experiential authority sizing the intensity of their research work: they spent
In ethnography, as in much quantitative research so many months in the field, interviewed countless
writing, the author disappears from view. Readers are individuals, worked hard to establish rapport, and so
told what the members of a group—the only people on. Drawing the reader’s attention to the hardships
directly visible in the text—say and do. An invisible of the fieldwork—the dangerous setting, the poor
author provides the narrative, giving the impression food, the disruptive effect on normal life, the feel-
that the findings presented are what any reasonable, ings of isolation and loneliness, and so on—creates
similarly placed researcher would have found. Read- sympathy for the author or (less charitably) deflects
ers have to accept that this is what the ethnographer potential criticism of the findings.
experienced. This strategy essentially plays down the Also worth mentioning are the extensive quo-
risk that the findings will be suspected of bias arising tations from conversations and interviews that
Van Maanen (1988) distinguished three major with presenting findings than with detailing how
types of ethnographic writing: research is carried out. Very often the confes-
• Realist tales: apparently definitive, confident, sional tale is published separately from the pro-
and dispassionate third-person accounts of a ject’s findings: for example, it may appear as a
culture and the behaviour of its members. This is chapter in a book of similar tales, while the main
the most prevalent form of ethnographic writing. findings are written up in realist tale form.
• Confessional tales: personalized accounts in • Impressionist tales: accounts that feature
which the ethnographer is fully implicated in “words, metaphors, phrasings, and . . . the ex-
the data-gathering and writing-up processes. pansive recall of fieldwork experience” (Van
These are warts-and-all accounts of the trials Maanen, 1988, p. 102). There is an emphasis
and tribulations of doing ethnography. They on stories of dramatic events that provide “a
have become more prominent since the 1970s representational means of cracking open the
and reflect a growing emphasis on reflexivity in culture and the fieldworker’s way of knowing
qualitative research. Several of the sources re- it” (Van Maanen, 1988, p. 102). However, as Van
ferred to in Chapter 10 are confessional tales (for Maanen noted, impressionist tales “are typ-
example, Armstrong, 1993; Giulianotti, 1995). ically enclosed within realist, or perhaps more
However, confessional tales are less concerned frequently, confessional tales” (1988, p. 106).
356 PART IV Transcending the Quantitative/Qualitative Divide and Some Practical Advice
invariably form part of the ethnographic report. Events I witnessed or took part in ranged from
These quotations serve not only as evidence in sup- the very routine (sitting around drinking coffee
port of the author’s findings, but also to establish the and eating junk food) to accompanying various
credibility of the report in that they demonstrate the women on visits to . . . the HIV clinic; I accom-
author’s ability to engage people in talk. Copious de- panied them when they were in court, and even
scriptive details—of places, patterns of behaviour, went flat-hunting with one woman. I went shop-
contexts, and so on—suggest that the author was ping with some, helping them choose clothes
ideally placed to bear witness to the people and ac- for their children and presents for their friends.
tivities described in the study. I visited them in their homes, rehabilitation
centres, and maternity wards, sat with them
Typical forms through withdrawals, watched them using
Authors will often generalize about recurring fea- drugs, and accompanied them when they went
tures of the group under study in order to illustrate “scoring” (buying drugs). (Taylor, 1993, p. 16)
a typical behaviour or thought pattern. Examples
based on particular incidents or people may be used, Interpretative omnipotence
but basically the emphasis is on recurrent forms of When writing ethnography, authors rarely suggest
behaviour. For example, in her conclusion to her alternative interpretations of an event or pattern of
ethnographic research on female drug users, cited behaviour. Instead, they tend to portray the phenom-
several times in Chapter 10, Taylor wrote: enon in question as having a single meaning or sig-
nificance, which only they have been able to identify.
Yet the control exercised over women through The evidence is carefully marshalled to support the
the threat to remove their children highlights singular interpretation placed on the event or pat-
a major factor differentiating female and male tern of behaviour. It seems obvious or inevitable that
drug users. Unlike male drug users, female anyone would draw the same inferences the author
drug users, like many other women, have two drew when faced with such evidence.
careers: one in the public sphere and one in the The above characteristics imply that the research
private, domestic sphere. (1993, p. 154) process is only the first stage in the creation of a con-
vincing realist tale. Persuading readers that one has
This statement portrays a pattern among women in fact figured out the nature of a different culture
drug users, making any individual woman import- also requires that the written ethnography represent
ant only insofar as she can be seen as illustrating the the fieldwork in an effective way. Recall that for post-
general tendency. modernists, any realist tale is merely one “spin”—one
interpretation that has been formulated in relation to
The subjects’ points of view the culture in question.
A commitment to seeing through the eyes of their
subjects is important for qualitative researchers be-
cause it is part of their strategy for getting at the
Practical Tip | Referring to
meaning of those people’s social reality. However, it
websites
is also helps to establish the authority of the ethnog-
rapher. Claiming to have taken the subjects’ points If you include a website in your bibliography,
of view implies that the author can speak authori- the entry should contain the date it was last
tatively about the group in question. Ethnographies consulted. This is done because someone who
frequently include numerous references to the steps reads your work and then checks out the site
taken by the ethnographer to get close to the people may find that it is no longer there or that it has
under investigation, and their success in that regard. changed.
In her research on female drug users, Taylor wrote:
15 | Writing Up Social Research 357
Key Points
• Good writing is probably just as important as good there is no external social reality. It is intended
research practice. Indeed, it should probably be merely to point out that readers’ understanding of
considered part of good research practice. that reality is profoundly influenced by the way the
• In writing up research, a clear statement of the re- writer represents it.
search questions and an effective structure for the • While postmodernism in particular has promoted
report are essential. recognition of this last point, writers working
• The writings of social scientists are designed to do within other traditions have also drawn attention
more than simply report findings; they must also to it.
convince and persuade. • Quantitative and qualitative research articles are
• Our emphasis on the importance of writing to broadly similar in structure, as are the rhetorical
convince and persuade is not meant to imply that devices used to produce them.
approached. Hypothetical results and conclusions café. The class is shown the picture or clip, and each
are given for points “f” and “g.” student then writes up a description of the physi-
Once that is done, students leave their groups cal setting that is depicted (maximum 250 words).
and the class is reconvened. Each person in the Three volunteers then read their descriptions to the
class is given 15–20 minutes to write a brief Intro- rest of the class. Next, the instructor facilitates a
duction to their group’s article. The Introduction discussion based on the following questions:
should not exceed 250 words and should be writ-
ten in such a manner that it attracts the reader’s a. How are the three descriptions similar? How
attention, gives a clear indication of the article’s are they different?
focus, and highlights the significance of the find- b. Why are the three descriptions not identical?
ings. When completed, the instructor asks for three How would you account for the differences?
volunteers who are willing to give a brief verbal c. What is “interpretive omnipotence”? Do
synopsis of their group’s study and have their any of the descriptions assume interpretive
Introduction read by the rest of the class. Volun- omnipotence?
teers with good keyboard skills are then asked to d. Which of the descriptions is most realistic?
transcribe the three Introductions into electronic e. Can one ever determine whether a particu-
format so they can be shown on the classroom lar description is more realistic than another?
screen. The first of the three volunteers then gives Does the term “realistic” even have a definite
a synopsis of their group’s study, and their Introduc- meaning?
tion is shown on the screen. The instructor then fa- f. Would the person or persons depicted in the
cilitates a discussion of the Introduction. The other picture or clip describe the setting in a way
two Introductions are presented in the same way. that differs from how the three volunteers de-
In an alternative version of this exercise that scribed it? How so? Why would the person or
would be appropriate for small classes, the in- persons depicted have a different description?
structor collects all the Introductions that the g. How could you learn about how the people in
class produced and provides written comments on the picture or clip view the setting?
each one, returning them in a later class. h. If multiple interpretations of a physical setting
are possible, would interpretations of people’s
2. Prior to class, the instructor selects a picture or a actions and experiences in the setting be even
brief video clip of a person or persons in a physical more numerous and variegated? Explain.
setting that would be appropriate for qualitative i. What are the implications of the responses to
research—for example, a market, a sports facility, a the previous question for qualitative research?
Relevant Websites
The London School of Economics and Political Sci- In this YouTube video, Jackie Hammill of the Univer-
ence Impact Blog provides tips on successful aca- sity of Prince Edward Island outlines how to organize
demic research and writing. your activities when starting the research for a paper
http://blogs.lse.ac.uk/impactofsocialsciences/2012 for a university course, and how to write things up as
/11/28/lupton-30-tips-writing you go along.
www.youtube.com/watch?v=BuJLRjd
Nick Fox of the University of Sheffield gives advice on
9vAc&NR=1
writing up a qualitative study.
www.academia.edu/3073153/How_to_write_ If you found the above video helpful, you may want to
and_structure_a_qualitative_paper_Powerpoint_ look at the next one in the series as well. This YouTube
2013_
15 | Writing Up Social Research 359
Chapter Overview
The goal of this chapter is to guide you through a small-scale research project from beginning to end. The
guidance offered goes beyond the discussions presented in previous chapters in that it focuses on the
practical aspects of doing research. A wide variety of issues is explored, including:
• time management when doing research;
• generating research questions; and
• dealing with existing literature on your subject.
When one of the authors of this book was a graduate and motivation will pick up even if they were minimal
student and had his first meeting with his dissertation when you began. It also helps if you think of the project
advisor, the advisor listened to his research proposal, as a series of small jobs rather than one monumental
offered a few suggestions, and then said, somewhat undertaking.
acerbically, “Just do it!” (This was years before that The same goes for the writing. Try to write some-
phrase became a marketing slogan.) thing related to the project—anything—every day.
Those three words contain a lot of wisdom. It’s easy And for your first draft, allow yourself to be sloppy.
to overthink things and waste time worrying that your Don’t worry about finding the perfect word or crafting
plans aren’t worthwhile or that you’re not up to the the perfect sentence or ensuring that the logic flows
task. Just assume that with proper guidance and a lot flawlessly. Those issues can be dealt with later, at the
of hard work, you’ll be just fine. But you must follow editing stage. When Michael Jackson was asked about
through with concrete actions, every day. Dreaming is how he wrote his songs, he replied, “Don’t get in the
not enough. way of the music.” That advice applies to completing a
Don’t wait for the perfect time to start. There is no research project too: don’t get in the way of the ideas.
such thing. Imagine you’re at work and your boss has Just let them flow. Write first and evaluate later. Just
given you an unbreakable deadline. You’ll find that do it.
once you get absorbed in the task, your inspiration
▲ Geber86/iStockphoto
16 | Conducting a Research Project 361
Research area
(Example: Concerns about risks to personal safety)
for research, talk things over with your advisor or greatly. Of course, students also vary in what they
course instructor, or take a look at the list of sources ask of their advisors. Our advice here is simple: use
for research questions provided in Box 16.1. supervisors or advisors to the fullest extent allowable
and give due consideration to their suggestions. Such
people are usually well versed in the research process
Using a supervisor and can provide help and feedback at all stages of the
Most institutions will assign a student writing a thesis project, subject to your institution’s strictures in this
or dissertation to a supervisor or advisor. However, the regard. If they criticize your research questions, inter-
kind and amount of assistance you can expect varies view schedule, early drafts, or whatever, try to respond
Marx (1997) suggested the following possible develop some research questions pertaining to
sources of research questions: the phenomenon in question, including ques-
tions about the validity of the conventional
• intellectual puzzles and contradictions
wisdom on the topic.
• the existing literature
• the counterintuitive; for example, cases in which
• replication
social scientific truths seem to fly in the face of
• a feeling that a certain theoretical perspective or
common sense, as they did when a doomsday
notable piece of work is misguided (in this case you
cult began to proselytize rather than slink away
would explore the reasons for the shortcomings)
in embarrassment when “doomsday” passed and
• a social problem in need of a social scientific
life carried on as usual (Festinger et al., 1956)
(for example, sociological) explanation
• “empirical examples that trigger amazement”
• “gaps between official versions of reality and
(1997, p. 114); for example, atypical events,
the facts on the ground” (1997, p. 113). Exam-
cases that contradict widely accepted theories,
ine the standard explanations offered by gov-
and so on
ernments, the media, or social researchers for
• new methods and theories—how can they be
social phenomena such as poverty, substance
applied in new settings?
abuse, international conflict, or anything else
• “new social and technical developments and
of interest. Are those explanations consistent
social trends” (1997, p. 114)
with what you know or have experienced? If not,
16 | Conducting a Research Project 363
positively. Criticism isn’t a personal attack. Such com- can your institution help with the cost of
ments are usually accompanied by the reasons for photocopying, postage, stationery, and so on?
them and some suggestions for revision. Be thankful Can it lend you hardware to record and tran-
for the opportunity to address problems before your scribe interviews? Does it have the necessary
work is formally examined. Supervisors have to go software, such as IBM SPSS Statistics (SPSS)
through the same process themselves whenever they or NVivo? This kind of information will help
apply for a research grant or submit an article or book you decide whether your research design and
for peer review. methods are financially feasible and practical.
Students who get stuck or fall behind in their The fictional “gym study” in Chapter 8 is an
work will sometimes avoid their supervisors. But the example of a project feasible within the kind of
longer they wait, the harder it is to ask for help. Then, time frame usually allocated to undergraduate
because the work has been put off, it has to be rushed projects. However, it would require resources
at the end, and in some cases is never completed. Try for photocopying covering letters and ques-
to avoid this situation by confronting any difficul- tionnaires; postage both for sending the ques-
ties head-on and getting advice on how to deal with tionnaires out and for follow-up letters; and
them as soon as possible. return postage for the questionnaires. In addi-
Undergraduate students doing course projects tion, quantitative data analysis would require
should follow the same basic advice. Consult your software such as SPSS.
course instructor if you run into difficulties, and the
sooner the better. Even if there are no apparent prob-
lems, it’s always worthwhile to discuss your research
Searching the existing
with your instructor, who may pick up on issues that literature
you have missed. Online bibliographical databases (accessible at most
university libraries) are an invaluable source of refer-
ences. The best one for sociology students is probably
Managing time and resources: Sociological Abstracts; databases for other disciplines
Start thinking early about the are available as well. They normally allow searches
by keyword (topic), author, title, journal, and other
research area descriptors. The focus is easily narrowed by speci-
Students are usually asked to start thinking about a fying language, year, and type of presentation. For
potential research topic well before they are expected example, “English, post-2010, journal articles only”
to start work on the project. That thinking is worth can be searched and listed from latest to earliest. The
doing, because all research is constrained by time. “journal articles only” restriction removes papers
Two pieces of advice are relevant here: presented at meetings and dissertations, which are
too hard for most students to access in any event.
• Work out a timetable, preferably with your Searches can be conducted using keywords in differ-
supervisor or instructor, detailing the differ- ent combinations with the Boolean operators “and,”
ent stages of the research (including the litera- “or,” and “not.” Thus a search for material on “sexism
ture review and the writing up). The timetable and universities” will produce articles that deal with
should specify the different stages and the both topics in the same piece. Choosing “sexism or
specific dates for starting and finishing them universities” will produce those articles plus arti-
(see Figure 16.2 for an example of a quantita- cles on each one alone. That is probably not a good
tive research outline). Some stages are likely to choice—it’s too broad. “Universities and sexism
be ongoing—for example, searching the litera- or prejudice” is better because it expands the first
ture for references (see below)—but that should option above to cover university prejudice in addi-
not delay the development of a timetable. tion to university sexism. Finally, “not” restricts the
• Find out what resources, if any, are available search. “Sexism not universities” will find articles
for carrying out your research. For example, about sexism in places other than universities. You
364 PART IV Transcending the Quantitative/Qualitative Divide and Some Practical Advice
Issue(s) to be researched
Consider whether a social survey is appropriate (if not, consider an alternative research design)
Revise questions
Finalize questionnaire/schedule
Analyze data
Interpret findings
Consider implications of findings for research questions, existing theories, future research,
and the discipline as a whole.
can experiment with the use of keywords and Bool- your credibility as someone knowledgeable in your
ean options; if you don’t find what you need, ask a chosen area.
librarian for help. As you explore the existing literature, look for an-
Following these procedures will produce a list swers to the following questions:
of journal articles in which the keyword(s) ap-
pears, with full citation details and an abstract, • What is already known about this area?
and sometimes a link to a full-text download. • What concepts and theories are relevant?
Other databases are more specialized: examples • What research methods and strategies have
include the Canadian Periodicals Index Quarterly been employed in studying this area?
(formerly the Canadian Periodical Index), the LGBT • Are there any significant controversies?
Life with Full Text, Family & Society Studies World- • Are there any inconsistencies in the findings
wide, and Contemporary Women’s Issues. Mark relating to this area?
the records you want and email them to yourself • Are there any unanswered research questions?
for a permanent record. The bibliographies in the
articles you find can provide additional sources While reading, take good notes, and be sure to
to examine. record full citations for all the materials you consult.
A particularly interesting tool is the Social Sci- It’s infuriating to find that you’ve forgotten some
ences Citation Index, which looks forward from the detail, such as the volume number of an article, when
original article to later ones that cite it, revealing what you come to assemble your bibliography. Also keep
others have thought of it. Perhaps some who cited the in mind that your written review of the literature
article were critical; perhaps others accepted or ex- should be critical as well as descriptive. How does a
panded on it. In this way a history of the research particular article relate to others you have read? Does
in the area of interest can be built up. To some pro- it contradict them? Are there any apparent strengths
fessors, this index provides a measure of how often or deficiencies—for example, in the methods used or
they have been cited. To those granting tenure and the conclusions drawn? What theoretical ideas influ-
raises, it can be a crude indicator of merit, following enced the author(s)?
the logic of “more citations, better researcher.” In some areas of research you will find huge num-
The library catalogue is an obvious tool for find- bers of references. Try to identify the major ones
ing books, even those held at other universities. and work outward from there. Move on to the next
Again, ask a librarian if you run into difficulty. How- stage of the research at the point you identified in
ever, since books are often dated, checked out, or your timetable (see above). This is not to say that
too specialized for student papers, the above sources your literature search will cease: only that you need
may be preferable. The rule of thumb is similar to the to force yourself to move on. Ask your supervisor or
idea of theoretical saturation in qualitative research: instructor for advice on whether you need to extend
stop searching when the same items keep appearing. your search of the literature.
Why review the existing literature? The most ob-
vious reason is to find out what is already known
about your area of interest, in order to avoid “re-
Preparing for research
inventing the wheel.” In addition, the process of In choosing a methodology you need to consider its
reviewing the literature will help you revise and ethical implications and what you will do to prevent
refine your research questions. Beyond this, the ex- any potential ethical violations. Keep in mind the
isting literature on your topic will help you develop various ethical issues discussed in Chapter 3 and
an argument regarding the significance of your re- flagged throughout this book.
search and where it will lead. The metaphor of a story Don’t begin collecting data until you have clearly
is sometimes used in this context (see below). A com- identified your research questions, and design your
petent review of the literature will help to establish data collection instruments with those questions
366 PART IV Transcending the Quantitative/Qualitative Divide and Some Practical Advice
in mind. If at all possible, conduct a small pilot study in rushed, last-minute writing. Writing under that
to see how well your research instruments work. You kind of pressure is not ideal, to say the least. How you
will also need to think about access to research sites. present your findings and conclusions is a crucial
If the research requires entry to a closed setting such aspect of the research process, and if you fail to pro-
as an organization, you need to get permission at the duce a convincing account of your findings because
earliest opportunity. This usually takes so long that of time pressures, you will not do your work justice.
many advisors do not recommend using closed set-
tings for student research. You also need to consider Get feedback
how you will go about gaining access to people. This Try to get as much feedback on your writing as pos-
issue leads you into sampling considerations, such as sible, and respond positively to whatever comments
the following: you receive. Your supervisor or instructor is likely to
be the main source of feedback; show as many drafts
• Whom do you need to study in order to answer of your work as regulations allow, leaving plenty of
the research questions? time for a response. You will not be the only student
• Is an adequate sampling frame available? seeking advice, and if comments are rushed, they
• What kind of sampling strategy is feasible— may not be particularly helpful. You can also ask
probability, quota, theoretical, convenience? others in your program or class to read your drafts
and comment on them, and they may ask you to do
the same. Their comments may be very useful, but
Writing up research those from your supervisor or instructor are likely to
Once the data gathering and analysis have been com- be the most valuable.
pleted, the findings must be conveyed to an audience.
The first bit of advice is to . . . Avoid sexist, racist, and other
prejudicial language
Start early Remember that writing should be free of sexist,
There are good reasons for beginning the writing racist, and other prejudicial language. The Social Sci-
process early on. It forces you to think about issues ences and Humanities Research Council of Canada
such as how best to present and justify the research publication On the Treatment of the Sexes in Research
questions, and how to structure the discussion of the (1985), by M. Eichler and J. Lapointe, is helpful.
literature cited. A second reason is entirely practical:
many people find it difficult to get started and tend Structure your writing
(often unwittingly) to use procrastination strategies. A 10,000- to 15,000-word research project entailing
But putting off the inevitable almost always results data collection is typically structured as follows.
16 | Conducting a Research Project 367
In 2002, a 19-year-old sociology student was sup- imminent (for example, through body language),
posed to interview a homeless person in Man- withdraw from the situation. Further guidelines on
chester. Because of safety concerns, her advisor these issues can be found in Craig (2004).
recommended that she take a friend with her and Sometimes researchers may be unaware of
conduct the interview in a public place. When she possible hazards. Lankshear (2000), for instance,
did not return, she was reported missing (Barkham found out only after her study had begun that her
& Jenkins, 2002). Four days later she showed up in research in a hospital laboratory could expose
Dublin, where she claimed to have fled after the her to dangerous pathogens. Be sure to do a thor-
man robbed her; however, it was later suggested ough risk assessment of the research environment
that her disappearance might have been sparked before you undertake your project.
by a family argument. Finally, once you’ve assessed the risk and
Whether or not her story was true, there is an are sure that you will not be placing yourself in
important lesson in it: social research can place harm’s way, it may be necessary to convince your
you in potentially dangerous circumstances. Avoid instructor, advisor, or research ethics board (REB)
situations where personal harm is a real possibil- that your safety will not be compromised. Small
ity. Just as you should ensure that no harm comes et al. (2014), who did ethnographic research with
to research participants, individuals conducting people in Vancouver who inject themselves with
research should not place themselves in unsafe illegal drugs, were told by their REB that they had
situations. The advice given by the student’s to consult with local police about safety issues and
advisor—to take someone with you and conduct provide police with the names of the ethnograph-
the interview in a public place—is sensible. Keep ers who would be observing the drug users. The
your cellphone nearby and switched on. Establish researchers had to go to great lengths to convince
a routine whereby you keep in regular contact the REB that proper precautions would be taken
with friends or family members. Even in seem- and that they would not be in danger, and that
ingly safe conditions, a researcher can be faced involving the police in their safety would make it
with a sudden outburst of abuse or threatening very difficult to recruit participants and ensure the
behaviour; you can’t always predict how people confidentiality of the results. Eventually the board
will react to a particular trigger, such as an inter- relented and the study went ahead as planned,
view question. If there are signs that trouble is but only after a prolonged review process.
• Additional reflections on producing a literature achieved (if relevant); specific procedures used (for
review are presented in Box 16.3. They were example, in the case of an online questionnaire,
derived from a review of qualitative studies of whether and how non-respondents were contacted
organizations, but they also apply to quantita- for follow-up); and, where relevant, the nature of
tive research. the questionnaire, interview schedule, partici-
pant observation, observation schedule, or coding
Research methods frame. Although copies of these usually appear in
The term “research methods” is a kind of catch- an appendix, you should comment on matters such
all for several things that need to be outlined: re- as your style of questioning or observation and
search design; sampling approach; how access was why you asked the questions you did. Other issues
In their examination of journal articles reporting Each of these strategies is designed to leave
qualitative research on organizations, Golden-Biddle room for the writer to make his or her own contri-
and Locke (1993; 1997) argued that good articles in bution to the field.
this area develop a story; that is, a clear and com-
• Problematizing the situation. The literature
pelling framework around which the writing is
is then criticized by identifying a problem, for
structured. This idea is very much in tune with Wol-
example:
cott’s (1990, p. 18) advice on writing up qualitative
»» Incomplete. There is a gap in the existing
research: “determine the basic story you are going
literature.
to tell.” Golden-Biddle and Locke suggest that es-
»» Inadequate. The existing literature has over-
tablishing how your research relates to the existing
looked useful ways of looking at the phenom-
literature on the topic is an important component of
enon; alternative perspectives or frameworks
storytelling. They identified two things that are done
are then introduced.
to effectively review the literature and convey the
relevance and significance of your research: According to Golden-Biddle and Locke, the au-
thors they studied used their accounts of the litera-
• Constructing inter-textual coherence. The
ture in various ways:
author shows how existing contributions to the
literature relate both to each other and to the re- »» They demonstrated their knowledge and
search that will be reported. Golden-Biddle and competence by referring to prominent writ-
Locke distinguished three approaches to the co- ings in the field.
herence question: »» They developed their review of the litera-
»» Synthesized coherence. The author pieces ture in such a way as to highlight the contri-
together theory and research previously re- bution they would make in the article.
garded as unconnected. »» They explained how the gaps or problems
»» Progressive coherence. The author traces they identified in the literature corres-
the building of consensus in a particular ponded to their research questions.
area of knowledge.
The idea of writing up research as storytelling
»» Non-coherence. The author emphasizes dis-
serves as a useful reminder that reviewing the lit-
agreements in the contributions to a certain
erature, which is part of the story, should link seam-
research program.
lessly with the rest of the article.
370 PART IV Transcending the Quantitative/Qualitative Divide and Some Practical Advice
Checklist
Checklist of issues to consider for writing up research
☐☐ Is there a good correspondence between the ☐☐ Do the conclusions clearly establish what
title of the project and its contents? your research contributes to the literature?
☐☐ Have you clearly specified your research ☐☐ Have you explained the limitations of your study?
questions? ☐☐ Do your conclusions consist solely of a sum-
☐☐ Have you clearly linked the literature cited to mary of findings? If so, rewrite them, explain-
your research questions? ing their significance.
☐☐ Is your discussion of the literature critical and ☐☐ Do the conclusions provide clear answers to
not just a summary of what you have read? your research questions?
☐☐ Have you clearly outlined your research ☐☐ Have you broken up the text in each chapter
design and research methods? Make sure you or section with appropriate subheadings?
have explained: ☐☐ Have you provided signposts so that readers
• why you chose a particular research design know what to expect next, and why it is there?
or method; ☐☐ Does the writing avoid sexist, racist, and other
• why you implemented your research design prejudicial language?
in the way you did (for example, how the ☐☐ Have you checked to ensure that the text
interview questions relate to the research does not make excessive use of jargon?
questions, or why you observed people in ☐☐ Have you included all the necessary appendi-
particular situations); ces (for example, interview schedule, letters
• how you selected your research participants; requesting access, communications with re-
• whether you experienced any problems with search participants)?
cooperation (for example, response rates);
☐☐ Does the list of references include all the
• if your research required access to an or-
items referred to in your text?
ganization, how and on what basis the
agreement was achieved; ☐☐ Does the format of the references follow pre-
cisely the style required by your institution or
• any difficulties encountered in imple-
instructor?
menting your research;
• the steps taken to ensure that the research ☐☐ Have you met all requirements regarding
was ethical; and matters such as word count (so that the paper
• how the data were analyzed. is neither too long nor too short) and inclusion
of an abstract and table of contents?
☐☐ Have you presented your data in such a way that
they clearly relate to your research questions? ☐☐ Have you ensured that you do not quote ex-
cessively when presenting the literature?
☐☐ Are your interpretations of the data fully sup-
ported with tables, figures, or segments from ☐☐ Have you fully acknowledged the work of others,
transcripts? so that you cannot be accused of plagiarism?
☐☐ Have you acknowledged, preferably in an
☐☐ Is every table and/or figure properly labelled
Acknowledgments section, the help of others
with a title and number?
where appropriate (for example, your super-
☐☐ Is every table and/or figure commented on in visor, people who helped with interviews,
the text? people who read drafts)?
☐☐ Does the discussion of the findings relate to ☐☐ Finally, have you checked to make sure that
the research questions? your tenses, margins, pagination, and capital-
☐☐ Does that discussion show how the findings ization are consistent? Many forms may be
shed light on the literature presented? acceptable, but all must be used consistently.
372 PART IV Transcending the Quantitative/Qualitative Divide and Some Practical Advice
Relevant Websites
Databases such as Sociological Abstracts, PsycINFO, In this YouTube video, Jared Wright walks you through
Worldwide Political Science Abstracts, Social Work the steps that need to be taken to write a sociological
Abstracts, the Canadian Periodicals Index Quarterly, research paper.
LGBT Life with Full Text, Family & Society Studies World- www.youtube.com/watch?v=n-5xIA71aRQ
wide, Contemporary Women’s Issues, and the Social Sci-
In this YouTube video, David Taylor discusses how to
ences Citation Index (through the Web of Knowledge)
choose and narrow down a research topic.
are all accessible online. To use them, log on to your
www.youtube.com/watch?v=jSHXb83Xtsk
institution’s library server and call them up from there.
(Websites accessed 2 November 2018)
Chapters 8 and 13 described a number of methods variables are in bold italics (e.g., gender or reasons).
for analyzing quantitative and qualitative data. Labels (longer versions of names; see pp. 376–77)
This appendix illustrates how to perform those an- given to variables or values are in bold but are not
alyses using IBM SPSS Statistics (SPSS) and NVivo italicized (e.g., reasons for visiting or male). A right-
software.* pointing arrow (→) means “left-click once with the
mouse” to make selections. Box A.1 presents a list of
basic operations in SPSS.
Quantitative data If you have SPSS software loaded onto the hard
analysis using SPSS disk of the computer you are using, click on the SPSS
The gym survey described in Chapter 8 will be icon on the Desktop screen or find the program
used to illustrate how SPSS software is used. Learn- using the Start button in the bottom left-hand corner
ing new software requires some perseverance, but of the screen. If you don’t have SPSS software on your
is well worth the effort. It would take far longer to computer, it may be available from a special server
perform the calculations by hand (even on the small provided by your university.
gym survey sample of 90) than to learn the soft- After SPSS loads, an opening dialogue box with
ware. With more advanced techniques and larger a number of options may appear. If so, for the pur-
samples, the time saved is even more substantial. poses of this exercise disable it: → Close. This will
Imagine calculating the mean age of 2500 people give you the SPSS Data Editor, made up of two com-
by hand! ponents: Data View and Variable View. Move be-
tween these two by selecting the appropriate button
at the bottom left of the screen. The Data View pro-
Getting started in SPSS vides a spreadsheet grid into which data are entered
(see Plate A.1). The columns represent variables—in
Introduction our example, information about the characteris-
SPSS is the software most widely used for this kind tics of each person in the gym study sample, such
of analysis, probably because it is relatively straight- as gender and age. Until data are entered, each
forward. It has been in existence since the mid-1960s column simply has var as its heading. The rows
and over the years has undergone many revisions, represent cases, which can be people (as in this ex-
particularly since the arrival of personal computers. ample) or other units of analysis. Each block in the
The version used in preparing this section is IBM grid is referred to as a cell. Note that when the data
SPSS Statistics 25. are entered in the SPSS spreadsheet, they will look
SPSS operations and other on-screen terms are different: for example, 1 becomes 1.00 (this can be
presented in bold (e.g., Analyze). The names given to adjusted if necessary).
* All screen captures in this appendix are reprinted Courtesy of International Business Machines Corporation, © International
Business Machines Corporation.
374 Appendix | Using IBM SPSS Statistics and NVivo Software
• IBM SPSS Statistics Data Editor. This is the mode • Value Label. This is a label attached to the
in which data may be entered and subsequently code used when entering the data. Thus for
edited and defined. It is made up of two screens: var00001, the label male can be attached to
Data View and Variable View. Move between 1 and the label female to 2. In output, such as
these two views by selecting the appropriate a frequency table or chart, the labels for each
button at the bottom left of the screen. value appear, making the interpretation of
• Data View. This is the spreadsheet into which the output easier than if the code were used. It is
data are entered. It appears when SPSS is started. generated from the Variable View.
• Variable View. Another spreadsheet, this one • Missing Values. If data for a particular variable
displays information about each of the vari- are missing, you must specify how missing
ables (such as the variable name, the variable values are coded so that the computer can omit
labels, and value labels) and allows that infor- them from the calculations. Missing values are
mation to be changed (see below). generated from the Variable View.
• IBM SPSS Statistics Viewer. After a statistical • Analyze. This is the button on the menu bar
analysis or the production of a graph (some- along the top of the Data Editor from which
times called a “chart” in SPSS), the output is (via a drop-down menu) the method of analysis
deposited here. The output window super- is selected. Note that whenever a menu item
imposes itself over the Data Editor after an an- appears with a right-pointing arrowhead after
alysis has been performed or a graph has been it, a further sub-menu is available.
generated. • Graphs. This is the button (at the top of the Data
• Variable Name. This is the name given to a vari- Editor) used to get access to the Chart Builder
able (for example, gender). Until they are given (via a drop-down menu), which can be used to
names, variables are referred to as var00001, create various graphs and charts.
var00002, and so on. Once a variable is given a • Chart Builder. This feature is used to make
name, the name appears in the column for that charts and graphs. A very useful tutorial is
variable in the Data View window. It is gener- available from the Data Editor by clicking Help
ated from the Variable View. → Topics. Use the search engine to find Tutorial
• Variable Label. This is a label attached to a → Creating and editing charts.
particular variable, and is optional. It is usually • Chart Editor. This is used to make various
longer than the variable name and therefore changes to graphs. To activate it, double-click
more explanatory. Spaces can be used (for ex- anywhere in the graph. A small Chart Editor
ample, reasons for visiting). The label appears window appears along with a version of the
in any output generated and comes from the main graph and stays until the Chart Editor is
Variable View. exited.
Entering data in the Data View number in that cell—that is, 1. This number goes
To input the data given in Box 8.2 into the Data directly into that cell and into the box beneath the
View, make sure that the top left-hand cell in the toolbar. Although you can use the mouse, many
grid is highlighted (Plate A.1). If it is not, simply people find it easier to use the arrow keys on their
click once in that cell. Then type the appropriate keyboard to move from cell to cell. If you make
Appendix | Using IBM SPSS Statistics and NVivo Software 375
a mistake, simply click once in the cell in ques- data from the gym survey entered (though only
tion, type in the correct value, and hit the Enter part of the data set is visible). The first row of data
key. The last piece of data goes into the bottom contains the coded answers (provided by the first
right-hand cell of what will be a perfect rectangle respondent) from the completed questionnaire in
of data. Plate A.2 shows the Data View with the Box 8.1.
To proceed further, SPSS works in the following Defining variables: Variable names,
typical sequence for selecting variables and analyzing missing values, variable labels, and
data. value labels
1. Make a selection from the menu bar at the top Once you have finished entering the data, prepare the
of the screen, for example, → Analyze. variables. The following steps explain how this is done:
2. From the menu that will appear, make a selec-
tion, for example, → Descriptive Statistics. 1. → Variable View tab at the bottom left of the Data
3. This will bring up another menu where you Editor [opens the window shown in Plate A.3]
will specify which statistics you want, for ex- 2. To name a variable, click on the current variable
ample, Frequencies. name (for example, var00003) and type a name
4. This will bring up a dialogue box in which you for it (for example, reasons). This name, which
provide further information, such as the vari- cannot include spaces, is the identification
ables to be analyzed. needed for instructing the computer to perform
5. Very often, further information is needed; → a an analysis with this variable. It can be any-
button that brings up a sub-dialogue box. thing you like, but you should choose a name
6. Provide the information in the sub-dialogue that indicates what the variable is measuring.
box and then go back to the dialogue box. 3. To give the variable a more detailed designa-
Sometimes, a further sub-dialogue box is re- tion, known in SPSS as a variable label, → the
quired before you can return to the dialogue cell in the Label column for this variable. Then
box. type in the label (for example, reasons for
visiting). This label will appear on your output,
When you have finished going through the entire just as a designer label appears on a piece of
procedure, → OK. The toolbar beneath the menu bar clothing. If no variable label is specified, the
allows shortcut access to certain SPSS operations. variable name will be used for the output.
PLATE A.3
The Data Editor
Variable View
Appendix | Using IBM SPSS Statistics and NVivo Software 377
othmins in box after weimins]; be sure to put 1. → Transform → Recode into Different Vari-
commas (but not spaces) between the variable ables [opens the Recode into Different
names, and to delete any “?” that remain in the Variables dialogue box shown in Plate A.7]
Numeric Expression box 2. → age → adjacent button [puts age in the
5. ‑→ OK Numeric Variable → Output Variable box];
→ box beneath Output Variable Name
The new variable totalmin is created and appears and type agegp → Change [puts agegp in
in the Data Editor. the Numeric Variable → Output Variable
box]; → Old and New Values . . . [opens the
Recoding variables Recode into Different Variables: Old and
Sometimes you will need to recode variables—for New Values sub-dialogue box shown in Plate
example, to group scores together, as was done to A.8]
produce Table 8.3 for an interval/ratio variable 3. If there are missing values for a variable (as
(var00002, now given the variable name age). SPSS in this case), in the Old Value box, → the
offers two choices: change age, or keep age as it is and circle by System- or user-missing, and in
create a new variable. The latter option is appropriate the New Value box, → System-missing, then
here, because you may need the data for age in its ori- → Add
ginal form for some other analysis. We will preserve 4. In the Old Value box, → circle by Range:
the original variable and create a new one, agegp, for LOWEST through value and type 20 in the
age groups, with five age categories, as in Table 8.3. box. In the New Value box, → circle by Value
and type 1; → Add [the new value appears in The new variable agegp is created and appears in
the Old → New box] the Data View. To generate value labels for the five
5. In the Old Value box, → first circle by Range age bands and a variable label, repeat the approach
and type 21, and in the box after through type described above for those procedures.
30. In the New Value box, → circle by Value
and type 2; → Add
6. In the Old Value box, → first circle by Range Using SPSS for data analysis
and type 31 and in the box after through type
40. In the New Value box, → circle by Value Generating a frequency table
and type 3; → Add To produce a frequency table like the one in Table 8.2:
7. In the Old Value box, → first circle by Range
and type 41, and in the box after through type 1. → Analyze → Descriptive Statistics → Fre-
50. In the New Value box, → circle by Value quencies . . . [opens the Frequencies dialogue
and type 4; → Add box shown in Plate A.9]
8. In the Old Value box, → circle by Range: value 2. → reasons → adjacent button [puts reasons into
through HIGHEST and type 51 in the box. Variable(s) box]
In the New Value box, → circle by Value and 3. → OK
type 5; → Add → Continue [closes the Recode 4. The table then appears in the IBM SPSS Statis-
into Different Variables: Old and New Values tics Viewer (see Plate A.10)
sub-dialogue box shown in Plate A.8 and re-
turns to the Recode into Different Variables Note that in the Frequencies dialogue box, the
dialogue box shown in Plate A.7] variable name appears. If you would prefer to work
9. → OK with variable labels (which provide more information
PLATE A.8 The Recode Into Different Variables: Old And New Values Sub-Dialogue Box
Appendix | Using IBM SPSS Statistics and NVivo Software 381
on what the variable measures), go into the Data Generating a bar chart
Editor, → Edit → Options → General, and in the top
left-hand corner you can select variable labels to be To produce a bar chart like the one in Figure 8.2:
used. Similarly, if you prefer that the variables be
arranged in alphabetical order, you can specify that 1. → Graphs → Chart Builder → Define Variable
there. Properties [opens the Define Variable Prop-
erties dialogue box shown in Plate A.11]
2. → reasons → adjacent button [puts reasons in
Click here to send selected the Variables to Scan box]
variables into Variable(s) list. 3. → Continue [opens the Define Variable Prop-
erties sub-dialogue box shown in Plate A.12].
This allows you to edit properties such as vari-
able labels, value labels, and so on before creat-
ing a graph, or to enter that information if you
have not already done so. We’ve already entered
all the information needed, so exit this box.
4. → Graphs → Chart Builder → OK [opens the
Chart Builder dialogue box shown in Plate A.13]
5. → Gallery → Bar, and drag the type of bar
chart you want into the “canvas,” the large area
above the Gallery
Select variable(s) to be 6. Drag reasons into the X-Axis? box at the
analyzed from here.
bottom of the canvas
PLATE A.9 The Frequencies dialogue box 7. → OK
As mentioned earlier, for some useful details on 3. Drag reasons into the Slice by? drop zone at
this procedure, do the tutorial available in the Data the bottom of the canvas
Editor at Help → Topics. Use the search engine to 4. → OK
find Tutorial → Creating and editing charts. 5. Double-click anywhere in the chart to bring up
the Chart Editor (Plate A.14)
Generating a pie chart 6. → Elements → Show Data Labels. This will
To produce a pie chart like the one in Figure 8.3: place the percentage in each category in the
appropriate slice of the pie (see Plate A.15).
1. → Graphs → Chart Builder → OK [opens the
Chart Builder dialogue box] This pie chart is in colour, but if you only have
2. → Gallery → Pie/Polar, and drag the pie chart access to a monochrome printer, you can change
into the “canvas,” the large area above the Gallery the colours into patterns. Get into the Chart
Appendix | Using IBM SPSS Statistics and NVivo Software 383
Editor, which enables editing of all figures: → Edit ages to be grouped together. The general procedure
→ Properties → Variables. By reasons for visiting for doing this is → Graphs → Chart Builder → OK
(the variable label) → Style: Colour, and → Style: → Gallery → Histogram, and drag the histogram
Pattern → Apply. Use the same procedure for all into the “canvas.” Then, → Element Properties
charts. → Statistic: Histogram → Set Parameters. However,
since we have already recoded age into the categories
Generating a histogram we want, you can use the newly created agegp vari-
Producing the histogram in Figure 8.4 is just as able. Although technically it is an ordinal variable,
simple, except that it requires you to define the to create a histogram it is necessary to define agegp
Appendix | Using IBM SPSS Statistics and NVivo Software 385
as a scale (interval/ratio) variable by right-clicking When the Print dialogue box appears, make sure
on it in the Variables list while in the Chart Builder. Selected output under Print Range is selected.
Drag agegp into the X-Axis? drop zone at the bottom
of the canvas; → OK. Other ways of presenting inter- Generating the arithmetic mean,
val/ratio data, such as measures of central tendency median, standard deviation, and range
and dispersion, will be d iscussed below. To produce the mean, median, standard deviation,
and the range for an interval/ratio variable like age,
Printing output follow these steps:
To print all the output in the IBM SPSS Statistics
Viewer, make sure that the Viewer is the active 1. → Analyze → Descriptive Statistics → Explore
window and then → File → Print . . . . When the . . . [opens the Explore dialogue box]
Print dialogue box appears, → OK. To print just 2. → age → the button to the left of Dependent
some of the output, hold down the Ctrl button on List [puts age in the Dependent List box];
the keyboard and click once on the parts of the → Statistics under Display; → OK
output you want to print. The easiest way to do this 3. The output also includes the 95 per cent confi-
is to select the elements from the output summary dence interval for the mean, which is based on
in the left-hand segment of the Viewer shown in the standard error of the mean. The output is
Plate A.10. Then bring up the Print dialogue box. in Table A.1.
TABLE A.2 | Contingency table for reasons for visiting by gender (SPSS output)
CROSSTABS
CASE PROCESSING SUMMARY
Cases
Valid Missing Total
Case Processing
N Per cent Summary Per cent N Per cent
reasons for visiting * 90 100.0% 0 .0% 90 100.0%
GENDER
CHI-SQUARE TESTS
This is the χ2 value Asymp. Sig.
referred to in the Value df [(2-sided)]
text
Pearson Chi-Square 22.726a 3 .000
Likelihood Ratio 25.805 3 .000
Linear-by-Linear Association 9.716 1 .002
N of Valid Cases 90
a
2 cells (25.0%) have expected count less than 5. The minimum expected count is 4.20.
SYMMETRIC MEASURES
Approx. Shows level of statistical
Value Sig. significance of computed value
Nominal by Phi .503 .000 of Cramér’s V
Must be selected for number of cases identified as likely to be the independent variable, it
in each cell to be included in table. is by convention placed on the X-axis (the horizontal
axis). Since age is bound to be the independent vari-
able, follow these steps:
Generating Pearson’s r
Select to give the percentage To produce Pearson’s r, in particular the correla-
of cases of each category of
tion between age, weimins, and cardmins shown in
the column variable.
Table 8.5, follow these steps:
PLATE A.17 The Crosstabs: Cell Display
Sub-Dialogue Box 1. → Analyze → Correlate → Bivariate . . . [opens
the Bivariate Correlations dialogue box
shown in Plate A.19]
Select to provide the 2 statistics
2. → age → button → weimins → button → card-
for the contingency table.
mins → button [age, weimins, and cardmins
should now be in the Variables box]; → Pear-
son (if not already selected) → OK
Comparing means
To produce a table like Table 8.6, follow these steps:
All variable(s) to be
correlated go here.
TABLE A.3 | Correlations output for age, weimins, and cardmins (SPSS output)
CORRELATIONS Correlations of p < 0.05 are “flagged” with asterisks
AGE WEIMINS CARDMINS
AGE Pearson Correlation 1.000 −.273 ** −.109
Sig. (2-tailed) . .010 .311
N 89 89 89
WEIMINS Pearson Correlation −.273 ** 1.000 −.161
Sig. (2-tailed) .010 . .130
N 89 90 90
CARDMINS Pearson Correlation −.109 −.161 1.000
Sig. (2-tailed) .311 .130 .
N 89 90 90
** Correlation is significant at the 0.01 level (2-tailed).
either opening one of the existing projects or starting fieldwork notes. NVivo 10 can accept documents in
a new one. If you are starting a new project, as in the both rich text (.rtf) and Word (.doc, .docx) formats.
example that follows, → File → New Project. The New To import the documents (see Plate A.23), → Inter-
Project dialogue box appears and you are asked to nals (below Sources at the top of the Navigation
provide a Title for your project. For this exercise, the view) → External Data on the ribbon (see Plate A.24
title “Disney Project” was chosen. You are also asked for a description of how to find the ribbon and other
to give a Description of the project, although this is key areas of the screen) → Documents button on the
an optional feature. When you have done this, → OK. find bar [opens the Import Internals dialogue box]
You then need to import the documents you → Browse to locate the documents that are to be im-
want to code. In this case, they will be interview ported. → the documents to be imported (you can
transcripts from the project on visitors to Disney hold down the Ctrl key to select several documents;
theme parks, referred to in Box 11.5 and Chapter 13. if you want to select all of them, hold down the Ctrl
Other kinds of documents can be imported such as key and tap the A key); → Open. The documents
Navigation view—provides
access to documents,
nodes, etc.
will then be visible in the document viewer (this is There are several ways of going about the coding
the term used to describe the general screen shown process in NVivo. The approach taken in relation to
in Plate A.24). Once the documents have been im- the coding of the Disney Project was as follows:
ported, they can be read and edited. All you need to
do is to double-click on the yellow icon to the left of • The interviews, both in printed form and in
each interview in the list view. the document viewer (Plate A.24), were read
through. The viewer is treated as having a
Coding number of different components or sections
Coding your data is obviously one of the key phases and these are highlighted in Plate A.24.
in the whole process of qualitative data analysis. For • Some codes that seemed relevant to the docu-
NVivo, coding is accomplished through nodes. NVivo’s ments were created.
help system in earlier releases defined coding as “the • The documents were then coded using NVivo.
process of marking passages of text in a project’s
documents with nodes” (emphasis added). Thus An alternative strategy is to code while browsing
nodes are the route by which coding proceeds. In the documents.
turn, the latest release defines a node as “a collection
of references about a specific theme, place, person or Creating nodes
other area of interest.” When a document has been The nodes that were relevant to the passage in Box 11.5
coded, the node will incorporate references to those are presented in Figure A.1. Prior to NVivo 9, when
portions of documents in which the code appears. creating a node, the researcher had to choose either
Once established, nodes can be changed or deleted. a free node or a tree node. The latter is a node that is
394 Appendix | Using IBM SPSS Statistics and NVivo Software
organized in a hierarchy of connected nodes, whereas 1. In the document viewer → Create in the ribbon
free nodes were not organized in this way. This dis- 2. → Node in the find bar [opens the New Node
tinction has been dropped in versions after NVivo 9, dialogue box shown in Plate A.25]
and the software assumes that a tree node is being 3. Enter the node Name (Not critical of Disney)
created. Two points are crucial to note here for users and a Description (the latter is optional)
of earlier releases of the software. First, the tendency 4. → OK
now is not to refer to “tree nodes” but to treat them
as hierarchically organized nodes. Second, free nodes Creating hierarchically organized nodes
(that is, nodes that are not hierarchically organized) To create a hierarchically organized node, the in-
can still be created; they are simply nodes without itial process is exactly the same as with a non-
“children” (to use the latest NVivo terminology). hierarchically organized node. The following example
Notice that there are three groups of hierarchi- will show how to create the hierarchically organized
cally organized nodes and two non-hierarchically node Class critique, which is a child of the hierarch-
organized nodes in Figure A.1. The nodes can be cre- ically organized node Ideological critique, which
ated in the following way. is itself a child of the hierarchically organized node
Critique (see Figure A.1). The following steps will
Creating non-hierarchically organized generate this node.
nodes
This sequence of steps demonstrates how to create 1. In the document viewer → Create in the ribbon
the non-hierarchically organized node Not critical 2. → Node in the find bar [opens the New Node
of Disney. dialogue box shown in Plate A.26]
3. Enter the node Name (Critique) and a
Description (the latter is optional)
Hierarchical nodes Critique
4. → OK
5. → Critique in the list of nodes in the list viewer
Aesthetic Ideological Content Other
6. → Node in the find bar [opens the New Node
critique critique critique critique dialogue box shown in Plate A.26]
7 Enter the node Name (Ideological critique)
and a Description (the latter is optional). This
Class Ethnicity Gender Nationality node will form a child of the hierarchically or-
critique critique critique critique
ganized node (make sure that in Hierarchical
name it reads Nodes\\Critique, as this will
Visitors
mean it is a child of Critique). See Plate A.26.
8. → Ideological critique in the list of nodes in
Visitors’ Visitors’ Visitors‘ Visitors‘ the list viewer
age gender class ethnicity 9. → Node in the find bar [opens the New Node
dialogue box shown in Plate A.26]
Significance
10. Enter the node Name (Class critique) and a
Description (the latter is optional). This node
will form a child of the hierarchically organ-
Educational Recreational Ideological Other
significance significance significance significance ized node (make sure that in Hierarchical
name it reads Nodes\\Critique\Ideologi-
Non-hierarchical nodes
cal critique, as this will mean it is a child of
Uncritical enthusiasm
Ideological critique, which is itself a child of
Not critical of Disney Critique). See Plate A.26.
FIGURE A.1 Nodes used in the Disney Project 11. → OK
Appendix | Using IBM SPSS Statistics and NVivo Software 395
Applying nodes in the coding process want to use has been created, NVivo 10 allows you
Coding is carried out by applying nodes to segments to drag and drop text into it (see Plate A.27). To do
of text. Once you have set up some nodes (remember this, highlight the text to be coded and then, holding
that you can add and alter them at any time), you down the left-hand button, drag the text over to the
can look at a document in the viewer and highlight appropriate node in the list view.
the area of the document that you want to code; then Alternatively, you can highlight the text you want
right-click on the mouse while holding the cursor to code and right-click over it. Then → Code Selec-
over the highlighted text. If the node that you want tion → Code Selection at Existing Nodes, which
to use has not been created yet, highlight the text opens the Select Project Items dialogue box (see
you want to code, right-click on the highlighted text, Plate A.28). → the node(s) you want to use. Thus, in
and then → Code Selection → Code Selection at the example in Plate A.28, the tick by Uncritical en-
New Node . . . . This opens the New Node dialogue thusiasm will code the highlighted text at that node.
box. You can then create a new node in the manner If you also wanted to use a hierarchically organized
outlined in the previous sections. If the code you node, you would need to find the appropriate parent
396 Appendix | Using IBM SPSS Statistics and NVivo Software
in the list of nodes within the list view and then → coding stripes. Selecting this feature allows you to
the plus sign to the left of it. To uncode at any point, see multi-coloured stripes that represent portions of
simply highlight the passage to be uncoded, and coded text and the nodes that have been used. Over-
→ the button with a red cross in it in the quick coding lapping codes do not represent any problem at all.
bar (see Plate A.24). Alternatively, you can right-click To activate this feature, → View in the ribbon
on the highlighted text and → Uncode. and then → Coding Stripes in the find bar → Nodes
Recently Coding. Plate A.29 shows these stripes.
Coding stripes We can see that some segments have been coded at
It’s very helpful to be able to see the areas of text two or more nodes (for example, Visitors’ ethnicity
that have been coded and the nodes applied to and Ethnicity critique). All the nodes that have been
them. To this end NVivo has a useful aid called used are clearly displayed.
Appendix | Using IBM SPSS Statistics and NVivo Software 397
Hint. To uncode at
any time, highlight the
text to be uncoded and
click on this button.
This will clear the
coding at that point.
3. Click on OK.
5. Select the
Advanced tab.
convenience sample A sample that is selected because of discourse analysis An approach to the analysis of talk
its availability to the researcher; a form of non-probability and other forms of communication that emphasizes the
sample. way language can create versions of reality.
conversation analysis The fine-grained analysis of dispersion The degree of variation in a set of scores.
talk (recorded in naturally occurring situations and then Commonly used measures of dispersion include the range
transcribed) to uncover the underlying structures in and the standard deviation.
interaction that make social order possible. Conversation ecological fallacy The error of assuming that infer-
analysis is grounded in ethnomethodology. ences about individuals can be made from data on the
correlation An approach to the analysis of relationships characteristics of groups.
between interval/ratio variables and/or ordinal variables empiricism An approach to the study of reality ac-
that seeks to assess the strength and direction of the rela- cording to which only knowledge gained by observation
tionship between the variables concerned. Pearson’s r and through the senses is acceptable.
Spearman’s rho are both correlational measures. The cor- epistemology, epistemological A branch of philoso-
responding term measure of association is often used phy concerned with what constitutes knowledge and how
with nominal variables. knowledge is to be acquired; see positivism, realism, and
covert research A term frequently used in connection interpretivism.
with ethnographic research in which the researcher does not eta A test of the strength of the relationship between two
reveal his or her true identity and/or intentions. Such re- variables. The independent variable is usually a nominal
search may violate the ethical principle of informed consent. variable while the dependent variable must be an interval
Cramér’s V A statistical measure used to assess the variable or ratio variable.
strength of the relationship between two nominal variables. ethnography, ethnographer, ethnographic Like par-
critical discourse analysis A type of content analysis ticipant observation, a research method in which the re-
that brings issues such as power hierarchies, structural in- searcher is immersed in a social setting for an extended
equalities, and historical political struggles to bear on the period of time, observing behaviour, asking questions,
analyses of texts. and analyzing what is said in conversations both between
critical realism A realist epistemology according to subjects and with the fieldworker. As a term, “ethnog-
which the study of the social world should be concerned raphy” is more inclusive than “participant observation,”
with the identification of social structures in order to which emphasizes the observational component. Written
change them and thereby counteract inequalities and accounts of ethnographic research are often referred to as
injustices. Unlike positivism, which is an empiricist epis- ethnographies.
temology, critical realism maintains that the structures ethnomethodology A sociological perspective con-
may not be directly perceivable. cerned with the way social order is established and main-
cross-sectional design A research design in which data tained through talk and interaction; the intellectual
are collected at a single point in time. foundation of conversation analysis.
deduction, deductive An approach to inquiry that evaluation research Research that concentrates on the
begins with the statement of a theory from which hypoth- evaluation of real-life interventions, such as policy changes.
eses may be derived and tested; compare with inductive. experiment A research design that rules out alternative
denotation A term used in semiotics to refer to the prin- explanations of findings deriving from it (in other words,
cipal and most manifest meaning of a sign; compare with that possesses internal validity) because it involves (a)
connotation. both an experimental group, which is exposed to a treat-
dependent variable A variable that is caused (or as- ment, and a control group, which is not, and (b) random
sumed to be caused) by an independent variable. assignment to the two groups.
diary A written memoir. Three types of diary are dis- experimental group See experiment.
cussed in this book: personal diaries that are spontaneously external validity A type of validity that is achieved if the
produced and can be analyzed as personal documents; results of a study can be generalized beyond the specific
diaries written at the behest of a researcher; and diaries research context in which they were generated.
written by researchers to log their activities and reflections. face validity A type of validity that is achieved if, on in-
dichotomous variable A variable with just two categories. spection, an indicator appears to measure the concept in
dimension An aspect of a concept. question.
Glossary 405
Kendall’s tau-b A test of the strength of the relationship multivariate analysis The examination of relationships
between two ordinal variables. among three or more variables.
key informant Someone who offers the researcher, usually narrative analysis An approach focused on the search
an ethnographer, particularly perceptive information about for and analysis of stories that people use to understand
the social setting, important events, and individuals. their lives and the world around them.
life history interview An unstructured interview that is naturalism, naturalistic A style of research designed to
similar to an oral history interview, but designed to gather minimize disturbance to the natural or everyday social world.
information on the entire biography of a respondent. negative relationship A relationship between two vari-
life history method A method (often referred to as the ables in which one decreases as the other increases.
biographical method) that emphasizes the inner experi- nominal variable A variable comprised of categories that
ence of individuals and its connections with larger societal cannot be ranked; also known as a categorical variable.
events throughout the life course. It usually entails life his- nomothetic An approach to explanation that utilizes
tory interviews and the use of personal documents as data. general law and principles, which are said to apply to some
Likert item, Likert scale A widely used format in which population beyond the people studied.
respondents are typically asked their degree of agreement non-probability sample A sample selected using a
with a series of attitude statements that together form a non-random sampling method. Essentially, this means
multiple-indicator measure. The scale is designed to meas- that some units in the population are more likely than
ure the intensity of respondents’ feelings about an issue. others to be selected.
longitudinal research A research design in which data non-response Occurs if someone in a sample refuses to
are collected on at least two separate occasions. participate in the study, cannot be contacted, or for some
mean See arithmetic mean. other reason does not supply the required data.
measure of central tendency A measure that summar- null hypothesis A hypothesis of no relationship between
izes the magnitude of a set of scores; examples includes the two variables; the hypothesis that the researcher hopes to
arithmetic mean, median, and mode. disprove.
measure of dispersion A measure (such as the range or objectivism, objectivist An ontological position ac-
standard deviation) that summarizes the amount of varia- cording to which social phenomena have an existence in-
tion in a set of scores. dependent of social actors or their perceptions; compare
measurement validity The degree to which a measure of with constructionism.
a concept actually measures what it is supposed to meas- observation schedule A device used in structured ob-
ure; see also face validity and concurrent validity. servation that specifies the categories of behaviour that are
median The mid-point in a set of scores that is arranged to be observed and gives instructions on how behaviour
in order. should be allocated to those categories.
member validation See respondent validation. official statistics Data compiled by (or on behalf of)
missing data Data that are not available, for example, state agencies in the course of conducting their business.
when a respondent in social survey research does not ontology, ontological A branch of philosophy concerned
answer a question. These are referred to as “missing values” with the nature of reality: for example, whether social en-
in IBM SPSS Statistics software. tities can and should be considered objective entities with
mode The score that occurs most frequently in a set of a reality external to specific social actors, or as social con-
scores. structions built up through the perceptions and actions of
moderated relationship A relationship between two these actors. See objectivism and constructionism.
variables is said to be “moderated” when the effect of the in- open, open-ended question In an interview schedule
dependent variable varies at different levels of a second in- or questionnaire, a format that does not present the re-
dependent variable (also known as a statistical interaction). spondent with a set of possible answers to choose from;
moderator The person who guides the questioning of a compare with closed question.
focus group, also called a facilitator. operationalization The way in which a concept is meas-
multiple-indicator measure A measure that uses more ured empirically; for example, one could operationalize
than one indicator. the concept “level of education” with a survey item that
multi-strategy research Research that combines quan- asks people to indicate their highest level of education
titative and qualitative approaches. by choosing from categories such as “some elementary,”
Glossary 407
“high school,” “university graduate,” etc.; see also oper- scientists can simply uncover, and argues that no single
ational definition. account of social reality can be authoritative. Postmodern-
operational definition A definition that spells out the ists prefer qualitative methods.
operations that are to be performed to measure a concept. pre-coded question Another name for a closed ques-
oral history interview A largely unstructured interview tion; used because a numerical code has been pre-assigned
in which respondents are asked to recall and reflect on to each of the predetermined answers (hence there is no
events they have experienced. need for a coding frame).
ordinal variable A variable whose categories can be probability sample A sample, selected at random, in
rank-ordered, but the distances between the categories are which each unit in the population has a known chance of
not equal or known across the range. being selected.
outlier An extreme value (either very high or very low) purposive sampling A form of non-probability sam-
in a distribution of scores. If a variable has an outlier, it pling in which cases are selected on the basis of their
will distort both the arithmetic mean and the range. ability to provide information relevant to the topic of the
paradigm A cluster of beliefs and assumptions, often study.
unstated, that influence views on what should be studied, QSR NVivo A CAQDAS package that derives from but
how research should be done, and how results should be goes beyond NUD*IST (Non-numerical Unstructured
interpreted. Data Indexing Searching and Theorizing).
participant observation Research in which fieldworkers qualitative content analysis An approach to construct-
are immersed in a social setting for an extended period of ing the meaning of documents and text that allows cat-
time, observing behaviour, asking questions, and analyz- egories to emerge out of data analysis and recognizes the
ing what is said in conversations both between the people significance of the context in which items appear.
under study and with the researcher. It usually includes qualitative research Inquiry that uses mainly words,
interviewing key informants and studying documents. In images, and other non-numerical symbols as data and
this book, “participant observation” refers to the observa- involves little or no quantification. As a research strategy
tional aspect of ethnography. it tends to be inductivist, constructivist, and interpretivist,
participatory action research Research in which local but qualitative researchers do not always subscribe to all
people affected by a particular social problem collaborate three features; compare with quantitative research.
as equals with professional researchers and government quantitative research Inquiry using quantitative
officials to generate knowledge pertinent to the problem data-gathering techniques and statistical analysis. As a
and to take action to ameliorate it. research strategy it tends to be deductivist and objectivist
Pearson’s r A measure of the strength and direction of and to incorporate a natural science model of the research
the relationship between two interval/ratio variables. process (in particular, one influenced by positivism), but
personal documents Documents not written for an quantitative researchers do not always subscribe to all three
official purpose that provide first-person accounts of the features; compare with qualitative research.
writer’s life and events within it. Examples include diaries, quasi-experiment A research design that resembles an
letters, and autobiographies. experiment but does not meet all the requirements and
population All cases or people covered by a theory or therefore does not exhibit complete internal validity.
explanation; the universe of units from which a sample is questionnaire A collection of written questions or re-
selected. sponse items that the respondent completes without the
positive relationship A relationship between two vari- aid of an interviewer.
ables whereby both move in the same direction, either in- quota sample A type of non-probability sample that
creasing or decreasing simultaneously. matches the proportions of people in different categories
positivism, positivist An epistemological position that in the population.
advocates using the methods of the natural sciences in the random assignment The random allocation of research
study of social reality. participants to either the experimental or the control
postal questionnaire A questionnaire that is sent out group(s) in an experiment.
and returned by regular (non-electronic) mail. random sampling The form of sampling in which
postmodernism A position that questions the notion all units of a population have the same chance of being
of a pre-existing external reality that dispassionate social selected.
408 Glossary
range The difference between the maximum and the sampling frame A listing of units in the population
minimum score in a set of scores associated with an inter- from which a sample is to be selected.
val or ratio variable. scale A term usually used interchangeably with index to
ratio variable An interval variable with a true zero refer to a multiple-indicator measure in which the scores
point. an individual respondent gives for each component indi-
reactivity, reactive effect The effect on research partici- cator are summed to produce a composite score for that
pants of knowing that they are being studied, which may person.
result in atypical or inauthentic behaviour. secondary analysis The analysis of data (quantitative or
realism, realist An epistemological position according qualitative) by researchers other than those responsible
to which reality is independent of the senses but is to some for their collection, often for purposes that the latter may
degree accessible to the researcher’s tools and theoretical not have envisaged.
speculations. See also critical realism. semiotics An approach to the analysis of documents
reflexive, reflexivity Terms used to refer to social re- and other materials that emphasizes the importance of
searchers’ awareness of the implications that their meth- signs and symbols, seeking out their deeper meaning and
ods, values, biases, decisions, and mere presence in the exploring their intended effects.
situations they investigate have for the knowledge they semi-structured interview A type of interview in which
generate. the interviewer has a series of questions in the general
relationship An association between two variables form of an interview guide and is able to vary the sequence
whereby the variation in one variable coincides with vari- of questions. The questions are typically more general than
ation in another variable. those specified by a structured interview schedule, and the
reliability The degree to which a measure of a concept is interviewer usually has some latitude to ask further ques-
stable or consistent. tions in response to replies that appear significant.
replication, replicability The degree to which a study sensitizing concept A concept that is not fixed in ad-
can be repeated using the same methods. vance (as with an operational definition) but rather is
representative sample A sample that is similar to the treated as a guide that suggests what may be relevant or
population in all important respects. important in an investigation.
research design The term used in this book to refer sign A term used in semiotics. Each sign has two parts:
to a framework for the collection and analysis of data. the signifier (manifestation of a sign) and the signified (the
The choice of research design reflects the goals that the deeper meaning to which the signifier refers).
researcher hopes to achieve (for example, generaliza- simple random sample A sample in which each unit
tion, establishing causality, or producing empathetic selected from the population, and each combination of
understanding). units, has an equal probability of being included.
respondent validation A process whereby researchers snowball sample A non-probability sample in which
provide the people on whom they conducted research with the researcher makes initial contact with a small group of
an account of their findings and request their feedback on people connected to the research topic and then uses them
it; sometimes called member validation. to establish contact with others.
response set A term for the tendency for some people, social desirability bias Distortion of data caused by re-
when responding to multiple-indicator measures, to re- spondents’ efforts to construct accounts that conform to a
spond to every item in the same way, suggesting that their socially acceptable model of behaviour.
answers are motivated by something other than their Spearman’s rho [ρ] A measure of the strength and dir-
actual feelings. Three of the most common response-set ection of the relationship between two ordinal variables.
effects are “acquiescence,” “social desirability,” and “laziness spurious relationship A relationship in which two vari-
or boredom.” ables are correlated but not causally related. It is produced
sample The segment or subset of the population selected by the impact of a common third variable on each of the
for research. The selection may be based on either prob- two variables. When the third variable is controlled, the
ability or non-probability sampling. relationship disappears.
sampling error Differences between the characteristics standard deviation A measure of how widely the scores
of a random sample and the population from which it is that make up a set are dispersed around the arithmetic
selected. mean.
Glossary 409
standard error of the mean An estimate of the degree to theoretical saturation In grounded theory, the point
which a sample mean is likely to differ from the population where emerging concepts have been fully explored and
mean. no new insights are being generated. See also theoretical
statistical significance (test of) A test (such as the chi- sampling.
square test) that allows analysts to estimate the probability thick description A term coined by Geertz (1973; see
that the results of a study of a randomly selected sample Chapter 9) to refer to detailed accounts of a social setting or
are generalizable to the population from which the sample people’s experiences that can form the basis for general state-
was drawn. Note that statistical significance has nothing ments about a culture and its significance in peoples’ lives.
to do with substantive significance or importance. Using a transcription, transcript The verbatim written record of
test of statistical significance to generalize from a sample a taped interview or focus group session.
to a population is known as statistical inference. triangulation The use of more than one method or source
stratified random sample A sample in which units are of data so that findings may be cross-checked.
randomly sampled from a population that has been previ- trustworthiness A general criterion (composed of four
ously divided into sub-groups (strata). more specific criteria) used by some writers in assessing
structured interview One in which all respondents are the quality of qualitative research.
asked exactly the same questions in the same order with turn-taking In conversation analysis, the idea that
the aid of a formal interview schedule. taking turns to speak is a rule that helps to maintain order
structured observation Often called systematic obser- in everyday conversation.
vation, this is a research method in which the researchers univariate analysis The analysis of one variable at a
follow explicitly formulated rules regarding not only what time.
they should look for, but when and where, and how they unobtrusive measures, unobtrusive methods, unobtru-
should record what they have observed. sive observations Measures that do not cause reactivity
survey research A research design in which data are col- because they do not make research participants aware that
lected from respondents, mainly by questionnaire or struc- they are being studied.
tured interview; typically, the data are then examined to unstructured interview An interview in which the
detect relationships among variables. interviewer is free to explore any topic, although an
symbolic interactionism A theoretical perspective in interview guide is often used. The questioning is usually
which social interaction is understood to be based on the informal and the content, phrasing, and sequencing of
meanings that actors attach to their actions and the con- questions may vary from one interview to the next.
texts in which they occur. validity A research criterion concerned with the integ-
systematic sample A probability sampling method in rity of the conclusions generated by a particular study.
which units are selected from a sampling frame at fixed There are several types of validity: see, in particular, meas-
intervals (for example, every fifth unit). urement validity, internal validity, and external validity.
text A written document or, in more recent years, any- When used on its own, validity is usually taken to refer to
thing (from paintings to buildings to theme parks) that measurement validity.
may be “read” and analyzed for its symbolic value. variable An attribute or characteristic that may vary
theoretical sampling A term used mainly in the context over time or from case to case. See also dependent variable
of grounded theory to refer to sampling carried out in such and independent variable; compare with constant.
a way that emerging theoretical considerations guide the vignette technique A method in which people are pre-
collection of data and/or the selection of cases (usually re- sented with “vignettes” (hypothetical scenarios) and then
search participants). This process is supposed to continue asked how they would respond if they faced the circum-
until a point of theoretical saturation is reached. stances depicted in the scenarios.
References
Chapter 1 Canada. Canadian Journal of Sociology, Lofland, J., & Lofland, L. (1995). Analyzing
35, 573–593. social settings: A guide to qualitative
Beagan, B. (2001). Micro inequities and Durkheim, E. (1938). The rules of sociological observation and analysis (3rd ed.).
everyday inequalities: “Race,” gender, method. S. Solavay & J. Mueller (Trans.). Belmont, CA: Wadsworth.
sexuality and class in medical school. New York: Free Press. Lyon, K., & Frohard-Dourlent, H. (2015).
Canadian Journal of Sociology, 26, Durkheim, E. (1952). Suicide: A study in Let’s talk about the institution: Same-
583–610. sociology. J. Spaulding & G. Simpson sex common-law partners negotiating
Bell, E. (2007). Separatism and quasi- (Trans.). London: Routledge & Kegan marriage equality and relationship
separatism in Alberta. P rairie Forum, Paul. (Original work published 1897) legitimacy. Canadian Review of Sociology,
32, 335–355. Foster, J. (1995). Informal social control and 52, 402–428.
Bikos, L. (2017, 17 May). It’s not just the community crime prevention. British Maiolino, E. (2015). Political pugilists:
RCMP: Police culture is toxic. Globe Journal of Criminology, 35, 563–583. Recuperative gender strategies in
and Mail. Retrieved from https://www Frisby, W., Crawford, S., & Dorer, T. (1997). Canadian electoral politics. Canadian
.theglobeandmail.com/opinion/its-not- Reflections on participatory action Review of Sociology, 52, 115–133.
just-the-rcmp-police-culture-is-toxic/ research: The case of low-income women Marx, K. (1998). The German ideology
article35014971/ accessing local physical activity services. including theses on Feuerbach and
Blumer, H. (1962). Society as symbolic Journal of Sport Management, 11, 8–28. Introduction to the critique of political
interaction. In A. Rose (Ed.), Human Giddens, A. (1984). The constitution of economy. A mherst, NY: Prometheus
Behavior and Social Processes. London: society. Cambridge, UK: Polity. Books. (Original work published 1845)
Routledge & Kegan Paul. Goffman, E. (1963). Stigma: Notes on the McKeganey, N., & Barnard, M. (1996). Sex
Charpentier, M., & Quéniart, A., (2017). management of spoiled identity. work on the streets. Buckingham: Open
Aging experiences of older immigrant Harmondsworth, UK: Penguin. University Press.
women in Québec (Canada): From Hallgrimsdottir, H., Phillips, R., & Benoit, C. Merton, R. (1938). Social structure and
deskilling to liberation. Journal of (2006). Fallen women and rescued anomie. American Sociological Review,
Women & Aging, 29, 437–447. girls: Social stigma and media narratives 3, 672–682.
Collins, R. (1994). Four sociological of the sex industry in Victoria, B.C., Merton, R. (1967). On theoretical sociology.
traditions (Rev. ed.). New York: Oxford from 1980 to 2005. C anadian Review New York: Free Press.
University Press. of Sociology and Anthropology, 43, Mies, M. (1993). Towards a methodology for
Creswell, J. (2013). Qualitative inquiry and 265–280. feminist research. In M. Hammersley
research design: Choosing among five Hay, D. (2014). An investigation into the (Ed.), Social research: Philosophy, politics
approaches (3rd ed.). Los Angeles: Sage. swiftness and intensity of recent secu- and practice. London: Sage.
Debbink, G., & Ornelas, A. (1997). Cows larization in Canada: Was Berger right? Milgram, S. (1963). A behavioural study
for Campesinos. In S.E. Smith & D.G. Sociology of Religion, 75, 136–162. of obedience. Journal of Abnormal and
Willms (Eds.), Nurtured by knowledge: Hier, S. (2002). Raves, risks and the ecstasy Social Psychology, 67, 371–378.
Learning to do participatory action- panic: A case study in the subversive Morgan, R. (2000). The politics of crimino-
research. New York and Ottawa: Apex nature of moral regulation. Canadian logical research. In R. King & E. Wincup
Press and International Development Journal of Sociology, 27, 33–57. (Eds.), Doing research on crime and jus-
Research Centre. Hochschild, A. (1983). The managed heart. tice. Oxford: Oxford University Press.
Demaiter, E.I., & Adams, T.L. (2009). “I Berkeley and Los Angeles: University of Nagra, B., & Maurutto, P. (2016). Cross-
really didn’t have any problems with the California Press. ing borders and managing racialized
male–female thing until . . .” : Successful Hughes, G. (2000). Understanding the pol- identities: Experiences of security and
women’s experiences in IT organizations. itics of criminological research. In V. surveillance among young Canadian
Canadian Journal of Sociology, 43, 31–53. Jupp, P. Davies, & P. Francis (Eds.), Doing Muslims. Canadian Journal of Sociology,
Dinovitzer, R., Hagan, J., & Parker, P. (2003). criminological research. London: Sage. 41, 165–194.
Choice and circumstance: Social capital Lantz, P., & Booth, K. (1998). The social con- Nietzsche, F. (1968). The will to power.
and planful competence in the attain- struction of the breast cancer epidemic. W. Kaufmann & R. J. Hollingdale
ments of immigrant youth. Canadian Social Sci. and Medicine, 46, 907–918. (Trans.). New York: Vintage Books.
Journal of Sociology, 28, 463–488. Layder, D. (1993). New strategies in social (Original work published 1883)
Duncan, L. (2010). Money and member- research. Cambridge, UK: Polity. Neuman, W.L. (2003). Social research
ship: Effects of neighbourhood poverty, Lewin, K. (1946). Action research and min- methods. Toronto: Pearson Canada.
income inequality and individual income ority problems. Journal of Social Issues, Oakley, A. (1981). Interviewing women:
on voluntary association membership in 2, 34–46. A contradiction in terms. In H. Roberts (Ed.),
References 411
Doing feminist research. London: Tastsoglou, E., & Miedema, B. (2003). possibility of consilience. Sociology, 51,
Routledge & Kegan Paul. Immigrant women and community 880–896.
Oakley, A. (1998). Gender, methodology development in the Canadian Maritimes: Bell, E., Woodley, M.A., Aitken Schermer,
and people’s ways of knowing: Some Outsiders within? Canadian Journal of J., & Vernon, P.A. (2012). Politics and the
problems with feminism and the Sociology, 28, 203–234. general factor of personality. Personality
paradigm debate in social science. Teevan, J., & Dryburgh, H. (2000). First and Individual Differences, 53, 546–551.
Sociology, 32, 707–731. person accounts and sociological explan- Berthoud, R. (2000). Introduction: The
Orwell, G. (1933). Down and out in Paris ations of delinquency. Canadian Review dynamics of social change. In R.
and London. Toronto: Harcourt. of Sociology and Anthropology, 37, 77–93. Berthoud & J. Gershuny (Eds.), Seven
Potter, J. (1996). Representing reality: Turnbull, P. (1973). The mountain people. years in the lives of British families:
Discourse, rhetoric and social construc- London: Cape. Evidence on the dynamics of social
tion. London: Sage. Weber, M. (1947). The theory of social and change from the British Household Panel
Pratt, A., & Valverde, M. (2002). From economic organization. A.M. Henderson & Survey. Bristol: Policy Press.
deserving victims to “masters of T. Parsons (trans.). New York: Free Press. Blaxter, M. (1990). Health and lifestyles.
confusion”: Redefining refugees in the Wilson, B. (2002). The Canadian rave scene London: Routledge.
1900s. Canadian Journal of Sociology, and five theses on youth resistance. Bowen, R. (2015). Squaring up: Experiences
27, 135–162. Canadian Journal of Sociology, 27, 373–412. of transition from off-street sex work to
Schutz, A. (1962). Collected papers I: The Wolcott, H. (2008). Ethnography: A way of square work and duality—concurrent
problem of social reality. The Hague: seeing (2nd ed.). Toronto: AltaMira Press. involvement in both—in Vancouver,
Martinus Nijhof. BC. Canadian Review of Sociology,
Smith, D. (2004). Women’s perspective 52, 429–449.
Chapter 2
as a radical critique of sociology. In Brannigan, A. (2004). The rise and fall of
S. Harding (Ed.), The feminist standpoint Alford, J.R., Funk, C.L., & Hibbing, J.R. social psychology: The use and misuse of
theory reader: Intellectual and political (2005). Are political orientations the experimental method. Hawthorne,
controversies. New York: Routledge. genetically transmitted? American NY: Aldine de Gruyter.
Smith, D. (2005). Institutional ethnography: Political Science Review, 99, 153–167. Carroll, L. (2009). Alice’s adventures in
A sociology for people. Walnut Creek, CA: Anderssen, E. (2015, 1 October). One way Wonderland. Toronto: Penguin Group.
AltaMira Press. to support Canadian families this (Original work published 1865)
Smith, S.E. (1997). Deepening participatory- campaign? Make life easier for single Charlton, T., Gunter, B., & Coles, D. (1998).
action research. In S.E. Smith & D.G. moms. The Globe and Mail. Retrieved Broadcast television as a cause of aggres-
Willms (Eds.), Nurtured by knowledge: from https://www.theglobeandmail sion? Recent findings from a naturalistic
Learning to do participatory action- .com/life/parenting/the-real-way- study. Emotional and Behavioural
research. New York and Ottawa: Apex to-support-canadian-families-this- Difficulties, 3, 5–13.
Press and International Development campaign-make-life-easier-for-single- Charlton, T., Coles, D., Panting, C., &
Research Centre. moms/article26613155/ Hannan, A. (1999). Behaviour of nursery
Statistics Canada. (2013). Cycle 27: Giving, Baer, D., Curtis, J., & Grabb, E. (2001). Has class children before and after the
volunteering and participating. General voluntary association membership availability of broadcast television: A
Social Survey. declined? Cross-national analyses for naturalistic study of two cohorts in a
Statistics Canada. (2017). Marital status and fifteen countries. Canadian Review of remote community. Journal of Social
opposite- and same-sex status by sex for Sociology, 38, 249–274. Behavior and Personality, 14, 315–324.
persons aged 15 and over living in private Bahr, H., Caplow, T., & Chadwick, B. (1983). Clairborn, W. (1969). Expectancy effects
households for both sexes, total, presence Middletown III: Problems of replication, in the classroom: A failure to replicate.
and age of children, 2016 counts, Canada, longitudinal measurement, and triangu- Journal of Educational Psychology, 60,
provinces and territories, 2016 Census – lation. Annual Review of Sociology, 9, 377–383.
100% data. Families, Households 243–264. Cook, T., & Campbell, D. (1979). Quasi-
and Marital Status Highlight Tables. Beardsworth, A., & Keil, T. (1992). The experimentation: Design and analysis for
Statistics Canada. Retrieved from vegetarian option: Varieties, conversions, field settings. Boston: Houghton Mifflin.
http://www12.statcan.gc.ca/census- motives and careers. Sociological Review, Demers, A. (1996). Effect of support groups
recensement/2016/dp-pd/hlt-fst/fam/ 40, 253–293. on family caregivers to the frail elderly.
Table.cfm?Lang=E&T=11&Geo=00&SP= Bell, E., Aitken Schermer, J., & Vernon, P.A. Canadian Journal on Aging, 15,
1&view=1&sex=1&presence=1 (2009). The origins of political attitudes 129–144.
Strauss, A. Schatzman, L., Ehrich, D., Bucher, and behaviours: An analysis using twins. Dinovitzer, R., Hagan, J., & Parker, P. (2003).
R., & Sabshin, M. (1973). The hospital Canadian Journal of Political Science, Choice and circumstance: Social capital
and its negotiated order. In G. Salaman & 42, 855–879. and planful competence in the attain-
K. Thompson (Eds.), People and organiza- Bell, E., Jansen, H., & Young, L. (2007). ments of immigrant youth. Canadian
tions. London: Longman. Sustaining a dynasty in Alberta: The Journal of Sociology, 28, 463–488.
Sugiman, P. (2004). Memories of the intern- 2004 provincial election. Canadian Dobash, R.E., & Dobash, R.P. (2000). Evalu-
ment: Narrating Japanese-Canadian Political Science Review, 1, 27–49. ating criminal justice interventions for
women’s life stories. Canadian Journal of Bell, E., & Kandler, C. (2017). The genetic domestic violence. Crime & Delinquency,
Sociology, 29, 359–388. and the sociological: Exploring the 46, 252–270.
412 References
Dyer, W., & Wilkins, A. (1991). Better stor- Lipset, S. (1990). Continental divide: The Seabrook, J., & Avison, W. (2015). Family
ies, not better constructs, to generate values and institutions of the United structure and children’s socioeconomic
better theory: A rejoinder to Eisenhardt. States and Canada. New York: Routledge. attainment: A Canadian sample.
Academy of Management Review, 16, Lynd, R., & Lynd, H. (1929). Middletown: A Canadian Review of Sociology, 52, 66–88.
613–619. study in contemporary American culture. Shalla, V. (2002). Jettisoned by design? The
Festinger, L., Riecken, H., & Schachter, S. New York: Harcourt Brace. truncated employment relationship of
(1956). When prophecy fails. New York: Lynd, R., & Lynd, H. (1937). Middletown in customer sales and service agents under
Harper & Row. transition: A study in cultural conflicts. airline restructuring. Canadian Journal
Fisher, R., & Ma, Y. (2014). The price of New York: Harcourt Brace. of Sociology, 27, 1–32.
being beautiful: Negative effects of Mead, M. (1928). Coming of age in Samoa. Stake, R.E. (1995). The art of case study
attractiveness on empathy for children in New York: Morrow. research. Thousand Oaks, CA: Sage.
need. Journal of Consumer Research, 41. Menard, S. (1991). Longitudinal research. Statistics Canada. (2010). Detailed informa-
doi:10.1086/676967 Newbury Park, CA: Sage. tion for 2008-2009 (Cycle 8). National
Foran, C. (2010). Mordecai: The life and Michaud, S. (2001). The National Longi- Longitudinal Survey of Children and
times. Toronto: Knopf Canada. tudinal Study of Children and Youth— Youth (NLSCY). Retrieved from
Goyder, J., Guppy, N., & Thompson, Overview and changes after three cycles. http://www23.statcan.gc.ca/imdb/p2SV
M. (2003). The allocation of male and Canadian Studies in Population, 28, .pl?Function=getSurvey&SDDS=4450
female occupational prestige in an 391–405. Smith, N., Lister, R., & Middleton, S. (2004).
Ontario urban area: A quarter-century Nemni, M., & Nemni, M. (2006). Young Longitudinal qualitative research. In
replication. Canadian Review of Sociology Trudeau: Son of Quebec, father of S. Becker & A. Bryman (Eds.), Under-
and Anthropology, 40, 417–439. Canada, 1919–1944. W. Johnson (Trans.). standing research for social policy and
Greene, J. (2000). Understanding social Toronto: Douglas Gibson Books. practice: Themes, methods, and ap-
programs through evaluation. In Pawson, R., & Tilley, N. (1997). Realistic proaches. Bristol: Policy Press.
N. Denzin & Y. Lincoln (Eds.), Handbook evaluation. London: Sage. Tilley, N. (2000). Doing realistic evalua-
of Qualitative Research (2nd ed.). Phinney, A., Moody, E., & Small, J. (2014). tion of criminal justice. In V. Jupp, P.
Thousand Oaks, CA: Sage. The effect of a community-engaged arts Davies, & P. Francis (Eds.), Doing
Hanson, R.K., Broom, I., & Stephenson, program on older adults’ well-being. criminological research. London: Sage.
M. (2004). Evaluating community sex Canadian Journal on Aging, 33, 336–345. Walsh, M., Hickey, C., & Duffy, J. (1999).
offender treatment programs: A 12-year Plomin, R., Defries, J.C., Knopik, V.S., & Influence of item content and stereotype
follow-up of 724 offenders. Canadian Neiderhiser, J.M., (2013). Behavioural situation on gender differences in math-
Journal of Sociology, 36, 87–96. genetics (6th ed.). New York: Worth. ematical problem solving. Sex Roles,
Houghton, E. (1998). Sex is good for you. Pratt, A., & Valverde, M. (2002). From 41, 219–240.
Guardian, 92, 14–15. deserving victims to “masters of Walters, D. (2004). A comparison of
Huey, L. (2003). Explaining Odlin Road: confusion”: Redefining refugees in the the labour market outcomes of post
Insecurity and exclusivity. Canadian 1900s. Canadian Journal of Sociology, 27, secondary graduates of various levels
Journal of Sociology, 28, 367–386. 135–162. and fields over a four-cohort period.
Hughes, E. (1943). French Canada in tran- Radley, A., & Chamberlain, K. (2001). Health Canadian Journal of Sociology, 29, 1–27.
sition. Chicago: University of Chicago psychology and the study of the case: White, J. (1990). Hospital strike. Toronto:
Press. From method to analytic concern. Social Thompson.
Kerr, D. (2004). Family transformations Science and Medicine, 53, 321–32. Wilson, B. (2002). The Canadian rave scene
and the well-being of children: Recent Rinehart, J. (1996). The tyranny of work and five theses on youth resistance. Can-
evidence from canadian longitudinal (3rd ed.). Toronto: Harcourt Brace. adian Journal of Sociology, 27, 373–412.
data. Journal of Comparative Family Rosenthal, R., & Jacobson, L. (1968). Pygma- Yin, R. (1984). Case study research: Design
Studies, 35, 73–90. lion in the classroom: Teacher expectation and methods. Beverly Hills, CA: Sage.
Kerr, D., & Michalski, J. (2007). Family and pupils’ intellectual development. Young, N., & Dugas, E., (2012). Comparing
structure and children’s hyperactivity New York: Holt, Rinehart & Winston. climate change coverage in Canadian
problems: A longitudinal analysis. Rosnow, R.L., & Rosenthal, R. (1997). People English- and French-language print
Canadian Journal of Sociology, 32, studying people: Artefacts and ethics media: Environmental values, media
85–112. in behavioral research. New York: W.H. cultures, and the narration of global
Knott, C. (2016). Contentious mobilities Freeman. warming. Canadian Journal of Sociology,
and cheap(er) labour: Temporary foreign Russell, R., & Tyler, M. (2002). Thank heaven 37, 25–54.
workers in a New Brunswick seafood for little girls: “Girl Heaven” and the
processing community. Canadian commercial context of feminine
Chapter 3
Journal of Sociology, 41, 375–397. childhood. Sociology, 36, 619–637.
Laplante, B. (2006). The rise of cohabitation Sanderson, S.K. (2001). The evolution of Aronson, E. (1992). The social animal
in Quebec: Power of religion and power human sociality: A Darwinian conflict (6th ed.). New York: W.H. Freeman and
over religion. Canadian Journal of Sociol- perspective. Lanham, MD: Rowman & Company.
ogy, 31, 1–24. Littlefield. Bailey, K.D. (1994). Methods of social
Lewis, O. (1961). The children of Sánchez. Scriven, M. (1991). Evaluation thesaurus research (4th ed.). Toronto: Maxwell
New York: Vintage. (4th ed.). Newbury Park, CA: Sage. Macmillan Canada.
References 413
Beagan, B. (2001). Micro inequities and guards in a simulated prison. Naval van den Hoonaard, W.C., & Tolich, M.
everyday inequalities: “Race,” gender, Research Reviews, 1–17. (2014). The New Brunswick Declaration
sexuality and class in medical school. Hessler, R., Downing, J., Beltz, C., Pelliccio, of Research Ethics: A simple and radical
Canadian Journal of Sociology, 26, A., Powell, M., & Vale, W. (2003). perspective. Canadian Journal of
583–610. Qualitative research on adolescent risk Sociology, 39, 87–97.
Canadian Institutes of Health Research, using e-mail: A methodological assess- Van Maanen, J. (1991). The smile factory:
Natural Sciences and Engineering ment. Qualitative Sociology, 26, 111–124. Work at Disneyland. In P. Frost, L.
Council of Canada, and Social Sciences Homan, R. (1991). The ethics of social Moore, et al. (Eds.), Reframing organiza-
and Humanities Research Council research. London: Longman. tional culture. Newbury Park, CA: Sage.
of Canada (1998, with 2000, 2002, Homan, R., & Bulmer, M. (1982). On the Vidich, A., & Bensman, J. (1968). Small town
2005 amendments). Tri-council policy merits of covert methods: A dialogue. in mass society. Princeton, NJ: Princeton
statement: Ethical conduct for research In M. Bulmer (Ed.), Social research ethics. University Press.
involving humans. Public Works and London: Macmillan. Von Hoffman, N. (1970, January 30). The
Government Services Canada. Humphreys, L. (1970). Tearoom trade: sociological snoopers. Washington Post.
Canadian Institutes of Health Research, Impersonal sex in public places. Chicago: Zimbardo, P. (1971). The psychological
Natural Sciences and Engineering Aldine. power and pathology of imprisonment.
Council of Canada, and Social Sciences Kimmel, A. (1988). Ethics and values in Statement prepared for the U.S. House
and Humanities Research Council of applied social research. Newbury Park, of Representatives Committee on the
Canada (2014). Tri-council policy state- CA: Sage. Judiciary, Subcommittee No. 3: Hearings
ment: Ethical conduct for research Kondro, W. (2016, 22 November). Canadian on Prison Reform, San Francisco.
involving humans. Her Majesty the researcher in legal battle to keep her
Queen in Right of Canada. interviews confidential. Sciencemag.org.
Castonguay, P. (2017, 7 June). Quebec ruling Retrieved from http://www.sciencemag. Appendix to Part I
reaffirms research data confidentiality. org/news/2016/11/canadian-researcher- Orwell, G. (2004). Why I write. London:
University Affairs. Retrieved from legal-battle-keep-her-interviews- Penguin. (Original work published 1946)
https://www.universityaffairs.ca/news/ confidential
news-article/quebec-ruling-reaffirms- Milgram, S. (1963). A behavioural study
research-data-confidentiality/ of obedience. Journal of Abnormal and
Chapter 4
Charbonneau, L. (2005, 6 June). Ethics Social Psychology, 67, 371–378.
boards harming survey research, says O’Connell Davidson, J., & Layder, D. (1994). Beardsworth, A., Bryman, A., Ford, J., &
York professor. University Affairs. Methods, sex, and madness. London: Keil, T. (n.d.). “The dark figure” in
Retrieved from http://www.university Routledge. statistics of unemployment and vacan-
affairs.ca/ethics-boards-harming-survey- Punch, M. (1994). Politics and ethics in cies: Some sociological implications.
research-says-york-professor.aspx qualitative research. In N. Denzin & Discussion paper, Department of Social
Desroches, F. (1990). Tearoom trade: A Y. Lincoln (Eds.), Handbook of Qualita- Sciences, Loughborough University.
research update. Qualitative Sociology, tive Research. Thousand Oaks, CA: Sage. Bell, E., Aitken Schermer, J., & Vernon, P.A.
13, 39–61. Rosenthal, R., & Jacobson, L. (1968). Pygma- (2009). The origins of political attitudes
Le Devoir. (2016, 3 November). L’affaire lion in the classroom: Teacher expectation and behaviours: An analysis using twins.
Maillé, ou l’avenir de la confidentialité and pupils’ intellectual development. Canadian Journal of Political Science, 42,
dans la recherche scientifique. Le Devoir. New York: Holt, Rinehart & Winston. 855–879.
Retrieved from https://www.ledevoir. Small, W., Maher, L., & Kerr, T. (2014). Berthoud, R. (2000). A measure of changing
com/opinion/libre-opinion/483756/l- Institutional ethical review and ethno- health. In R. Berthoud & J. Gershuny
affaire-maille-ou-l-avenir-de-la- graphic research involving injection (Eds.), Seven years in the lives of British
confidentialite-dans-la-recherche- drug users: A case study. Social Science & families: Evidence on the dynamics
scientifique Medicine, 104, 157–162. of social change from the British
Dinovitzer, R., Hagan, J., & Parker, P. (2003). Totten, M. (2001). Legal, ethical, and clinical Household Panel Survey. Bristol: Policy
Choice and circumstance: Social capital implications of doing fieldwork with Press.
and planful competence in the attain- youth gang members who engage in Bourdieu, P. (1987). The historical genesis of
ments of immigrant youth. Canadian serious violence. Journal of Gang a pure aesthetic. Journal of Aesthetics and
Journal of Sociology, 28, 463–488. Research, 8, 35–49. Art Criticism, 46 (Analytic Aesthetics),
Festinger, L., Riecken, H., & Schachter, S. van den Hoonaard, W.C. (2001). Is research- 201–210.
(1956). When prophecy fails. New York: ethics review a moral panic? Canadian Blumer, H. (1956). Sociological analysis and
Harper & Row. Review of Sociology and Anthropology, the “variable.” American Sociological
Gans, H. J. (1962). The urban villagers. 38, 19–36. Review, 21, 683–690.
New York: Free Press. van den Hoonaard, W.C., & Connolly, A. Brannigan, A. (2004). The rise and fall of
Goode, E. (1996). The ethics of deception in (2006). Anthropological research in light social psychology: The use and misuse of
social research: A case study. Qualitative of research-ethics review: Canadian the experimental method. Hawthorne,
Sociology, 19, 11–33. master’s theses, 1995–2004. Journal of New York: Aldine de Gruyter.
Haney, C., Banks, C., & Zimbardo, P. (1973, Empirical Research on Human Research Bridgman, P. (1927). The logic of modern
September). A study of prisoners and Ethics, 1, 59–69. physics. New York: Macmillan.
414 References
Bryman, A., & Cramer, D. (2001). Quanti- importance of neighbourhood: A central Andrews, K., Smith, L., Henzi, D., & Demps,
tative data analysis with SPSS release 10 city/suburb comparison. Canadian E. (2007). Faculty and student percep-
for Windows: A guide for social scientists. Journal of Sociology, 38, 359–383. tions of academic integrity at U.S. and
London: Routledge. LaPiere, R.T. (1934). Attitudes vs. actions. Canadian dental schools. Journal of
Bulmer, M. (1984). Facts, concepts, theories Social Forces, 13, 230–237. Dental Education, 71, 1027–1039.
and problems. In M. Bulmer (Ed.), Social McArdle, M. (2017, 15 July). Having an affair Bampton, R., & Cowton, C.J. (2002). The
research methods. London: Macmillan. seems to be going out of style. National e-interview. Forum: Qualitative Social
Cicourel, A. (1964). Method and measurement Post. Retrieved from http://nationalpost. Research, 3. Retrieved from http://www
in sociology. New York: Free Press. com/opinion/having-an-affair-seems-to- .qualitative-research.net/index.php/fqs/
Cicourel, A. (1982). Interviews, surveys, be-going-out-of-style article/view/848
and the problem of ecological validity. Murphy, L., & Fedoroff, J. (2013). Sexual Beagan, B. (2001). Micro inequities and every-
American Sociologist, 17, 11–20. offenders’ views of Canadian sex day inequalities: “Race,” gender, sexuality
Cramer, D. (1998). Fundamental statistics for offender registries: A survey of a clinical and class in medical school. Canadian
social research. London: Routledge. sample. Canadian Journal of Behavioural Journal of Sociology, 26, 583–610.
Deng, J., Walker, G., & Swinnerton, G. Science, 45, 238–249. Beardsworth, A., & Keil, T. (1997). Sociology
(2006). A comparison of environmental Reiner, R. (2000). Crime and control in on the menu: An invitation to the study of
values and attitudes between Chinese Britain. Sociology, 34, 71–94. food and society. London: Routledge.
in Canada and Anglo-Canadians. Ricciardelli, R., & Clow, K. (2009). Men, Beardsworth, A., Keil, T., Goode, J., Haslam,
Environment and Behavior, 38, 22–47. appearance, and cosmetic surgery: The C., & Lancashire, E. (2002). Women, men
Edin, K., & Kefalas, M. (2005). Promises role of self-esteem and comfort with the and food: The significance of gender for
I can keep: Why poor women put mother- body. Canadian Journal of Sociology, 34, nutritional attitudes and choices. British
hood before marriage. Berkeley: 105–134. Food Journal, 104, 470–491.
University of California Press. Schieman, S., & Narisada, A. (2014). In control Bell, E., Aitken Schermer, J., & Vernon, P.A.
Foddy, W. (1993). Constructing questions or fatalistically ruled? The sense of mastery (2009). The origins of political attitudes
for interviews and questionnaires: among working Canadians. C anadian and behaviours: An analysis using twins.
Theory and practice in social research. Review of Sociology, 51, 343–374. Canadian Journal of Political Science, 42,
Cambridge: Cambridge University Press. Schuman, H., & Presser, S. (1981). Questions 855–879.
Frazer, R., & Wiersma, U. (2001). Prejudice and answers in attitude surveys: Experi- Bell, E., & Kandler, C. (2017). The genetic
vs. discrimination in the employment ments on question form, wording, and and the sociological: Exploring the
interview: We may hire equally, but our context. San Diego, CA: Academic Press. possibility of consilience. Sociology, 51,
memories harbour prejudice. Human Silverman, R., Sacco, V., & Teevan, J. (2000). 880–896.
Relations, 54, 173–191. Crime in Canadian society. Toronto: Beran, T., Hughes, G., & Lupart, J. (2008).
Gabor, T., Hung, K., Mihorean, S., & St-Onge, Harcourt Canada. A model of achievement and bullying:
C. (2002). Canadian homicide rates: A Sobel, N. (2015). A typology of the changing Analyses of the Canadian National
comparison of two data sets. Canadian narratives of Canadian citizens through Longitudinal Survey of Children and
Journal of Criminology, 44, 351–363. time. Canadian Ethnic Studies, 47, 11–39. Youth data. Educational Research, 50,
Gazso-Windlej, A., & McMullin, J. (2003). Sutton, R.I. , & Rafaeli, A. (1992). How we 25–39.
Doing domestic labour: Strategizing in a untangled the relationship between dis- Beres, M.A., Crow, B., & Gotell, L. (2009).
gendered domain. Canadian Journal of played emotion and organizational sales: The perils of institutionalization in
Sociology, 28, 341–366. A tale of bickering and optimism. In neoliberal times: Results of a national
George, U., & Chaze, F. (2014). Discrimina- P. Frost & R. Stablein (Eds.), Doing exem- survey of Canadian sexual assault and
tion at work: Comparing the experiences plary research. Newbury Park, CA: Sage. rape crisis centres. Canadian Journal of
of foreign-trained and locally-trained Teevan, J., & Dryburgh, H. (2000). First Sociology, 34, 135–163.
engineers in Canada. Canadian Ethic person accounts and sociological explan- Blauner, R. (1964). Alienation and freedom:
Studies, 46, 1–21. ations of delinquency. Canadian Review The factory worker and his industry.
Goyder, J., Guppy, N., & Thompson, M. of Sociology and Anthropology, 37, 77–93. Chicago: University of Chicago Press.
(2003). The allocation of male and female Boivin, R., & Leclerc, C. (2016). Domestic
occupational prestige in an Ontario violence reported to the police: Correl-
Chapter 5
urban area: A quarter-century replica- ates of victims’ reporting behavior and
tion. Canadian Review of Sociology and Ames, M., Rawana, J., Gentile, P., & Morgan, support to legal proceedings. Violence
Anthropology, 40, 417–439. A. (2015). The protective role of opti- and Victims, 31, 402–415.
Guppy, L., & Siltanen, J. (1977). A comparison mism and self-esteem on depressive Bottomore, T.B., & Rubel, M. (1963). Karl
of the allocation of male and female oc- symptom pathways among Canadian Marx: Selected writings in sociology
cupational prestige. Canadian Review of Aboriginal youth. Journal of Youth and and social philosophy. Harmondsworth,
Sociology and Anthropology, 14, 320–330. Adolescence, 44, 142–154. UK: Penguin.
Hay, D. (2014). An investigation into the Anderson, K., Sebaldt, R., Lohfeld, L., Brannigan, A., Gemmell, W., Pevalin, D., &
swiftness and intensity of recent secu- Burgess, K., Donald, F., & Kaczorowski, Wade, T. (2002). Self-control and social
larization in Canada: Was Berger right? J. (2006). Views of family physicians in control in childhood misconduct and
Sociology of Religion, 75, 136–162. southwestern Ontario on preventive care aggression: The role of family structure
Kennedy, E., Krogman, N., & Krahn, H. services and performance incentives. and hyperactivity. Canadian Journal of
(2013). Sustainable consumption and the Family Practice, 23, 469–471. Criminology, 44, 119–142.
References 415
Brownridge, D.A., Taillieu, T., Afifi, T., Goyder, J., Guppy, N., & Thompson, M. Légaré, J., Décarie, Y., & Bélanger, A. (2014).
Chan, K.L., Emery, C., Lavoie, J., & Elgar, (2003). The allocation of male and female Using microsimulation to reassess aging
F. (2017). Child maltreatment and intim- occupational prestige in an Ontario trends in Canada. Canadian Journal on
ate partner violence among Indigenous urban area: A quarter-century replica- Aging, 33, 208–219.
and non-Indigenous Canadians. Journal tion. Canadian Review of Sociology and Li, P. (2003). Initial earnings and catch-up
of Family Violence, 32, 607–619. Anthropology, 40, 417–439. capacity of immigrants. Canadian Public
Cobanoglu, C., Ward, B., & Moreo, P.J. Grabb, E., & Curtis, J. (2004). Regions apart: Policy, 29, 319–337.
(2001). A comparison of mail, fax and The four societies of Canada and the U.S. Mann, C., & Stewart, F. (2000). Internet
web-based survey methods. International Don Mills, ON: Oxford University Press. communication and qualitative research:
Journal of Market Research, 43, 441–452. Hay, D. (2014). An investigation into the A handbook for researching online.
Coleman, C., & Moynihan, J. (1996). Under- swiftness and intensity of recent secu- London: Sage.
standing crime data: Haunted by the dark larization in Canada: Was Berger right? Marin, A., & Hayes, S. (2017). The occupational
figure. Buckingham: Open University Sociology of Religion, 75, 136–162. context of mismatch: Measuring the link
Press. Hessler, R., Downing, J., Beltz, C., Pelliccio, between occupations and areas of study
Corti, L. (1993). Using diaries in social A., Powell, M., & Vale, W. (2003). Quali- to better understand job-worker match.
research. Social Research Update, 2. tative research on adolescent risk using Canadian Journal of Sociology, 42, 1–22.
Crook, C., & Light, P. (2002). Virtual society e-mail: A methodological assessment. Mata, F., & Pendakur, R. (2017). Of intake
and the cultural practice of study. In Qualitative Sociology, 26, 111–124. and outcomes: Wage trajectories of im-
S. Woolgar (Ed.), Virtual society? Holbrook, A., Green, M., & Krosnick, J. migrant classes in Canada. Journal of
Technology, cyberbole, reality. Oxford: (2003). Telephone versus face-to-face International Migration and Integration,
Oxford University Press. interviewing of national probability 18, 829–844.
Curasi, C. (2001) A critical exploration of samples with long questionnaires: Com- Mayhew, P. (2000). Researching the state of
face-to-face interviewing vs. computer- parisons of respondent satisficing and crime. In R. King & E. Wincup (Eds.),
mediated interviewing. International social desirability response bias. Public Doing research on crime and justice.
Journal of Market Research, 43, 361–375. Opinion Quarterly, 67, 79–125. Oxford: Oxford University Press.
Dale, A., Arber, S., & Proctor, M. (1988). Ipsos Reid. (2015). Three in ten (28%) O’Connor, H., & Madge, C. (2001).
Doing secondary analysis. London: Canadians can’t confirm Canada’s Con- Cyber-mothers: Online synchronous
Unwin Hyman. federation year. Retrieved from https:// interviewing using conferencing soft-
Durkheim, E. (1952). Suicide: A study in www.historicacanada.ca/sites/default/ ware. Sociological Research Online, 5.
sociology. J. Spaulding & G. Simpson files/PDF/polls/Historica%20Canada%20 Retrieved from www.socresonline.org.
(Trans.). London: Routledge & Kegan SJAM%20Factum.pdf uk/5/4/o’connor.html.
Paul. (Original work published 1897) Kelley, J., & De Graaf, N. (1997). National con- O’Connor, H., & Madge, C. (2003). “Focus
Elliott, H. (1997). The use of diaries in text, parental socialization, and religious groups in cyberspace”: Using the Inter-
sociological research on health experi- belief: Results from 15 nations. American net for qualitative research. Qualitative
ence. Sociological Research Online, 2. Sociological Review, 62, 639–659. Market Research, 6, 133–143.
Retrieved from http://www.socresonline Kent, R., & Lee, M. (1999). Using the Internet O’Sullivan, L., Udell, W., & Patel, V. (2006).
.org.uk/socresonline/2/2/7.html for market research: A study of private Young urban adults’ heterosexual risk
Finch, J. (1987). The vignette technique in trading on the Internet. Journal of the encounters and perceived risk and safety:
survey research. Sociology, 21, 105–114. Market Research Society, 41, 377–385. A structured diary study. Journal of Sex
Fournier, P., Cutler, F., Soroka, S., & Stolle, Kerr, D. (2004). Family transformations and Research, 43, 343–351.
D. (2015). The 2015 Canadian Election the well-being of children: Recent evidence Pahl, J. (1990). Household spending, personal
Study. Retrieved from https://ces-eec from canadian longitudinal data. Journal spending and the control of money in
.arts.ubc.ca/english-section/surveys/ of Comparative Family Studies, 35, 73–90. marriage. Sociology, 24, 119–138.
Fowler, F. (1993). Survey research methods Kevins, A., & Soroka, S. (2018). Growing Robinson, J., & Lee, C. (2014). Society’s
(2nd ed.). Newbury Park, CA: Sage. apart? Partisan sorting in Canada, (virtually) time-free transition into the
Frey, J. (2004). Telephone surveys. In M. 1992–2015. Canadian Journal of Political digital age. Social Indicators Research,
Lewis-Beck, A. Bryman, & T. Liao (Eds.), Science, 51, 103–133. 117, 939–965.
The Sage Encyclopedia of social science Kingsbury, M., & Coplan, R. (2012). Robinson, J., Shaver, P., & Wrightsman, L.
research methods. Thousand Oaks, CA: Mothers’ gender-role attitudes and their (1999). Measures of political attitudes.
Sage. responses to young children’s hypothet- Toronto: Academic Press.
Gidengil, E., Everitt, J., Blais, A., Fournier, ical display of shy and aggressive behav- Schaeffer, D., & Dillman, D. (1998). Develop-
P., & Nevitte, N. (2006). Gender and vote iours. Sex Roles, 66, 506–517. ment of a standard e-mail methodology.
choice in the 2006 Canadian election. Krosnick, J., Holbrook, A. et al. (2002). The Public Opinion Quarterly, 62, 378–397.
Paper prepared for the Annual Meeting impact of “no opinion” response options Schuman, H., & Presser, S. (1981). Questions
of the American Political Science Associ- on data quality: Non-attitude reduction and answers in attitude surveys: Experi-
ation, Philadelphia. or an invitation to satisfice? Public ments on question form, wording, and
Goodyear-Grant, E., & Croskill, J., (2011). Opinion Quarterly, 66, 371–403. context. San Diego, CA: Academic Press.
Gender affinity effects in vote choice in Laurenceau, J.-P., & Bolger, N. (2005). Sheehan, K. and Hoy, M. (1999). Using
westminster systems: Assessing “flexible” Using diary methods to study marital e-mail to survey Internet users in the
voters in Canada. Politics and Gender, 7, and family processes. Journal of Family United States: Methodology and assess-
223–250. Psychology, 19, 86–97. ment. Journal of Computer-Mediated
416 References
Communication, 4. Retrieved from www over a four-cohort period. Canadian change, say lawyers, academics. Montreal
.ascusc.org/jcmc/vol4/issue3/sheehan.html Journal of Sociology, 29, 1–27. Gazette. Retrieved from http://montreal
Shuy, R.W. (2002). In-person versus telephone Wright, K.B. (2005). Researching Internet- gazette.com/news/local-news/quebecs-
interviewing. In J.F. Gubrium & J.A. based populations: Advantages and dis- clogged-justice-system-needs-big-
Holstein (Eds.), Handbook of interview advantages of online survey research, culture-change-say-lawyers-academics
research: Context and method. Thousand online questionnaire authoring software Webb, E.J., Campbell, D.T., Schwartz, R.D., &
Oaks, CA: Sage. packages, and web survey services. Sechrest, L. (1966). Unobtrusive measures:
Simons, P., & Clancy, C. (2017, 29 June). On Journal of Computer-Mediated Com- Non-reactive measures in the social sciences.
point: Fifty years ago, Canada changed munication, 10. https://doi.org/10.1111/ Chicago: Rand McNally.
its immigration rules and in doing so j.1083-6101.2005.tb00259.x Weber, M. (1947). The theory of social and eco-
changed the face of this country. Ed- nomic organization. A.M. Henderson &
monton Journal. Retrieved from http:// T. Parsons (Trans.). New York: Free Press.
Chapter 6
edmontonjournal.com/news/insight/ Whyte, W.F. (1955). Street corner society
on-point-fifty-years-ago-canada-changed- Bales, Robert (1951). Interaction process an- (2nd ed.). Chicago: University of Chicago
its-immigration-rules-and-in-doing-so- alysis. Cambridge, MA: Addison-Wesley. Press.
changed-the-face-of-this-country Bell, E. (2007). Separatism and quasi-
Sinclair, M., O’Toole, J., Malawaraarachchi, separatism in Alberta. Prairie Forum, 32,
Chapter 7
M., & Leder, K. (2012). Comparison of 335–355.
response rates and cost-effectiveness Charlton, T., Gunter, B., & Coles, D. (1998). Beagan, B. (2001). Micro inequities and every-
for a community-based survey: postal, Broadcast television as a cause of day inequalities: “Race,” gender, sexuality
Internet and telephone modes with aggression? Recent findings from a nat- and class in medical school. Canadian
generic or personalized recruitment uralistic study. Emotional and Journal of Sociology, 26, 583–610.
approaches. BMC Medical Research Behavioural Difficulties, 3, 5–13. Becker, H. (1963). Outsiders: Studies in the
Methodology, 12, 1–8. Galton, M., Simon, B., & Croll, P. (1980). sociology of deviance. New York: Free
Smith, D.J., & McVie, S. (2003). Theory and Inside the primary classroom. London: Press.
method in the Edinburgh Study of Youth Routledge & Kegan Paul. Bell, E. (2007). Separatism and quasi-
Transitions and Crime. British Journal of Katz, J., Kuffel, S., & Coblentz, A. (2002). separatism in Alberta. Prairie Forum, 32,
Criminology, 43, 169–195. Are there gender differences in sustain- 335–355.
Statistics Canada. (2013). The General Social ing dating violence? Journal of Family Bowen, R. (2015). Squaring up: Experiences
Survey: An overview. Retrieved from Violence, 17, 247–271. of transition from off-street sex work to
http://www.statcan.gc.ca/pub/ LaPiere, R.T. (1934). Attitudes vs. actions. square work and duality—concurrent
89f0115x/89f0115x2013001-eng.htm Social Forces, 13, 230–237. involvement in both—in Vancouver,
Sudman, S., & Bradburn, N. (1982). Asking McCall, M.J. (1984). Structured field obser- BC. Canadian Review of Sociology, 52,
questions: A practical guide to question- vation. Annual Review of Sociology, 10, 429–449.
naire design. San Francisco: Jossey-Bass. 263–282. Bryman, A. (1995). Disney and his worlds.
Sullivan, O. (1996). Time co-ordination, the Normand, S., Schneider, B., Lee, M., Maison- London: Routledge.
domestic division of labour and affective neuve, M., C hupetlovska-Anastasova, A., Bryman, A. (1999). Global Disney. In
relations: Time use and the enjoyment of Kuehn, S., & Robaey, P. (2013). Continu- P. Taylor & D. Slater (Eds.), The American
activities within couples. Sociology, 30, ities and changes in the friendships of century. Oxford: Blackwell.
79–100. children with and without ADHD: Buchanan, C. (2014). A more national
Tourangeau, R., & Smith, T.W. (1996). A longitudinal, observational study. representation of place in Canadian daily
Asking sensitive questions: The impact Journal of Abnormal Child Psychology, newspapers. Canadian Geographer, 58,
of data collection mode, question format, 41, 1161–1175. 517–530.
and question context. Public Opinion Perrucci, R., Belshaw, R., DeMerritt, A., Butcher, B. (1994). Sampling methods: An
Quarterly, 60, 275–304. Frazier, B., Jones, J., Kimbrough, J., . . . overview and review. Survey Methods
Tse, A. (1998). Comparing the response rate, Williams, B. (2000). The two faces of Centre Newsletter, 15, 4–8.
response speed and response quality of racialized space at a predominantly white Charlton, J. (2017, 3 February). Researchers
two methods of sending questionnaires: university. International Journal of shine a light on challenges facing Sask
E-mail vs. mail. Journal of the Market Contemporary Sociology, 37, 230–244. migrant workers. Saskatoon StarPhoenix.
Research Society, 40, 353–361. Roethlisberger, F., & Dickson, W. (1939). Retrieved from http://thestarphoenix.
Walby, S., & Myhill, A. (2001). New survey Management and the worker. Cambridge, com/business/local-business/researchers-
methodologies in researching violence MA: Harvard University Press. shine-a-light-on-challenges-facing-
against women. British Journal of Rosenhan, D.L. (1973). On being sane in sask-migrant-workers
Criminology, 41, 502–522. insane places. Science, 179, 350–8. Czaja, R., & Blair, J. (1996). Designing sur-
Walklate, S. (2000). Researching victims. Schulenberg, J.L. (2014). Systematic social veys: A guide to decisions and procedures.
In R. King & E. Wincup (Eds.), Doing observation of police decision-making: Thousand Oaks, CA: Sage.
research on crime and justice. Oxford: The process, logistics, and challenges in a Dinovitzer, R., Hagan, J., & Parker, P. (2003).
Oxford University Press. Canadian context. Quality and Quantity, Choice and circumstance: Social capital
Walters, D. (2004). A comparison of the 48, 297–315. and planful competence in the attain-
labour market outcomes of postsecond- Valiante, G. (2017, 26 February). Quebec’s ments of immigrant youth. Canadian
ary graduates of various levels and fields clogged justice system needs big culture Journal of Sociology, 28, 463–488.
References 417
Finch, J., & Hayes, L. (1994). Inheritance, development in the Canadian Maritimes: Chapter 9
death and the concept of the home. Outsiders within? Canadian Journal of
Sociology, 28, 417–433. Sociology, 28, 203–234. Abramovich, A. (2016). Preventing, reducing
Fowler, F. (1993). Survey research methods Teitler, J., Reichman, N., & Sprachman, S. and ending LGBTQ2S youth homeless-
(2nd ed.). Newbury Park, CA: Sage. (2003). Costs and benefits of improv- ness: The need for targeted strategies.
Frey, J. (2004). Telephone surveys. In M. ing response rates for a hard-to-reach Social Inclusion, 4, 86–96.
Lewis-Beck, A. Bryman, & T. Liao (Eds.), population. Public Opinion Quarterly, 67, Armstrong, G. (1993). Like that Desmond
The Sage encyclopedia of social science 126–138. Morris? In D. Hobbs & T. May (Eds.),
research methods. Thousand Oaks, CA: Warde, A. (1997). Consumption, food and Interpreting the field: Accounts of ethnog-
Sage. taste. London: Sage. raphy. Oxford: Clarendon Press.
Gaudet, S., Cooke, M., & Jacob, J. (2011). Whalen, H., & Schmidt, G. (2016). The Belk, R., Sherry, J., & Wallendorf, M. (1988).
Working after childbirth: A lifecourse women who remain behind: challenges in A naturalistic inquiry into buyer and
transition analysis of Canadian women the LDC lifestyle. Rural Society, 25, 1–14. seller behaviour at a swap meet. Journal
from the 1970s to the 2000s. Canadian of Consumer Research, 14, 449–470.
Review of Sociology, 48, 153–180. Bell, E. (2007). Separatism and quasi-
Chapter 8 separatism in Alberta. Prairie Forum,
Hallett, R., & Barber, K. (2014). Ethnographic
research in a cyber era. Journal of Con- Anderssen, E. (2014, 2 October). Big data is 32, 335–355.
temporary Ethnography, 43, 306–30. watching you. Has online spying gone Bloor, M. (1997). Addressing social problems
Johnson, T., & Wislar, J. (2012). Response too far? The Globe and Mail. Retrieved through qualitative research. In D.
rates and nonresponse errors in surveys. from https://www.theglobeandmail.com/ Silverman (Ed.), Qualitative research:
Journal of the American Medical life/relationships/big-data-is-watching- Theory, method and practice. London: Sage.
Association, 307, 1805–1806. you-has-online-spying-gone-too-far/ Blumer, H. (1954). What is wrong with social
Kennedy, E., Krogman, N, & Krahn, H. article20894498/?page=all theory? American Sociological Review,
(2013). Sustainable consumption and the Agresti, A., & Finlay, B. (2009). Statistical 19, 3–10.
importance of neighbourhood: A central methods for the social sciences. Upper Bochner, A., & Ellis, C. (2003). An introduc-
city/suburb comparison. Canadian Saddle River, NJ: Prentice-Hall. tion to the arts and narrative research: Art
Journal of Sociology, 38, 359–383. Bryman, A. and Cramer, D. (2001). Quanti- as inquiry. Qualitative Inquiry, 9, 506–514.
MacKinnon, N., & Luke, A. (2002). Changes tative data analysis with SPSS release 10 Branker, R. (2017). Labour market dis-
in identity attitudes as reflections of for Windows: A guide for social scientists. crimination: The lived experiences of
social and cultural change. Canadian London: Routledge. English-speaking Caribbean immigrants
Journal of Sociology, 27, 299–338. Cao, L. (2014). Aboriginal people and con- in Toronto. Journal of International
Mairs, K., & Bullock, S. (2013). Sexual-risk fidence in the police. Canadian Journal Migration & Integration, 18, 203–222.
behaviour and HIV testing among of Criminology and Criminal Justice, 56, Bryman, A. (1994). The Mead/Freeman
Canadian snowbirds who winter in 499–525. controversy: Some implications for
Florida. Canadian Journal on Aging, 32, Dhillon, S. (2015, 29 September). Researchers qualitative researchers. In R. Burgess
145–158. to study big data collection used on (Ed.), Studies in Qualitative Methodology
Mangione, T.W. (1995). Mail surveys: Canadians. The Globe and Mail. Re- (Vol. 4). Greenwich, CT: JAI Press.
Improving the quality. Thousand Oaks, trieved from https://www.theglobeand Bryman, A., Haslam, C., & Webb, A. (1994).
CA: Sage. mail.com/news/british-columbia/ Performance appraisal in UK universi-
Marsh, C., & Scarbrough, E. (1990). Testing researchers-will-study-big-data- ties: A case of procedural compliance?
nine hypotheses about quota sampling. collection-used-on-canadians/ Assessment and Evaluation in Higher
Journal of the Market Research Society, article26595244/ Education, 19, 175–188.
32, 485–506. Healey, J., & Prus, S.G. (2016). Statistics: Burman, M., Batchelor, S., & Brown, J.
Parnaby, P. (2003). Disaster through dirty A tool for social research. (3rd Cdn ed.). (2001). Researching girls and violence:
windshields: Law, order, and Toronto’s Toronto: Nelson Education. Facing the dilemmas of fieldwork. British
squeegee kids. Canadian Journal of Rippeyoung, P. (2013). Can breastfeeding Journal of Criminology, 41, 443–459.
Sociology, 28, 281–307. solve inequality? The relative mediating Coloma, R., & Pino, F. (2016). “There’s
Schieman, S., & Narisada, A. (2014). In impact of breastfeeding and home en- hardly anything left”: Poverty and the
control or fatalistically ruled? The sense vironment on poverty gaps in Canadian economic insecurity of elderly Filipinos
of mastery among working Canadians. child cognitive skills. Canadian Journal in Toronto. Canadian Ethnic Studies, 48,
Can. Review of Sociology, 51, 343–374. of Sociology, 38, 65–85. 71–97.
Shuy, R.W. (2002). In-person versus tele- Rojek, C. (1995). Decentring leisure: Rethink- Finlay, L. (2006). “Rigour,” “ethical integrity”
phone interviewing. In J.F. Gubrium ing leisure theory. London: Sage. or “artistry”? Reflexively reviewing cri-
& J.A. Holstein (Eds.), Handbook of Ryerson University. (n.d.). About. Privacy teria for evaluating qualitative research.
interview research: Context and method. and Big Data Institute. Retrieved from British Journal of Occupational Therapy,
Thousand Oaks, CA: Sage. https://www.ryerson.ca/pbdi/About/ 69, 319–326.
Smith, T.W. (1995). Trends in non-response Sutton, R.I., & Rafaeli, A. (1988). Untangling Foster, J. (1995). Informal social control and
rates. International Journal of Public the relationship between displayed community crime prevention. British
Opinion Research, 7, 157–171. emotions and organizational sales: The Journal of Criminology, 35, 563–583.
Tastsoglou, E., & Miedema, B. (2003). case of convenience stores. Academy of Gans, H. J. (1962). The urban villagers.
Immigrant women and community Management Journal, 31, 461–487. New York: Free Press.
418 References
Geertz, C. (1973). Thick description: Toward the public sector. Canadian Journal of Becker, H. (1970). Practitioners of vice and
an interpretive theory of culture. In Sociology, 37, 357–379. crime. In R. Habenstein (Ed.), Pathways
The Interpretation of Cultures. New York: McKee, L., & Bell, C. (1985). Marital to data. Chicago: Aldine.
Basic Books. and family relations in times of male Bell, E. (2007). Separatism and quasi-
Green, A.I. (2010). Queer unions: Same-sex unemployment. In B. Roberts, separatism in Alberta. Prairie Forum, 32,
spouses marrying tradition and innova- R. Finnegan, & D. Gallie (Eds.), New 335–355.
tion. Canadian Journal of Sociology, 35, approaches to economic life. Manchester: Breckenridge, J., & Jones, D. (2009).
399–436. Manchester University Press. Demystifying theoretical sampling in
Guba, E., & Lincoln, Y. (1994). Competing O’Reilly, K. (2000). The British on the Costa grounded theory research. Grounded
paradigms in qualitative research. In del Sol: Transnational identities and local Theory Review, 8, 113–126.
N. Denzin & Y. Lincoln (Eds.), Handbook communities. London: Routledge. Browne, A.J. (2007). Clinical encounters
of qualitative research. Thousand Oaks, Pettigrew, A. (1997). What is a processual between nurses and First Nations women
CA: Sage. analysis? Scandinavian Journal of in a western Canadian hospital. Social
Hallgrimsdottir, H., Phillips, R., & Benoit, C. Management, 13, 337–348. Science & Medicine, 64, 2165–2176.
(2006). Fallen women and rescued girls: Renzetti, E. (2017, 6 January). Women killed Campbell, M. (2000). Research on health
Social stigma and media narratives of the by their spouses are not casualties in care experiences of people with disabil-
sex industry in Victoria, B.C., from 1980 someone else’s story. The Globe and Mail. ities: Exploring the social organization
to 2005. Canadian Review of Sociology Retrieved from https://www.theglobe of service delivery. Canadian Journal of
and Anthropology, 43, 265–280. andmail.com/opinion/women-killed- Sociology, 1, 131–154.
Hammersley, M. (1992). Deconstructing by-their-spouses-are-not-casualties-in- Charmaz, K. (2000). Grounded theory:
the qualitative–quantitative divide. In someone-elses-story/article33535098 Objectivist and constructivist methods.
What’s wrong with ethnography? London: Sharma, S., Reimer-Kirkham, S., & Meyerhoff, In N. Denzin & Y. Lincoln (Eds.), Hand-
Routledge. H. (2011). Filmmaking with Aboriginal book of qualitative research (2nd ed.).
Hiller, H., & DiLuzio, L. (2004). The inter- youth for type 2 diabetes prevention. Thousand Oaks, CA: Sage.
viewee and the research interview: Pimatisiwin: A Journal of Aboriginal and Coffey, A. (1999). The ethnographic self:
Analyzing a neglected dimension in Indigenous Community Health, 9, 423–440. Fieldwork and the representation of
research. Canadian Review of Sociology Shelden, D., Angell, M., Stoner, J., & reality. London: Sage.
and Anthropology, 41, 1–26. Roseland, B. (2010). School principals’ Desroches, F. (1990). Tearoom trade: A re-
Hochschild, A. (1983). The managed heart. influence on trust: Perspectives of moth- search update. Qualitative Sociology, 13,
Berkeley and Los Angeles: University of ers of children with disabilities. Journal 39–61.
California Press. of Educational Research, 103, 159–170. Devault, M., (2006). Introduction: What is
Karabanow, J. (2002). Open for business: Silverman, D. (1993). Interpreting qualitative institutional ethnography? Social Prob-
Exploring the life stages of two Canadian data: Methods for analysing qualitative lems, 53, 294–298.
street youth shelters. Journal of Sociology data. London: Sage. Ditton, J. (1977). Part-time crime: An ethnog-
and Social Welfare, 29, 99–116. Skeggs, B. (1994). Situating the production raphy of fiddling and pilferage. London:
LeCompte, M., & Goetz, J. (1982). Problems of feminist ethnography. In M. Maynard Macmillan.
of reliability and validity in ethnographic & J. Purvis (Eds.), Researching women’s Fine, G. (1996). Justifying work: Occupa-
research. Review of Educational Research, lives from a feminist perspective. London: tional rhetorics as resources in kitchen
52, 31–60. Taylor & Francis. restaurants. Administrative Science
Leidner, R. (1993). Fast food, fast talk: Service Skeggs, B. (1997). Formations of class and Quarterly, 41, 90–115.
work and the routinization of everyday gender. London: Sage. Gans, H. J. (1968). The participant-observer
life. Berkeley and Los Angeles: University Taylor, A. (1993). Women drug users: An as human being: Observations on
of California Press. ethnography of an injecting community. the personal aspects of field work. In
Lewis, O. (1961). The children of Sánchez. Oxford: Clarendon Press. H. Becker (Ed.), Institutions and the
New York: Vintage. Tracy, S. (2010). Qualitative quality: Eight person: Papers presented to Everett C.
Lincoln, Y., & Guba, E. (1985). Naturalistic “big tent” criteria for excellent qualita- Hughes. Chicago: Aldine.
inquiry. Beverly Hills, CA: Sage. tive research. Qualitative Inquiry, 16, Gerson, K., & Horowitz, R. (2002). Obser-
Lofland, J., & Lofland, L. (1995). Analyzing 837–851. vation and interviewing: Options and
social settings: A guide to qualitative Whyte, W.F. (1955). Street corner society (2nd choices. In T. May (Ed.), Qualitative
observation and analysis. (3rd ed.). Bel- ed.). Chicago: University of Chicago Press. research in action. London: Sage.
mont, CA: Wadsworth. Williams, M. (2000). Interpretivism and Giulianotti, R. (1995). Participant observa-
Lyon, K., & Frohard-Dourlent, H. (2015). generalization. Sociology, 34, 209–224. tion and research into football hooli-
Let’s talk about the institution: Same- ganism: Reflections on the problems of
sex common-law partners negotiating entrée and everyday risks. Sociology of
Chapter 10
marriage equality and relationship legit- Sport Journal, 12, 1–20.
imacy. Canadian Review of Sociology, 52, Armstrong, G. (1993). Like that Desmond Glaser, B., & Strauss, A. (1967). The discovery
402–428. Morris? in D. Hobbs & T. May (Eds.), of grounded theory: Strategies for qualita-
Mason, J. (2002). Qualitative researching Interpreting the field: accounts of tive research. Chicago: Aldine.
(2nd ed.). London: Sage. ethnography. Oxford: Clarendon Press. Glucksmann, M. (1994). The work of know-
McDonough, P., & Polzer, J. (2012). Habitus, Atkinson, P. (1981). The clinical experience. ledge and the knowledge of women’s
hysteresis, and organizational change in Farnborough, UK: Gower. work. In M. Maynard & J. Purvis (Eds.),
References 419
Armstrong, G. (1993). Like that Desmond In J. Gubrium & J. Holstein (Eds.), between research participants. Sociology
Morris? in D. Hobbs & T. May (Eds.), Handbook of interview research: Context of Health and Illness, 16, 103–121.
Interpreting the field: Accounts of ethnog- and method. Thousand Oaks, CA: Sage. Krueger, R. (1998). Moderating focus groups.
raphy. Oxford: Clarendon Press. Clapper, D., & Massey, A. (1996). Electronic Thousand Oaks, CA: Sage.
Atkinson, M. (2002). Pretty in ink: Conform- focus groups: A framework for explora- Kvale, S. (1996). Interviews: An introduction
ity, resistance and negotiation in women’s tion. Information and Management, 30, to qualitative research interviewing.
tattooing. Sex Roles, 47, 219–235. 43–50. Thousand Oaks, CA: Sage.
Atkinson, M. (2004). Tattooing and civilizing Curasi, C. (2001). A critical exploration of Lareau, A., & Weininger, E., (2003). Cultural
processes. C anadian Review of Sociology face-to-face interviewing vs. computer- capital in educational research: A critical
and Anthropology, 41, 125–146. mediated interviewing. International assessment. Theory and Society, 32, 185–196.
Atkinson, P. (2004). Life story interview. In M. Journal of Market Research, 43, 361–375. Lett, D., Hier, S.P., & Walby, K. (2010). CCTV
Lewis-Beck, A. Bryman, & T. Liao (Eds.), Davies, P. (2000). Doing interviews with surveillance and the civic conversation:
The Sage encyclopedia of social science re- female offenders. In V. Jupp, P. Davies, & A study in public sociology. Canadian
search methods. Thousand Oaks, CA: Sage. P. Francis (Eds.), Doing criminological Journal of Sociology, 35, 437–462.
Bampton, R., & Cowton, C.J. (2002). The research. London: Sage. Lewis, O. (1961). The children of Sánchez.
e-interview. Forum: Qualitative Social Deacon, D., Pickering, M., Golding, P., & New York: Vintage.
Research, 3. Retrieved from http://www Murdock, G. (1999). Researching com- Little, M. (2001). A litmus test for democracy:
.qualitative-research.net/index.php/fqs/ munications: A practical guide to methods The impact of Ontario welfare changes
article/view/848 in media and cultural analysis. London: for single mothers. Studies in Political
Barter, C., & Renold, E. (1999). The use of Arnold. Economy, 66, 9–36.
vignettes in qualitative research. Social Eady, A., Dobinson, C., & Ross, L. (2011). Lofland, J., & Lofland, L. (1995). Analyzing
Research Update, 25. Bisexual people’s experiences with social settings: A guide to qualitative ob-
Beardsworth, A., & Keil, T. (1992). The mental health services: A qualitative servation and analysis (3rd ed.). Belmont,
vegetarian option: Varieties, conversions, investigation. Community Mental Health CA: Wadsworth.
motives and careers. Sociological Review, Journal, 47, 378–389. Macnaghten, P., & Jacobs, M. (1997). Public
40, 253–293. Frohlich, K., Potvin, L., Chabot, P., & Corin, identification with sustainable develop-
Becker, H., & Geer, B. (1957). Participant E. (2002). A theoretical and empirical ment: Investigating cultural barriers
observation and interviewing: A com- analysis of context: Neighbourhoods, to participation. Global Environmental
parison. Human Organization, 16, 28–32. smoking, and youth. Social Science and Change, 7, 5–24.
Bell, E. (2007). Separatism and quasi- Medicine, 54, 1401–1417. Madge, C., & O’Connor, H. (2002). On-line
separatism in Alberta. Prairie Forum, 32, Gubrium, J., & Holstein, J. (1997). The new with e-mums: Exploring the Internet as a
335–355. language of qualitative method. New medium of research. Area, 34, 92–102.
Bloor, M., Frankland, S., Thomas, M., & York: Oxford University Press. Madriz, M. (2000). Focus groups in feminist
Robson, K. (2001). Focus groups in social Hanna, Paul. (2012). Using Internet technolo- research. In N. Denzin & Y. Lincoln
research. London: Sage. gies (such as Skype) as a research medium: (Eds.), Handbook of qualitative research
Bryman, A. (1999). Global Disney. In P. A research note. Qualitative Research, 12, (2nd ed.). Thousand Oaks, CA: Sage.
Taylor & D. Slater (Eds.), The American 239–242. Malbon, B. (1999). Clubbing: Dancing,
century. Oxford: Blackwell. Hine, V. (2000). Virtual ethnography. London: ecstasy and vitality. London: Routledge.
Bryman, A., Stephens, M., & Campo, C. Sage. Mann, C., & Stewart, F. (2000). Internet
(1996). The importance of context: Quali- Holbrook, B., & Jackson, B. (1996). Shopping communication and qualitative research:
tative research and the study of leader- around: Focus group research in North A handbook for researching online.
ship. Leadership Quarterly, 7, 353–370. London. Area, 28, 136–142. London: Sage.
Canadian Press (2017, 20 August). Teachers Hughes, R. (1998). Considering the vignette Markham, A. (1998). Life online: Research-
lack confidence to talk about residential technique and its application to a study ing the real experience in virtual space.
schools, study says. CBC.ca. Retrieved of drug injecting and HIV risk and London and Walnut Creek, CA: AltaMira
from http://www.cbc.ca/news/canada/ safer behaviour. Sociology of Health and Press.
edmonton/residential-schools- Illness, 20, 381–400. Mason, J. (2002). Qualitative interviewing:
edmonton-1.4254947 Kelly, L., Burton, S., & Regan, L. (1994). Asking, listening, interpreting. In T. May
Carter, A., Greene, S., Nicholson, V., Researching women’s lives or studying (Ed.), Qualitative Research in Action.
O’Brien, N., Dahlby, J., de Pokomandy, women’s oppression? Reflections on London: Sage.
A., . . . Kaida, A., (2016). “It’s a very what constitutes feminist research. In McKeganey, N., & Barnard, M. (1996). Sex
isolating world”: The journey to HIV M. Maynard & J. Purvis (Eds.), Research- work on the streets. Buckingham, UK:
care for women living with HIV in ing women’s lives from a feminist perspec- Open University Press.
British Columbia, Canada. Gender, tive. London: Taylor & Francis. Millen, D. (1997). Some methodological
Place & Culture, 23, 941–954. Kitzinger, J. (1993). Understanding AIDS: and epistemological issues raised by
Cassano, R., & Dunlop, J., (2005). Partici- Researching audience perceptions of ac- doing feminist research on non-feminist
patory action research with South Asian quired immune deficiency syndrome. In J. women. Sociological Research Online, 2.
immigrant women: A Canadian example. Eldridge (Ed.), Getting the message: News, Retrieved from www.socresonline.org.
Critical Social Work, 6. truth and power. London: Routledge. uk/socresonline/2/3/3.html
Charmaz, K. (2002). Qualitative inter- Kitzinger, J. (1994). The methodology of focus Miller, R.L. (2000). Researching life stories
viewing and grounded theory analysis. groups: The importance of interaction and family histories. London: Sage.
References 421
Milne, E. (2016). “I have the worst fear of Sugiman, P. (2004). Memories of the intern- Atkinson, J.M., & Drew, P. (1979). Order in
teachers”: Moments of inclusion and ment: Narrating Japanese-Canadian court: The organization of verbal interaction
exclusion in family/school relationships women’s life stories. Canadian Journal of in judicial settings. London: Macmillan.
among Indigenous families in southern Sociology, 29, 359–388. Bell, E. (2007). Separatism and quasi-
Ontario. Canadian Review of Sociology, Sweet, C. (2001). Designing and conducting separatism in Alberta. Prairie Forum, 32,
53, 270–289. virtual focus groups. Qualitative Market 335–355.
Milne, E. (2017). Implementing Indigenous Research, 4, 130–135. Bell, E., Jansen, H., & Young, L. (2007).
education policy directives in Ontario Tastsoglou, E., & Miedema, B. (2003). Sustaining a dynasty in Alberta: The
public schools: Experiences, challenges Immigrant women and community de- 2004 provincial election. Canadian
and successful practices. International velopment in the Canadian Maritimes: Political Science Review, 1, 27–49.
Indigenous Policy Journal, 8, 1–20. Outsiders within? Canadian Journal of Bhuyan, R., Jeyapal, D., Ku, J., Sakamoto,
Miraftab, F. (2000). Sheltering refugees: The Sociology, 28, 203–234. I., & Chou, E. (2017). Branding “Can-
housing experience of refugees in Metro- Trow, M. (1957). Comment on “Partici- adian experience” in immigration policy:
politan Vancouver, Canada. Canadian pant Observation and Interviewing: A Nation building in a neoliberal era.
Journal of Urban Research, 9, 42–63. Comparison.” Human Organization, Journal of International Migration &
Morgan, D. (1998). Planning focus groups. 16, 33–35. Integration, 18, 47–62.
Thousand Oaks, CA: Sage. Tse, A. (1999). Conducting electronic focus Billig, M. (1991). Ideology and opinions: Stud-
Morgan, D. (2002). Focus group interviewing. group discussions among Chinese re- ies in rhetorical psychology. Cambridge:
In J. Gubrium & J. Holstein (Eds.), Hand- spondents. Journal of the Market Cambridge University Press.
book of interview research: Context and Research Society, 41, 407–415. Blaikie, A. (2001). Photographs in the cul-
method. Thousand Oaks, CA: Sage. Wachholz, S., & Miedema, B. (2000). Risk, tural account: Contested narratives and
Morgan, D., & Spanish, M. (1985). Social fear, harm: Immigrant women’s percep- collective memory in the Scottish islands.
interaction and the cognitive organiza- tions of the “policing solution” to woman Sociological Review, 49, 345–367.
tion of health-relevant behaviour. Sociol- abuse. Crime, Law, and Social Change, Blommaert, J., & Bulcaen, C. (2000). Critical
ogy of Health and Illness, 7, 401–422. 34, 301–317. Discourse Analysis. Annual Review of
O’Connor, H., & Madge, C. (2001). Wane, N. (2004). Black Canadian feminist Anthropology, 29, 447–466.
Cyber-mothers: Online synchronous thought: Tensions and possibilities. Bryman, A. (1995). Disney and his worlds.
interviewing using conferencing soft- Canadian Woman Studies, 23, 145–153. London: Routledge.
ware. Sociological Research Online, 5. Wilkinson, S. (1998). Focus groups in Buchanan, C. (2014). A more national rep-
Retrieved from www.socresonline.org feminist research: Power, interaction, and resentation of place in Canadian daily
.uk/5/4/o’connor.html. the co-production of meaning. Women’s newspapers. Canadian Geographer, 58,
O’Connor, H., & Madge, C. (2003). “Focus Studies International Forum, 21, 111–125. 517–530.
groups in cyberspace”: Using the Inter- Wilkinson, S. (1999a). Focus group method- Clayman, S., & Gill, V.T. (2004). Conversa-
net for qualitative research. Qualitative ology: A review. International Journal of tion analysis. In M. Hardy & A. Bryman
Market Research, 6, 133–143. Social Research Methodology, 1, 181–203. (Eds.), Handbook of data analysis.
Parker, M. (2000). Organizational culture Wilkinson, S. (1999b). Focus groups: A fem- London: Sage.
and identity. London: Sage. inist method. Psych. of Women Quarterly, Clayton, R., Nagurney, A., & Smith, J. (2013).
Poland, B.D. (1995). Transcription quality as 23, 221–244. Cheating, breakup, and divorce: Is
an aspect of rigor in qualitative research. Wilson, B. (2002). The Canadian rave scene Facebook use to blame? Cyberpsychology,
Qualitative Inquiry, 1, 290–310. and five theses on youth resistance. Behaviour, and Social Networking, 16,
Rafaeli, A., Dutton, J., Harquail, C.V., & Canadian Journal of Sociology, 27, 717–720.
Mackie-Lewis, S. (1997). Navigating 373–412. Coté, J., & Allahar, A. (1994). Generation on
by attire: The use of dress by female Wright, R., Alaggia, R., & Krygsman, A. hold: Coming of age in the late twentieth
administrative employees. Academy of (2014). Five-year f ollow-up study of the century. Toronto: Stoddart.
Management Journal, 40, 9–45. qualitative experiences of youth in an Crook, C., & Light, P. (2002). Virtual society
Reinharz, S. (1992). Feminist methods in afterschool arts program in low-income and the cultural practice of study. In
social research. New York: Oxford communities. Journal of Social Science S. Woolgar (Ed.), Virtual society? Tech-
University Press. Research, 40, 137–146. nology, cyberbole, reality. Oxford: Oxford
Ristock, J. (2001). Decentring heterosexual- University Press.
ity: Responses of feminist counsellors to Dickinson, H. (1993). Accounting for Augus-
Chapter 12
abuse in lesbian relationships. Women tus Lamb: Theoretical and methodological
and Therapy, 23, 59–72. Abraham, J. (1994). Bias in science and med- issues in biography and historical sociol-
Schlesinger, P., Dobash, R.E., Dobash, R.P., ical knowledge: The Opren controversy. ogy. Sociology, 27, 121–132.
& Weaver, C. (1992). Women viewing Sociology, 28, 717–736. Disney Miller, D. (1956). The story of Walt
violence. London: British Film Institute. Altheide, D.L. (1996). Qualitative media Disney. New York: Dell.
Silverman, D. (1993). Interpreting qualitative analysis. Thousand Oaks, CA: Sage. Dunning, E., Murphy, P., & Williams, J.
data: Methods for analysing qualitative Altheide, D.L. (2004). Ethnographic content (1988). The roots of football hooliganism:
data. London: Sage. analysis. In M. Lewis-Beck, A. Bryman, & An historical and sociological study.
Smith, R. (2008). Pain in the act: The mean- T. Liao (Eds.), The Sage encyclopedia London: Routledge.
ings of pain among professional wrestlers. of social science research methods. Drouin, M., Tobin, E., & Wygant, K. (2014).
Qualitative Sociology, 31, 129–148. Thousand Oaks, CA: Sage. “Love the way you lie”: Sexting deception
422 References
in romantic relationships. Computers in Hatton, E., & Trautner, M. (2013). Images local-business/census-language-
Human Behaviour, 35, 542–547. of powerful women in the age of “choice households-marital-status
Fenton, N., Bryman, A., & Deacon, D. (1998). feminism.” Journal of Gender Studies, Lynch, M., & Bogen, D. (1997). Sociology’s
Mediating social science. London: Sage. 22, 65–78. asociological “core”: An examination of
Finkel, E., Eastwick, P., Karney, B., Reis, Heath, C. (1997). The analysis of activities textbook sociology in the light of the soci-
H., & Sprecher, S. (2012). Online dating: in face to face interaction using video. ology of scientific knowledge. American
A critical analysis from the perspective In D. Silverman (Ed.), Qualitative re- Sociological Review, 62, 481–493.
of psychological science. Psychological search: Theory, method and practice. Marcic, D. (2002). Respect: Women and
Science in the Public I nterest, 13, 3–66. London: Sage. popular music. New York: Texere.
Forster, N. (1994). The analysis of company Heritage, J. (1984). Garfinkel and ethnometh- McGuigan, J. (1992). Cultural populism.
documentation. In C. Cassell & G. odology. Cambridge, UK: Polity. London: Routledge.
Symon (Eds.), Qualitative methods in Heritage, J. (1987). Ethnomethodology. In Miller, N., & Morgan, D. (1993). Called to
organizational research. London: Sage. A. Giddens & J. Turner (Eds.), Social account: The CV as an autobiographical
Garfinkel, H. (1967). Studies in ethnometh- theory today. Cambridge, UK: Polity. practice. Sociology, 27, 133–143.
odology. Englewood Cliffs, NJ: Hier, S. (2000). The contemporary structure of Morton, K. (2016). Hitchhiking and missing
Prentice-Hall. Canadian racial supremacism; Networks, and murdered Indigenous women: A
Gill, R. (1996). Discourse analysis: Practical strategies, and new technologies. Can- critical discourse analysis of billboards
implementation. In J. Richardson (Ed.), adian Journal of Sociology, 25, 471–494. on the Highway of Tears. Canadian
Handbook of qualitative research meth- Hier, S. (2002). Raves, risks and the ecstasy Journal of Sociology, 41, 299–326.
ods for psychology and the social sciences. panic: A case study in the subversive Nettleton, S., Pleace, N., Burrows, R., Muncer,
Leicester: BPS Books. nature of moral regulation. Canadian S., & Loader, B. (2002). The reality of vir-
Gill, R. (2000). Discourse analysis. In M. Journal of Sociology, 27, 33–57. tual social support. In S. Woolgar (Ed.),
Bauer & G. Gaskell (Eds.), Qualitative Hirsch, J. (1981). Family photographs. New Virtual society? Technology, cyberbole,
researching with text, image and sound. York: Oxford University Press. reality. Oxford: Oxford University Press.
London: Sage. Hitsch, G., Hortaçscu, A., & Ariely, D. Olukoju, S. (2017, 17 July). Refugees settling
Giulianotti, R. (1997). Enlightening the North: (2010). Matching and scoring in online well in Manitoba. Winnipeg Free Press.
Aberdeen fanzines and local football dating. American Economic Review, 100, Retrieved from https://www.winnipeg
identity. In G. Armstrong & R. Giulianotti 130–163. freepress .com/our-communities/
(Eds.), Entering the field: New perspectives Hodson, R. (1996). Dignity in the workplace souwester/correspondent/Refugees-
on world football. Oxford: Berg. under participative management. settling-well-in-Manitoba-434975983
Gottdiener, M. (1997). The theming of American Sociological Review, 61, 719–738. .html
America: Dreams, visions and commer- Hutchby, I., & Wooffitt, R. (1998). Conversa- Parnaby, P. (2003). Disaster through dirty
cial spaces. Boulder, CO: Westview Press. tion analysis. Cambridge, UK: Polity. windshields: Law, order and Toronto’s
Gubrium, J., & Holstein, J. (1997). The new Jagger, E. (1998). Marketing the self, buying squeegee kids. Canadian Journal of
language of qualitative method. New an other: Dating in a post modern, con- Sociology, 28, 281–307.
York: Oxford University Press. sumer society. Sociology, 32, 795–814. Pauwels, L. (2008). A private visual prac-
Hales, J. (2006). An anti-colonial critique Jagger, E. (2001). Marketing molly and tice going public? Social functions and
of research methodology. In G. Dei & melville: Dating in a postmodern, con- sociological research opportunities of
A. Kempf (Eds.), Anti-colonialism and sumer society. Sociology, 35, 39–57. web-based family photography. Visual
education: The politics of resistance. Rot- John, I.D. (1992). Statistics as rhetoric in Studies, 23, 34–49.
terdam: Sense. psychology. Australian Psychologist, 27, Phillips, N., & Brown, J. (1993). Analyzing
Hall, J., Park, N., Song, H., & Cody, M. 144–149. communications in and around organiz-
(2010). Strategic misrepresentation in Jones, K. (2000). Constructing rBST in ations: A critical hermeneutic approach.
online dating: The effects of gender, self- Canada: Biotechnology, instability, and Academy of Management Journal, 36,
monitoring, and personality traits. Jour- the management of nature. Canadian 1547–1576.
nal of Social and Personal Relationships, Journal of Sociology, 25, 311–341. Potter, J. (1996). Representing reality: Dis-
27, 117–135. Kempf, A. (2006). Anti-colonial historiog- course, rhetoric and social construction.
Hallgrimsdottir, H., Phillips, R., & Benoit, C. raphy: Interrogating colonial education. London: Sage.
(2006). Fallen women and rescued girls: In G. Dei & A. Kempf (Eds.), Anti- Potter, J. (1997). Discourse analysis as a way
Social stigma and media narratives of the colonialism and education: The politics of analysing naturally occurring talk. In
sex industry in Victoria, B.C., from 1980 of resistance. Rotterdam: Sense. D. Silverman (Ed.), Qualitative research:
to 2005. Canadian Review of Sociology Kennedy, R. (2006). Researching the inter- Theory, method and practice. London: Sage.
and Anthropology, 43, 265–280. section between collective identity and Potter, J. (2004). Discourse analysis. In
Harding, R. (2006). Historical representa- conceptions of post-separation and di- M. Hardy & A. Bryman (Eds.), Handbook
tions of Aboriginal people in the Can- vorced fatherhood: A case study fathers for of data analysis. London: Sage.
adian news media. Discourse & Society, justice, fathers for just us, or fathers are us. Potter, J., Wetherell, M., & Chitty, A. (1991).
17, 205–235. Qualitative Sociology Review, 2, 75–97. Quantification rhetoric—Cancer on tele-
Harding, R. (2010). The demonization of Lee-Young, J. (2017, 2 August). Census 2016: vision. Discourse and Society, 2, 333–365.
Aboriginal child welfare authorities in A rise and changes in one-person house- Potter, J., & Hepburn, A. (2004). The analysis
the news. Canadian Journal of Com- holds. Vancouver Sun. Retrieved from of NSPCC call openings. In S. Becker &
munication, 35, 85–108. http://vancouversun.com/business/ A. Bryman (Eds.), Understanding research
References 423
for social policy and practice: Themes, (Eds.), Discourse analytic research: Coffey, A., Holbrook, B., & Atkinson, P. (1996).
methods, and approaches. Bristol: Policy Readings and repertoires of text. London: Qualitative data analysis: Technologies
Press. Routledge. and representations. Sociological Research
Potter, J., & Wetherell, M. (1994). Analyzing Online, 2. Retrieved from www.socres
discourse. In A. Bryman & R. Burgess online.org.uk/socresonline/1/1/4.html
Chapter 13
(Eds.), Analyzing qualitative data. Glaser, B., & Strauss, A. (1967). The discovery
London: Routledge. Ashbourne, L., & Baobaid, M, (2014). Parent- of grounded theory: Strategies for qualita-
Psathas, G. (1995). Conversation analysis: The adolescent storytelling in Canadian- tive research. Chicago: Aldine.
study of talk-in-interaction. Thousand Arabic immigrant families (Part 1): A Hesse-Biber, S. (1995). Unleashing Franken-
Oaks, CA: Sage. grounded theory. The Qualitative Report, stein’s monster? The use of computers in
Reed, M. (2000). The limits of discourse an- 19, 1–21. qualitative research. Studies in Qualita-
alysis in organizational analysis. Organiz- Atkinson, M. (2002). Pretty in ink: Conform- tive Methodology, 5, 25–41.
ation, 7, 524–530. ity, resistance and negotiation in women’s Locke, K. (1996). Rewriting The discovery of
Riches, G., & Dawson, P. (1998). Lost children, tattooing. Sex Roles, 47, 219–235. grounded theory after 25 years? Journal
living memories: The role of photographs Atkinson, M. (2004). Tattooing and civilizing of Management Inquiry, 5, 239–245.
in processes of grief and adjustment processes. C anadian Review of Sociology Lofland, J., & Lofland, L. (1995). Analyzing
among bereaved parents. Death Studies, and Anthropology, 41, 125–146. social settings: A guide to qualitative ob-
22, 121–140. Bryman, A., & Burgess, R. (1994). Reflections on servation and analysis (3rd ed.).
Schegloff, E. (1997). Whose text? Whose con- qualitative data analysis. In A. Bryman & Belmont, CA: Wadsworth.
text? Discourse and Society, 8, 165–187. R. Burgess (Eds.), Analyzing qualitative Lonkila, M. (1995). Grounded theory as an
Scott, J. (1990). A matter of record. Cam- data. London: Routledge. emergent paradigm for computer-
bridge, UK: Polity. Bryman, A., Gillingwater, D., & McGuinness, assisted qualitative data analysis. In U.
Seale, C. (2002). Cancer heroics: A study I. (1996). Industry culture and strategic Kelle (Ed.), Computer-aided qualitative
of news reports with particular reference response: The case of the British bus data analysis. London: Sage.
to gender. Sociology, 36, 107–126. industry. Studies in Cultures, Organiza- Miles, M.B. (1979). Qualitative data as an
Silverman, D. (1984). Going private: Cere- tions and Societies, 2, 191–208. attractive nuisance. Administrative Sci-
monial forms in a private oncology Bryman, A., Stephens, M., & Campo, C. ence Quarterly, 24, 590–601.
clinic. Sociology, 18, 191–204. (1996). The importance of context: Quali- Miller, R.L. (2000). Researching life stories
Silverman, D. (1993). Interpreting qualitative tative research and the study of leader- and family histories. London: Sage.
data: Methods for analysing qualitative ship. Leadership Quarterly, 7, 353–370. Okely, J. (1994). Thinking through fieldwork.
data. London: Sage. Bury, M. (2001). Illness narratives: Fact or In A. Bryman & R. Burgess (Eds.), Analyz-
Silverman, D. (1994). Analysing naturally oc- fiction? Sociology of Health and Illness, ing qualitative data. London: Routledge.
curring data on AIDS counselling: Some 23, 263–285. Pidgeon, N., & Henwood, K. (2004). Grounded
methodological and practical issues. In Catterall, M., & Maclaran, P. (1997). Focus theory. In M. Hardy & A. Bryman (Eds.),
M. Boulton (Ed.), Challenge and innov- group data and qualitative analysis pro- Handbook of data analysis. London: Sage.
ation: Methodological advances in social grams: Coding the moving picture as Riessman, C.K. (1993). Narrative analysis.
research on HIV/AIDS. London: Taylor & well as snapshots. Sociological Research Newbury Park, CA: Sage.
Francis. Online, 2. Retrieved from www.socres Riessman, C.K. (2004a). Narrative inter-
Sturgeon, R., & Morrissette, P. (2010). A online.org.uk/socresonline/2/1/6.html viewing. In M. Lewis-Beck, A. Bryman, &
qualitative analysis of suicide ideation Charmaz, K. (1983). The grounded theory T. Liao (Eds.), The Sage encyclopedia of
among Manitoba farmers. Canadian method: An explication and interpreta- social science research methods. Thousand
Journal of Counselling, 44, 191–207. tion. In R. Emerson (Ed.), Contemporary Oaks, CA: Sage.
Sugiman, P. (2004). Memories of the intern- field research: A collection of readings. Riessman, C.K. (2004b). Narrative analysis.
ment: Narrating Japanese-Canadian Boston: Little, Brown. In M. Lewis-Beck, A. Bryman, & T. Liao
women’s life stories. Canadian Journal of Charmaz, K. (1997). Identity dilemmas of (Eds.), The Sage encyclopedia of social
Sociology, 29, 359–388. chronically ill men. In A. Strauss & science research methods. Thousand
Sutton, R.I. (1992). Feelings about a Disney- J. Corbin (Eds.), Grounded theory in Oaks, CA: Sage.
land visit: Photography and the recon- practice. Thousand Oaks, CA: Sage. Roberts, B. (2002). Biographical research.
struction of bygone emotions. Journal of Charmaz, K. (2000). Grounded theory: Buckingham, UK: Open University
Management Inquiry, 1, 278–287. Objectivist and constructivist methods. Press.
Thompson, T., & Zerbinos, E. (1995). Gender In N. Denzin & Y. Lincoln (Eds.), Hand- Sosteric, M. (2017, 15 September). Suicide
roles in animated cartoons: Has the pic- book of qualitative research (2nd ed.). often has roots in childhood abuse.
ture changed in 20 years? Sex Roles, 32, Thousand Oaks, CA: Sage. Winnipeg Free Press. Retrieved from
651–673. Charmaz, K. (2004). Grounded theory. In M. https://www.winnipegfreepress.com/
Wetherell, M. (1998). Positioning and inter- Lewis-Beck, A. Bryman, & T. Liao (Eds.), opinion/analysis/suicide-often-has-roots-
pretative repertoires: Conversation an- The Sage encyclopedia of social science re- in-childhood-abuse-444592623.html
alysis and post-structuralism in dialogue. search methods. Thousand Oaks, CA: Sage. Stanley, L., & Temple, B. (1995). Doing the
Discourse and Society, 9, 387–412. Coffey, A., & Atkinson, P. (1996). Making business? Evaluating software packages
Widdicombe, S. (1993). Autobiography and sense of qualitative data: Complementary to aid the analysis of qualitative data sets.
change: Rhetoric and authenticity of research strategies. Thousand Oaks, CA: Studies in Qualitative Methodology, 5,
“gothic” style. In E. Burman & I. Parker Sage. 169–197.
424 References
Strauss, A. (1987). Qualitative analysis for In N. Denzin & Y. Lincoln (Eds.), Hand- Hodson, R. (1999). Analyzing documentary
social scientists. New York: Cambridge book of Qualitative Research (2nd ed.). accounts. Thousand Oaks, CA: Sage.
University Press. Thousand Oaks, CA: Sage. Hughes, K., MacKintosh, A.M., Hastings, G.,
Strauss, A., & Corbin, J. (1998). Basics of Deacon, D., Bryman, A., & Fenton, N. (1998). Wheeler, C., Watson, J., & Inglis, J. (1997).
qualitative research: Techniques and Collision or collusion? A discussion of Young people, alcohol, and designer drinks:
procedures for developing grounded the unplanned triangulation of quanti- A quantitative and qualitative study. British
theory. Thousand Oaks, CA: Sage. tative and qualitative research methods. Medical Journal, 314, 414–418.
Strauss, A., & Corbin, J. (1990). Basics of quali- International Journal of Social Research Jamieson, J. (2000). Negotiating danger in
tative research: Grounded theory procedures Methodology, 1, 47–63. fieldwork on crime: A researcher’s tale.
and techniques. Newbury Park, CA: Sage. Denzin, N. (1970). The research act in sociol- In G. Lee-Treweek & S. Linkogle (Eds.),
Tastsoglou, E., & Miedema, B. (2003). ogy. Chicago: Aldine. Danger in the field: Risk and ethics in
Immigrant women and community de- Fenton, N., Bryman, A., & Deacon, D. (1998). social research. London: Routledge.
velopment in the Canadian Maritimes: Mediating social science. London: Sage. Johnson, J., Bottorff, J., Moffat, B., Ratner,
Outsiders within? Canadian Journal of Festinger, L., Riecken, H., & Schachter, S. P., Shoveller, J., & Lovato, C. (2003).
Sociology, 28, 203–234. (1956). When prophecy fails. New York: Tobacco dependence: Adolescents’ per-
Weaver, A., & Atkinson, P. (1995). Micro- Harper & Row. spectives on the need to smoke. Social
computing and qualitative data analysis. Finch, J. (1985). “It’s great to have someone Science and Medicine, 56, 1481–1492.
Aldershot, UK: Avebury. to talk to”: The ethics and politics of Kuhn, T. (1970). The structure of scientific
Whitehead, K. (2010). “Hunger hurts but interviewing women. In C. Bell & H. revolutions (2nd ed.). Chicago: University
starving works”: A case study of gendered Roberts (Eds.), Social researching: of Chicago Press.
practices in the online pro-eating- Politics, problems, practice. London: Lantz, P., & Booth, K. (1998). The social con-
disorder community. Canadian Journal Routledge & Kegan Paul. struction of the breast cancer epidemic.
of Sociology, 35, 595–626. Gabriel, Y. (1998). The use of stories. In Social Science and Medicine, 46, 907–918.
Whitehead, P., & Carpenter, D. (1999). Ex- G. Symon & C. Cassell (Eds.), Qualitative MacKinnon, N., & Luke, A. (2002). Changes
plaining unsafe sexual behaviour: Cultural methods and analysis in organizational in identity attitudes as reflections of
definitions and health in the military. research. London: Sage. social and cultural change. Canadian
Culture, Health, and Sexuality, 1, 303–315. Gephart, R. (1988). Ethnostatistics: Qualita- Journal of Sociology, 27, 299–338.
Whiting, J., Parker, T., & Houghtaling, A. tive foundations for quantitative research. Mason, J. (1994). Linking qualitative and
(2014). Explanations of a violent relation- Newbury Park, CA: Sage. quantitative data analysis. In A. Bryman
ship: The male perpetrator’s perspective. Gilbert, G., & Mulkay, M. (1984). Opening & R. Burgess (Eds.), Analyzing qualita-
Journal of Family Violence, 29, 277–286. Pandora’s box: A sociological analysis tive data. London: Routledge.
of scientists’ discourse. Cambridge: Milkman, R. (1997). Farewell to the factory:
Cambridge University Press. Auto workers in the late twentieth century.
Chapter 14 Glaser, B. (1992). Basics of grounded theory Los Angeles: University of California Press.
Barnard, M., & Frischer, M. (1995). Com- analysis. Mill Valley, CA: Sociology Press. Miller, R.L. (2000). Researching life stories
bining quantitative and qualitative Glock, C. (1988). Reflections on doing survey and family histories. London: Sage.
approaches: Researching HIV-related research. In H. O’Gorman (Ed.), Survey- Morgan, D. (1998). Practical strategies for
risk behaviours among drug injectors. ing social life. Middletown, CT: Wesleyan combining qualitative and quantitative
Addiction Research, 2, 351–362. University Press. methods: applications for health research.
Barron, I. (2013). The potential and challen- Gorski, P. (2013). What is critical realism? Qualitative Health Research, 8, 362–376.
ges of critical realist ethnography. Inter- And why should you care? Contemporary Nowicki, E., Brown, J., & Stepien, M. (2014).
national Journal of Research and Method Sociology, 42, 658–670. Children’s structured conceptualizations
in Education, 36, 117–130. Hammersley, M. (1996). The relationship of their beliefs on the causes of learning
Bell, E. (2007). Separatism and quasi- between qualitative and quantitative difficulties. Journal of Mixed Methods
separatism in Alberta. Prairie Forum, 32, research: Paradigm loyalty versus meth- Research, 8, 69–82.
335–355. odological eclecticism. In J. Richardson Phelan, P. (1987). Comparability of qualita-
Bhaskar, R. (1975). A realist theory of science. (Ed.), Handbook of research methods tive and quantitative methods: Studying
London: Verso. for psychology and the social sciences. child sexual abuse in America. Education
Bhaskar, R. (1979). The possibility of natural- Leicester, UK: BPS Books. and Urban Society, 20, 35–41.
ism: A philosophical critique of the contem- Hansen, D. (2017, 25 August). Gender stero- Phoenix, A. (1994). Practising feminist
porary human science. Atlantic Highlands, typing of work roles hurts both women research: The intersection of gender and
NJ: Humanities Press. and men. The Globe and Mail. Retrieved “race” in the research process. In M.
Bhaskar, R. (1986). Scientific realism and from https://beta.theglobeand mail.com/ Maynard & J. Purvis (Eds.), Researching
human emancipation. Abingdon, UK: report-on-business/careers/business- women’s lives from a feminist perspective.
Routledge. education/gender-stereotyping-of- London: Taylor & Francis.
Bloor, M. (1997). Addressing social problems work-roles-hurts-both-women-and-men/ Platt, J. (1986). Functionalism and the survey:
through qualitative research. In D. Silver- article36091414/?ref=http://www The relation of theory and method. Socio-
man (Ed.), Qualitative research: Theory, .theglobeandmail.com& logical Review, 34, 501–536.
method and practice. London: Sage. Hodson, R. (1996). Dignity in the workplace Platt, J. (1996). A history of sociological re-
Charmaz, K. (2000). Grounded theory: under participative management. American search methods in America 1920–1960.
Objectivist and constructivist methods. Sociological Review, 61, 719–738. Cambridge: Cambridge University Press.
References 425
Porter, S. (1993). Critical realist ethnography: motives and careers. Sociological Review, Wolcott, H. (1990). Writing up qualitative
The case of racism and professionalism in 40, 253–293. research. Newbury Park, CA: Sage.
a medical setting. Sociology, 27, 591–609. Bryman, A. (1998). Quantitative and quali-
Porter, S. (2002). Critical realist ethnography. tative research strategies in knowing the
Chapter 16
In T. May (Ed.), Qualitative research in social world. In T. May & M. Williams
action. London: Sage. (Eds.), Knowing the social world. Buck- Barkham, P., & Jenkins, R. (2002, 13
Robinson, J., Shaver, P., & Wrightsman, L. ingham, UK: Open University Press. December). Fears for fresher who van-
(1999). Measures of political attitudes. Denzin, N. (1994). Evaluating qualitative ished on mission to talk to the homeless.
Toronto: Academic Press. research in the poststructural moment: The Times [London].
Ross, L., O’Gorman, L., MacLeod, M., Bauer, The lessons James Joyce teaches us. Inter- Beardsworth, A., & Keil, T. (1992). The vege-
G., MacKay, J., & Robinson, M. (2016). national Journal of Qualitative Studies in tarian option: Varieties, conversions,
Bisexuality, poverty and mental health: A Education, 7, 295–308. motives and careers. Sociological Review,
mixed methods analysis. Social Science & Dwyer, J., Mayer, L., Dowd, K., Kandel, R., & 40, 253–293.
Medicine, 156, 64–72. Mayer, J. (1974). The new vegetarians: The Becker, H. (1986). Writing for social scien-
Schrøder, K.C. (1999). The best of both worlds? natural high? Journal of the American tists: How to start and finish your thesis,
Media audience research between rival Dietetic Association, 65, 529–536. book, or article. Chicago: University of
paradigms. In P. Alasuutari (Ed.), Rethink- Giulianotti, R. (1995). Participant observa- Chicago Press.
ing the media audience. London: Sage. tion and research into football hooligan- Craig, G. (2004). Managing safety in policy
Silverman, D. (1985). Qualitative methodology ism: Reflections on the problems of entrée research. In S. Becker & A. Bryman
and sociology: Describing the social world. and everyday risks. Sociology of Sport (Eds.), Understanding research for
Aldershot, UK: Gower. Journal, 12, 1–20. social policy and practice: Themes, meth-
Strauss, A., & Corbin, J. (1998). Basics of Harding, R. (2006). Historical representations ods, and approaches. Bristol: Policy Press.
qualitative research: Techniques and of Aboriginal people in the Canadian news Eichler, M., & Lapointe, J. (1985). On the
procedures for developing grounded media. Discourse & Society, 17, 205–235. treatment of the sexes in research.
theory. Thousand Oaks, CA: Sage. Harding, R. (2010). The demonization of Ottawa: Social Sciences and Humanities
Tastsoglou, E., & Miedema, B. (2003). Aboriginal child welfare authorities in Research Council of Canada.
Immigrant women and community de- the news. Canadian Journal of Communi- Festinger, L., Riecken, H., & Schachter, S.
velopment in the Canadian Maritimes: cation, 35, 85–108. (1956). When prophecy fails. New York:
Outsiders within? Canadian Journal of Kelley, J., & De Graaf, N. (1997). National Harper & Row.
Sociology, 28, 203–234. context, parental socialization, and re- Golden-Biddle, K. (1997). Composing quali-
Wajcman, J., & Martin, B. (2002). Narratives ligious belief: Results from 15 Nations. tative research. Thousand Oaks, CA:
of identity in modern management. American Sociological Review, 62, 639–59. Sage.
Sociology, 36, 985–1002. Lincoln, Y., & Denzin, N. (1994). The fifth Golden-Biddle, K., & Locke, K. (1993).
Walklate, S. (2000). Researching victims. moment. In N. Denzin & Y. Lincoln Appealing work: An investigation of
In R. King & E. Wincup (Eds.), Doing (Eds.), Handbook of qualitative research. how ethnographic texts convince.
research on crime and justice. Oxford: Thousand Oaks, CA: Sage. Organization Science, 4, 595–616.
Oxford University Press. Manning, P. (1995). The challenge of post- Kelley, J., & De Graaf, N. (1997). National con-
Webb, E.J., Campbell, D.T., Schwartz, R.D., modernism. In J. Van Maanen (Ed.), text, parental socialization, and religious
& Sechrest, L. (1966). Unobtrusive meas- Representation in ethnography. Thousand belief: Results from 15 nations. American
ures: Non-reactive measures in the social Oaks, CA: Sage. Sociological Review, 62, 639–659.
sciences. Chicago: Rand McNally. Richardson, L. (1994). Writing: A method Lankshear, G. (2000). Bacteria and babies:
Wilson, B. (2002). The Canadian rave scene of inquiry. In N. Denzin & Y. Lincoln A personal reflection on researcher’s
and five theses on youth resistance. Can- (Eds.), Handbook of qualitative research. risk in a hospital. In G. Lee-Treweek &
adian Journal of Sociology, 27, 373–412. Thousand Oaks, CA: Sage. S. Linkogle (Eds.), Danger in the field:
Wilson, E.O. (1998). Consilience: The unity of Rosenau, P.M. (1992). Post-modernism and the Risk and ethics in social research.
knowledge. New York: Knopf. social sciences: Insights, inroads, and intru- London: Routledge.
sions. Princeton: Princeton University Press. Lofland, J., & Lofland, L. (1995). Analyzing
Stoller, P. (1989). The taste of ethnographic social settings: A guide to qualitative
Chapter 15 things. Philadelphia: University of observation and analysis (3rd ed.).
Armstrong, G. (1993). Like that Desmond Pennsylvania Press. Belmont, CA: Wadsworth.
Morris? In D. Hobbs and T. May (Eds.), Taylor, A. (1993). Women drug users: An Marx, G.T. (1997). Of methods and man-
Interpreting the Field: Accounts of ethnography of an injecting community. ners for aspiring sociologists: 37 moral
Ethnography. Oxford: Clarendon Press. Oxford: Clarendon Press. imperatives. American Sociologist, 28,
Atkinson, P., & Coffey, A. (1995). Realism and Van Maanen, J. (1988). Tales of the field: On 102–125.
its discontents: On the crisis of cultural writing ethnography. Chicago: University Small, W., Maher, L., & Kerr, T. (2014).
representation in ethnographic texts. In of Chicago Press. Institutional ethical review and ethno-
B. Adam & S. Allan (Eds.), Theorizing van Stapele, N. (2014). Intersubjectivity, graphic research involving injection
culture: An interdisciplinary critique after self-reflexivity and agency: Narrating drug users: A case study. Social Science &
postmodernism. London: UCL Press. about “self” and “other” in feminist Medicine, 104, 157–162.
Beardsworth, A., & Keil, T. (1992). The vege- research. Women’s Studies International Wolcott, H. (1990). Writing up qualitative
tarian option: Varieties, conversions, Forum, 43, 13–21. research. Newbury Park, CA: Sage.
Index
Note: Page numbers in italics indicate illustrations.
Abraham, J., 277–8 Barron, I., 327
abstract, 347, 348, 367 Beagan, B., 21, 65, 159
access: ethnography and, 236; to research subjects, 20–1 Beardsworth, A.: et al., 126; and T. Keil, 41, 249, 265, 350–2, 368
accuracy, 137; documents and, 273, 279 Becker, H., 160, 161, 368; and B. Geer, 264, 266
acknowledgments, 367 behaviour: quantitative/qualitative distinction and, 329; quanti-
activism, 13 tative research and, 213; questions on, 98, 101, 105; sampling
adjacency pairs, 293, 403 and, 162; stated v. actual, 90, 134
Adriaenssens, C. and L. Cadman, 259 Behaviour Genetics Association, 49
Adshade, Marina, 291 Belk, R. et al., 206
advisors, 362–3 Bell, E., 3, 135, 164, 218, 224, 226, 242, 287, 336; et al., 40, 44, 78,
Ajodhia-Andrews, A., 250 109, 116, 207, 277; and C. Kandler, 126
Alberta Survey, 79, 111 Beres, M.A. et al., 114
Altheide, D.L., 287–8 Berthoud, R., 82
Amazon’s Mechanical Turk, 116, 167 beta weights, 191
American Sociological Review, 347 Bhuyan, R. et al., 279
analysis: bivariate, 170, 180–5, 403; cluster, 78; conversation, bias, 18; documents and, 281; non-response, 156; qualitative
199, 271, 290–4, 299–300, 301, 302, 404; critical discourse, research and, 211; questions and, 97; sampling, 111, 147–9;
12, 199, 271, 280, 298–9, 300, 302, 404; factor, 78, 405; inter- social desirability, 105, 134, 139, 408
actional, 318; multivariate, 127, 170, 189–93, 406; mutual, bibliographies, 365; websites and, 356
331–2; narrative, 305, 318–20, 406; performative, 318; sec- Bibliography section, 347
ondary, 122–8, 130, 408; structural, 318; thematic, 318, 332; Bikos, Lesley, 9
unit of, 114, 127, 287; univariate, 170, 177–80, 409; see also biographical method, 273, 403; see also life history
content analysis; data analysis; discourse analysis Blaxter, M., 38
analytic induction, 306–7, 326, 403 Bloor, M., 205–6, 335
Anderson, K. et al., 116 Bloor, M. et al., 257–8
Andrews, K. et al., 119–20 Blumer, H., 11, 91, 201–3
anecdotalism, 317, 332 Boivin, R. and C. Leclerc, 128
anonymity, 57, 140 Bourdieu, P., 88
ANOVA , 191, 192, 193 Bowen, R., 29
anxiety: covert research and, 220 Boyd, Monica, 129
appendices, 370 Brannigan, A., 90
Armstrong, G., 207, 224, 227, 234, 266 Browne, A.J., 229
Ashbourne, L. and M. Baobaid, 315 Bryman, A., 161, 247–8, 278, 347; et al., 209, 266; and D. Cramer,
association, 39 79–80, 176; D. Gillingwater, and I. McGuiness, 313
Atkinson, M., 319 Buchanan, C., 164, 272
Atkinson, P., 218, 242; and A. Coffey, 353 Bullfighter’s Braid, The (photograph), 230
attitudes: quantitative study of, 327–8; questions on, 98–100 Bulmer, M., 77; et al., 207
auditing, 206 Bury, M., 320
authenticity, 213, 354; documents and, 273, 278, 279, 300, 301;
mass media and, 280; qualitative research and, 204; websites Cambridge-Summerville Youth Study, 63–4
and, 281 Campbell, M., 232, 236
autobiographies, 273–5 Canadian Election Study, 112, 123, 132, 158
authority, experiential, 355–6 Canadian Institutes of Health Research (CIHR), 50
average. See mean Canadian Research Institute for Social Policy (CRISP), 94
Canadian Sociological Association: Code of Ethics, 51, 68
Baer, D. et al., 42, 46 Cancer: Your Money or Your Life (film), 297–8
Bales, Robert, 135, 136 Cao, L., 193–4
Bampton, R. and C.J. Cowton, 113 CAQDAS . See computer-assisted qualitative data analysis soft-
Barker, Joshua, 227 ware (CAQDAS)
Barnard, M. and M. Frischer, 338 Carmichael, Jason, 141
Index 427
232–4; structure and, 209; types of, 218; virtual, 225; visual, Gabriel, Y., 332
229–31; writing and, 346, 353, 354–6; see also participant Galton, M. et al., 137
observation Gans, H.J., 56, 199, 224
ethnomethodology, 271, 290–1, 404 Garfinkel, H., 290
ethnostatistics, 331 gatekeepers, 20–1, 220, 221, 405
evaluation research, 36–7, 38–9, 404 Gaudet, S. et al., 163
exhaustive rule, 178, 285 Gazso-Windlej, A. and J. McMullin, 76, 82
experimental group, 31, 37, 38, 76, 404 Geertz, C., 206
experiments, 27, 29–37, 51, 329, 404; causality and, 86; classic, generalizability, 47, 146, 405; case studies and, 45–6; experi-
30–4, 31; ethics and, 30; field, 30, 141, 143, 405; informed ments and, 33; qualitative research and, 204, 211–12
consent and, 54; laboratory, 30, 34; logic of comparison, 37; generalization, 7, 74, 405; multi-strategy research and, 338;
manipulation and, 29–30, 31; natural, 35–6; notation in, 31; quantitative research and, 87, 213; sampling and, 154, 155,
qualitative, 47; quantitative, 47; quasi-, 34–7; random assign- 163, 185
ment, 31; significance of, 37; validity and, 29, 32, 33, 34 General Social Survey (GSS), 84, 91, 111, 124–5, 127, 164; 2015
explanations: idiographic, 29, 47; nomothetic, 28, 47; research Time-Use Study, 122, 132; Victimization Study, 132, 193
designs and, 28 genetics, behavioural, 36, 40
George, U. and F. Chaze, 82
Gephart, R., 331
Facebook, 225, 281
Gerson, K. and R. Horowitz, 222
facilitation: quantitative/qualitative distinction and, 334, 336–7
Giddens, A., 5, 15
facilitator, 250–1, 254–5
Gilbert, G. and M. Mulkay, 331
feedback, on writing, 366
Gill, R., 295, 296
feminism, 12, 20, 353; ethnography and 235–6; focus groups
Giulianotti, R., 221, 222, 224–5, 234, 290
and, 263–4; interviewing and, 262–4, 268; qualitative re-
Glaser, B. and A. Strauss, 234, 307
search and, 20, 24; quantitative research and, 341; survey
Glock, C., 329
research and, 128–9; values and, 18–20
Glucksmann, M., 218
Fenton, N. et al., 301, 336–7, 340
Goffman, E., 5
Festinger, L. et al., 45, 63, 326
Golden-Biddle, K. and K. Locke, 369
field notes, 216, 227–8, 236, 306, 405
Goode, E., 56
field studies, 133
Gottdiener, M., 288–9
Finch, J., 107, 338
Goyder, J. et al., 89, 112
findings, 72; grounded theory and, 311; postmodernism and, 353; Grabb, E. and J. Curtis, 126
quantitative research and, 76; research reports and, 201, 350 Grayson, J. Paul, 65
Findings section, 351–2 Green, A.I., 202
Fine, G., 224 grounded theory, 7, 201, 305, 306, 307–13, 320, 326, 405; coding
Fisher, R. and Y. Ma, 34, 35, 37 and, 307–9; criticisms of, 311–13; defining, 307; outcomes of,
flexibility, 209–10, 250, 264 309–11, 310; theory and, 310, 312–13
focus groups, 110, 199, 239, 240, 250–61, 267, 405; advantages of, group effects, 258
251; artificial v. natural, 330; conducting, 251; feminism and, Guba, E. and Y. Lincoln, 204
263–4; group interaction in, 251, 256–7, 258, 267; interviews Gubrium, J. and J. Holstein, 291
v., 258; limitations of, 257–8; moderators and, 250–1, 254–5,
257, 259, 267; number of, 251, 254; online, 258, 259, 260; Hales, J., 281–2
online v. face-to-face, 260–1; questions in, 254–5; recording Hallett, R. and K. Barber, 225
and transcription, 255, 258; research on, 252; selection of Hallgrimsdottir, H. et al., 14, 18, 207, 211, 279
participants in, 253–4; size of, 251–3, 259 Hammersley, M., 334
Foddy, W., 81 Hammersley, M. and P. Atkinson, 223
Foran, C., 44 Haney, C., C. Banks, and P. Zimbardo, 64–5
Forster, N., 279 Hanna, Paul, 259
Foster, J., 11, 207 Hansen, D., 333
Foucault, M., 294 Hanson, R.K. et al., 37
Fowler, F., 113, 155 Harding, R., 279, 298, 299, 301
Fox, Nick, 358 harm: ethics and, 51, 63–4, 65
fragmentation: coding and, 313, 316, 317–18 Hawthorne effect, 139
Frazer, R. and U. Wiersma, 90 Hay, D., 8, 80, 100
frequency tables, 177–8, 405; in SPSS , 380–1 Health Canada, 61
Frisby, W. et al., 13 Help, The (film), 160
Frohlich, K. et al., 252 Heritage, J., 292
F statistic, 189 hermeneutics, 271, 289–90, 301, 405
funding, research, 20 Hessler, R. et al., 54, 121, 224
430 Index
Hier, S., 17, 234, 280–1, 282, 290 in-depth, 243; interviewers, 110–11, 116, 244–5; introductory
Highway of Tears, 299 statement and, 114–15; longitudinal, 266; narrative, 242;
Hiller, H. and L. DiLuzio, 206–7 online, 112–13, 260–1; oral history, 240, 243, 407; probing,
Hine, V., 259 115–16; prompting, 116; qualitative research and, 239–67;
histograms, 179, 179; in SPSS , 384–5 quantitative v. qualitative, 240; rapport and, 114; retro-
Historica Canada, 100 spective, 265; semi-structured, 208, 239, 240, 241–3, 247,
history, 32; effects, 33 272, 253, 262, 325, 408; telephone, 111–12, 166–7, 259; un-
Hitsch, G. et al., 284 structured, 41, 199, 208, 239, 240, 241, 242–3, 262, 272, 409;
Hobbs, D., 226 see also qualitative interviews; structured interviews
Hochschild, A., 5, 203–4 interview schedule, 76, 108, 110, 113, 135, 171, 240, 405
Hodson, R., 331–2 Introduction section, 348, 350–1, 352, 368
Holbrook, B. and B. Jackson, 254 invalidity: convergent, 84; see also validity
Holbrook, A. et al., 112 iterative, 7, 72, 201, 306
Holtby, A. et al., 230
Homan, R., 54; and M. Bulmer, 59 Jagger, E., 282, 283, 301
Huey, L., 44 Jamieson, J., 337
Hughes, E., 44 Johnson, J. et al., 336
Hughes, G., 20, 21 Jones, K., 277
Hughes, K. et al., 335 journals: publication and, 21, 89, 347; searching articles in, 365;
Hughes, R., 247 social science, 26
Human Genome Project, 49 justice: ethics and, 50, 62–6
Humphreys, L., 59, 66, 224
hypotheses, 6, 186, 405; grounded theory and, 309; multi- Karabanow, J., 208, 211
strategy research and, 336; qualitative research and, 326, Kelley, J. and N. De Graaf, 125–6, 347–50, 368, 370
336; quantitative research and, 76 Kelly, L. et al., 262
Kendall’s tau-b, 184, 187, 406
IBM SPSS Statistics (SPSS), 177–8, 194, 196, 284, 316, 363, 373–91, Kennedy, E. et al., 78, 82, 83, 156
405; basic operations in, 374; data analysis with, 380–90; Kennedy, R., 287
getting started in, 373–80 Kerr, D., 42; and J. Michalski, 42
idiographic explanations, 29, 48, 405 key informants, 216, 222–3, 236, 406
illegality: ethics and, 57 keywords, 282, 284, 314, 363, 365
incentives, 34, 167 Kimmel, A., 63–4
incidents, 266, 310, 332; recording, 137 Kingsbury, M. and R. Coplan, 106
index, 405; see also scale Kitzinger, J., 251, 252, 254, 256
indexicality, 290 Knott, C., 44
indexing. See coding knowledge, xix, xx; consilience and, 324; creation of, 212
indicators, 77, 405; multiple, 78–9, 80, 85, 105, 406 Krueger, R., 258
induction, 2, 7, 7, 9, 24, 352, 405; case studies and, 44; positivism Kuhn, T., 334
and, 10; qualitative research and, 16, 198, 199; quantitative Kvale, S., 244–5, 248
research and, 76; research questions and, 22
informants, key, 216, 222–3, 236, 406 labour, emotional, 5, 203–4
information sheet, 53 Lamb, Augustus, 273, 274
instrumentation, 32 language: as constructive, 296; content analysis and, 290–300;
interaction, 190–1; focus groups and, 256–7, 267 conversation analysis and, 290; discourse analysis and, 290;
International Social Survey Programme (ISSP), 125–6 sexist, racist, and other prejudicial, 366, 368
Internet, 17, 259; content analysis and, 280–2; ethics and, 58, Lankshear, G., 367
188; focus groups and, 258–61; interviews and, 258–61; sam- Lantz, P. and K. Booth, 14, 328
pling and, 167; see also websites LaPiere, R.T., 90, 137, 141
interpretivism, 10–12, 198, 405; qualitative research and, 16, Laplante, B., 44
199; quantitative/qualitative distinction and, 324, 327–8 Lareau, A. and E. Weininger, 257
intersubjectivity, 10, 405 Laurenceau, J.-P. and N. Bolder, 121
Inter-University Consortium for Political and Social Research, laws, 28
123, 132 Lee-Treweek, G., 226
interval, confidence, 156, 157, 186; 95 per cent, 154 Leidner, R., 204, 219
interviewer effects, 117 Lett, D. et al., 252
interview guides, 239, 243–5, 247–8, 267, 405 letters; cover, 166; as data, 273–5; of information, 53, 53
interviews, 95; computer-assisted, 112; conducting, 109–16; Lewin, K., 13
contexts of, 109–13; email, 112–13; ethnography and, 240, Lewis, O., 44, 208, 242
264–6, 268; face-to-face, 111–12, 260–1; focus groups v., 258; library catalogue, 365
Index 431
life history, 44, 208, 318, 406; documents and, 273; interviews, measurement validity, 47, 75, 406; case studies and, 44; quantita-
240, 242–3, 406 tive research and, 83–5; structured observation and, 138–9
Likert, Rensis, 79 measures: of central tendency, 179, 406; of dispersion, 179–80,
Likert items/scale, 78, 79, 96, 110, 176, 406; format of, 79, 406; multiple-indicator, 78–9, 80, 85, 105, 406; unobtrusive,
98–100, 120, 121; question order and, 104 140, 409
Lincoln, Y. and N. Denzin, 353, 354 media: content analysis and, 272, 279–80, 282; sampling and,
Lincoln, Y. and E. Guba, 204, 206 164; see also social media
listening, 246 median, 179, 406; in SPSS , 385
literature review, 69–71; report writing and, 351, 366, 368–9 memos, 311, 313; analytic, 228; in NVivo, 401
literature search, 363–5 Menard, S., 44
Little, M., 263 Merton, R., 4
Lofland, J. and L. Lofland, 208, 228, 243, 249 methodology: as power, xx
longitudinal research, 27, 41–4, 47, 87, 406; qualitative inter- Methods and Models section, 349
viewing and, 266; secondary analysis and, 123–4 Mies, M., 18
Longitudinal Immigration Database (IMDB), 123 Miles, M.B., 306
lying, 34 Milgram, S., 3, 54–6, 65
Lynch, M. and D. Bogen, 287 Milkman, R., 337–8, 341
Lynd, R. and H. Lynd, 42 Millen, D., 262
Lyon, K. and H. Frohard-Dourlent, 3, 199–201 Miller, Diane Disney, 273
Miller, N. and D. Morgan, 288
McArdle, M., 91 Miller, R.L., 242, 320, 326–7
McCall, M.J., 137, 141 Milne, Emily, 255, 257
McDonough, P. and J. Polzer, 207 Miraftab, F., 252
McKee, L. and C. Bell, 208 mixed-methods research, 325; see also multi-strategy research
McKeganey, N. and M. Barnard, 5, 265 mode, 179, 406
Mackenzie King, William Lyon, 275 moderator, 250–1, 254–5, 257, 267, 406; of online focus
MacKinnon, N. and A. Luke, 159, 337 group, 259
Macnaghten, P. and M. Jacobs, 252, 254 Mojtehedzadeh, Sara, 219
Madriz, M., 258, 264 Moody, Geraldine, 276
Maillé, Marie-Ève, 66 Morgan, D., 251–3, 254, 258, 337, 337; and M. Spanish, 256
Mairs, K. and S. Bullock, 159 Morgan, R., 20
Malbon, B., 241 mortality, 32
Mangione, T.W., 166 Morton, K., 299
manipulation: cross-sectional designs and, 37, 40; experiments multiple linear regression, 191, 193–4; examples of results,
and, 29–30, 31, 32–3 192, 193
Mann, C. and F. Stewart, 112, 259 multi-strategy research, 324, 325, 332–4, 341, 343, 406; ap-
Manning, P., 353 proaches to, 334–41; arguments against, 333–4; different
Markham, A., 258–9 aspects of phenomena, 338–41; facilitation and, 334, 336–7;
Marx, G.T., 362 filling in gaps, 337; generality and, 338; micro/macro issues
Marx, Karl, 12, 101 and, 338; participants and, 337–8; planning, 337; research-
Marxism, 12 ers and, 337–8; static and process features, 337; triangulation
Mason, J., 204, 246 and, 334–6
Mason, J., 246, 338–9 multivariate analysis, 127, 170, 189–93, 406
maturation, 32 Murphy, L. and J. Fedoroff, 80–1
Mayhew, P., 104 mutual analysis, quantitative/qualitative, 331–2
Mead, George Herbert, 11
Mead, Margaret, 45 Nagra, B. and P. Maurutto, 3
mean: arithmetic, 385, 403; eta and, 185; population, 154; narrative analysis, 305, 318–20, 406
sample, 154, 156; in SPSS , 385, 388, 390; statistical signifi- narratives: illness, 320; meta-, 353; mini-, 353
cance and, 189 National Graduates Survey, 123
meaning: connotative, 288; denotative, 288; documents and, National Inquiry into Missing and Murdered Indigenous
273, 275, 278, 279, 300, 301; manifest/latent, 300; media and, Women and Girls, 198–9
280; qualitative research and, 213; quantitative/qualitative National Longitudinal Survey of Children and Youth (NLSCY),
distinction and, 329; quantitative research and, 327–8; semi- 42, 49, 123
otics and, 288 National Population Health Survey, 123
measurement: concepts and, 76–81; direct/indirect, 77–8; naturalism, 406; conversation analysis and, 291; ethnography
multiple-item, 78–9, 85; quantitative research and, 85, 212; and, 265; focus groups and, 251, 263; qualitative research
qualitative research and, 336 and, 199, 208; quantitative/qualitative distinction and, 330
Measurement section, 349 natural science model, 326–7
432 Index
Natural Sciences and Engineering Council of Canada (NSERC), 50 participant observation, 11, 29, 135, 199, 208, 216, 325, 407; arti-
Nemni, M. and M. Nemni, 44 ficial v. natural, 330; roles in, 223, 223–4; as term, 217
neo-positivism, 326 participants: of focus groups, 253–4, 259, 261; in qualitative
Nietzsche, F., 13 research, 212; in quantitative research, 75–6
nodes: in NVivo, 393–6, 398–9 participatory action research (PAR), 12, 13, 199, 210, 230, 407
nomothetic explanations, 28, 47, 406 Pauwels, L., 281
non-response, 123, 148, 406; bias, 156; reducing, 165–7; sample Pawson, R. and N. Tilley, 38–9
size and, 155–7 payment; ethics and, 65; focus groups and, 258
non-spuriousness, 28–9 Pearson’s r, 182, 184, 187, 407; in SPSS , 388
“no opinion,” 103, 112 periodicity, 152
Normand, S. et al., 138 Perrucci, R. et al., 134
Norris, C., 224 Pettigrew, A., 208
“not applicable,” 173, 178 Phelan, P., 336
note-taking: covert research and, 220; qualitative interviews Phillips, N. and J. Brown, 289–90
and, 245 Phinney, A. et al., 38
Nowicki, E. et al., 340 Phoenix, A., 330
null hypothesis, 186–7, 406 photographs, 275–7, 278; ethnography and, 229–31; field notes
NVivo, 305, 316, 322, 363, 391–402, 407; coding in, 393–8; get- and, 228; on websites, 281
ting started, 391–3; memos in, 401; opening in, 402; output, photovoice, 230, 231
399; saving in, 401–2; searching in, 398–9 Pidgeon, N. and K. Henwood, 309
pilot studies, 107–8, 130, 136, 159, 287, 366
Oakley, A., 18 Pink, S., 229–30
objectivism, 13–14, 406; quantitative/qualitative distinction Platt, J., 328
and, 325; quantitative research and, 16, 75, 92 Poland, B.D., 249
objectivity, 10, 212, 278, 279 police: statistics, 127–8, 285
observation research: non-participant, 135; systematic, 135; policies: evaluation research and, 38–9; quasi-experiments and,
systematic social, 142; types of, 135; unobtrusive, 135, 409; 36, 37
unstructured, 86, 135, 143; see also participant observation; politics: research and, 20–1, 24
structured observation polysemy, 288, 289
observation schedules, 133, 135, 136–7, 143, 171, 285, 406 population, 146, 148, 407; generalizing to, 154, 185; heterogen-
observer-as-participant, 224, 233 eity of, 157; large, 153, 167, 213; mean, 154; stratification of,
O’Connell Davidson, J. and D. Layder, 60 152, 154, 161
O’Connor, H. and C. Madge, 260 Porter, S., 327
Olivieri, Nancy, 61 positivism, 9–10, 11, 12, 407; conversation analysis and, 291;
omnipotence, interpretive, 355, 356 covert, 326; neo-, 326; quantitative/qualitative distinction
ontology, 2, 13–15, 24, 291, 324, 406; constructionist position, and, 326; quantitative research and, 16, 75, 76
13–14; objectivist position, 13–14; qualitative research and, post-coding, 80
16, 17; quantitative/qualitative distinction and, 324, 328–9; postmodernism, 346, 353–4, 356, 407
quantitative research and, 16, 17; “soft constructionist” pos- post-testing, 31, 32, 76
ition, 13–14 Potter, J., 15, 294, 295; et al., 297–8; and A. Hepburn, 295; and
operationalization, 76, 406–7 M. Wetherell, 296
opinion polls, 161 Pratt, A. and M. Valverde, 18, 44, 45–6
ORACLE (Observational Research and Classroom Learning praxis, 12
Evaluation), 137 preference organization, 293–4
oral history interviews, 240, 243, 407 preformulation, 298
O’Reilly, K., 209, 222 pre-testing, 76, 33, 34, 287
Orwell, George, 3, 69 principles, 28
O’Sullivan, L. et al., 121 privacy: covert research and, 220; ethics and, 57–61; Internet
“other,” as response option, 98 and, 58
outliers, 179, 407 privilege, 353
probabilistic statements, 4
Pahl, J., 111 probability sampling, 163, 167, 185, 407; ethnography and, 233;
panel conditioning effect, 43–4 qualities of, 153–5; sampling error and, 149–50; types of,
panel studies, 27, 42–3 150–3; validity and, 163
paradigm argument, 333–4 probing, 115–16
paradigms, 333–4, 407 problematizing the situation, 369
Parker, M., 250 production, means of, 12, 179
Parnaby, P., 164, 279, 282, 288 prompting, 116, 248; photographs and, 276
participant-as-observer, 224 Psathas, G., 291
Index 433
pseudonyms, 59, 261 questions: ambiguous terms in, 100–1; attitudes and, 98–100;
publication, 76; journals and, 21, 89, 347 balanced answers to, 103; behaviour, 98, 101, 105; beliefs and,
Punch, M., 56 100; closed (closed-ended), 95, 97–8, 102, 129, 403; closed
v. open, 96, 99; common mistakes in, 106; design rules for,
Qian, Yue, 291 100–9; direct, 246, 248; double-barrelled, 101, 102; ending,
QSR NVivo. See NVivo 246; existing, 108–9; factual, 98; filter, 103, 118; focus group,
qualitative interviews, 199, 239–67, 267–8; checklist for, 267; 254–5; follow-up, 245; formulating, 245; general, 101; general
ethnography v., 264–6, 268; feminism and, 262–4; length indirect, 246; intermediate, 246; interpreting, 246; inter-
of, 249–50; online, 258–9; recording and transcription, views and, 245–7; introducing, 245; knowledge and, 100,
244, 248–50; setting of, 244; structured interviews v., 240; 103; leading/loaded, 101–2, 244; long, 101; negative, 102;
as term, 243; websites about, 270; see also focus groups; open (open-ended), 81, 83, 95, 96–7, 106, 129, 246, 406; order
interviews of, 103–5, 130; overlapping answers, 103; overstretching
qualitative research, 16, 24, 198–213, 407; case studies and, 47; people’s memories, 103; personal, 98, 105; pilot studies and,
content analysis and, 272–3, 328, 407; context in, 207–8; cri- 107–8; pre-coded, 171, 407; pre-testing, 95; probing, 245;
tiques of, 211–12; cross-sectional design and, 41; data analy- prompting, 116, 248; response sets and, 105, 130; specific,
sis in, 72, 305–20; description in, 207–8; design and, 27, 47; 101; standardization and, 97, 98, 112, 129; structuring, 246;
empiricism and, 326; ethics and, 52, 56, 59, 62; evaluating, technical terms and, 102–3; triple-barrelled, 101, 102; two (or
204–6, 213; feminism and, 20, 24; flexibility and, 209–10; more) questions, 102; types of, 98–100; vignette, 95, 105–7,
goals of, 198, 206–11; inductive approach and, 44; longitud- 247; yes/no, 106; see also research questions
inal designs and, 44; multi-method approach, 199; mutual Quiet Rage (film), 68
analysis and, 331–2; natural science model and, 326–7; pro-
cess and, 70, 71, 208; quantification and, 332; quantitative “race”: official statistics and, 128
research and, 212–13, 326–7, 331–2, 336–7; questions in, Radley, A. and D. Taylor, 231
22, 23; reliability and, 204, 206; report writing and, 350–2; random assignment, 31, 32, 35, 407
sample size and, 19; steps in, 199–201, 200; theoretical work random-digit dialling, 111
in, 200–1; types of, 199; validity and, 204, 206; see also quan- randomization: interviewing and, 112; see also samples/
titative/qualitative distinction sampling, random
quantification: content analysis and, 272; qualitative research random response technique, 57–9
and, 332; quasi-, 330, 332 range, 180, 385, 408
quantification rhetoric, 296, 297–8 reactive effects/reactivity, 33, 54, 139, 266, 330, 408; documents
quantitative/qualitative distinction, 2, 16–17, 324, 343; artifi- and, 273; ethnography and, 220; measurement, 139;
cial v. natural, 330; behaviour v. meaning, 329; debate over, participant observation and, 224; quantitative/qualitative
334; epistemological argument against, 334; multi-strategy distinction and, 330; structured observation and, 139–41,
research and, 332–4; mutual analysis, 331–2; numbers v. 143; threat to external validity, 33
words, 329–30, 332; problems with, 329–30; technical argu- realism, 408; critical, 300, 404; ethnographic reports and,
ment against, 334; theory testing and, 329 354–5; quantitative/qualitative distinction and, 326; visual
quantitative research, 16, 24, 38, 74–92, 407; concepts and, materials and, 229
76–81; conclusions and, 72; correlation and, 28–9; construc- realist tales, 354–6; experiential authority, 355–6; interpretive
tionism and, 328; content analysis and, 272; critiques of, omnipotence, 355, 356; subjects’ point of view, 355, 356;
89–92; cross-sectional designs and, 41; data analysis and, typical forms, 355, 356
72, 76, 170–94; deductive approach and, 44, 75; design and, reality: ontology and, 13–15
47, 76; epistemological assumptions of, 75; ethics and, 52; recordings, 261; field notes and, 227, 228; focus groups and, 255
feminism and, 20, 341; generalization and, 87, 213; goals Reed, M., 300
of, 85–9; internal validity, 85–7; interpretivism and, 327–8; References section, 370
longitudinal designs and, 41; measurement as goal of, 85; reflexivity, 18, 290, 354, 408; postmodernism and, 353; in quali-
measurement validity and, 83–5; mutual analysis and, 331–2; tative interviews, 250; visual materials and, 230
ontological assumptions of, 75; process and, 70, 71; qualita- regression: multiple linear, 191, 193–4; in SPSS , 390
tive research and, 212–13, 331, 336–7; questions in, 22; reli- Reinharz, S., 235, 236, 262
ability and, 81–3; replication and, 87–9; report writing and, relational statements, 4
347–50; steps in, 75–6, 75; variables in, 170, 173–7; see also relationships, 338, 408; causality v., 181–2; direction of, 181, 182;
quantitative/qualitative distinction linear, 182; moderated, 406; negative, 38, 182, 183, 406; posi-
quasi-experiments, 27, 34–7, 139, 407 tive, 86, 182, 183–4, 183, 190, 407; spurious, 28, 182, 189–90,
questionnaires, 29, 37, 116–22, 130, 171, 325, 407; “bad,” 109; 190, 408
checklist for, 110; clear presentation and, 120; design of, reliability, 74, 77, 85, 408; coding and, 331; content analysis
120–1; embedded/attached, 118; focus groups v., 258; in- and, 285, 287; inter-coder, 285, 287, 405; internal, 82, 405;
structions for, 120–1; mailed, 119, 166–7, 209, 407; online, inter-observer consistency, 82–3; intra-coder, 285, 287, 405;
116, 118–20; quantitative research and, 76; reminders, 167; official statistics and, 285; qualitative research and, 204,
respondent problems with, 120–1; response rate and, 166–7; 206; quantitative research and, 81–3; stability and, 81–2;
sample, 172–3; structured interviews v., 116, 117–18 structured observation and, 133, 137
434 Index