Case Study Research
Case Study Research
Continuum
London and New York
ContinuUDl
Wellington House 370 Lexington Avenue
125 Strand New York
L~ndon WC2R OBB NY 10017-6503
Contents
© 2000 Bill Gillham
v
1
Series Foreword Case Study Research:
Underlying Principles
The bookS in this series are intended for those doing small- What is a case study?
scale research in real-life settings. No previous knowledge of
research methods is assumed and the series is particularly Perhaps we should first ask: what is a case? The word 'case'
suited to practitioners studying for a higher degree or who (like 'intelligence' and 'neurosis') is one we all use, and feel
want to research some aspect of their practice. The thinking we understand, but is rather challenging to define. Here is an
underlying the series reflects a major shift in social science attempt:
research methods over the past fifteen years - away from a
• a unit of human activity embedded in the real world;
natural-sciences style which emphasizes deductive theory-
• which can only be studied or understood in context;
testing, a prior determination of method (usually experi-
• which exists in the here and now; .
mental) and 'generalizable' results, towards a recognition
• that merges in with its context so that precise boundaries
that such requirements are often unworkable and inap-
are difficult to draw.
propriate in the real world.
This is not a defect, because the traditionally 'scientific' A case can be an individual: it can be a group - such as a family,
methods are often not adequate as a way of understanding or a class, or an office, or a hospital ward; it can be an
how people behave 'in context'. This does not mean that one institution - such as a school or a children's home, or a factory;
should give up an empirical, evidence-based research tradi- it can be a large-scale community - a town, an industry, a
tion but adapt to what is possible and, more importantly, profession. All of these are single cases; but you can also study
what is likely to yield a truer picture. multiple cases: a number of single parents; several schools; two
different professions. It all depends what you want to find
Bill Gillham out - which leads us on.
A case study is one which investigates the above to answer
specific research questions (that may be fairly loose to~begin
with) and which seeks a range of different kinds of evidence,
evidence which is there in the case setting, and which has to
be abstracted and collated to get the best possible answers
VI
Case Study Research Methods Underlying Principles
to the research questions. No one kind or source of evidence experiments or other carefully controlled investigations
is likely to be sufficient (or sufficiently valid) on its own. which are presumed to yield 'proven' results of potentially·
This use of multiple sources of evidence, each with its great importance. Note that 'research evidence' for most
strengths and weaknesses, is a key characteristic of case people equates 'scientific' evidence. Second, the word
study research. 'evidence' is used in courts of law and judicial inquiries,
Another fundamental characteristic is that you do not as part of the process of judging whether allegations or
start out with a priori theoretical notions (whether derived concerns are likely to be true or not. Much of the process
from the literature or not) - because until you get in there of criminal and civil law is concerned with defining and
and get hold of your data, get to understand the context, you testing evidence.
won't know what theories (explanations) work best or make One big difference is that 'scientific' evidence is, in a sense,
the most sense. manufactured. It is an outcome of the investigative methods
used: it didn't exist before. Judicial evidence is there in the
case being investigated and has to be uncovered and tested,
FundaD1entals of research usually by reasonable argument. Evidence that is a result of
the techniques of investigation (for example, asking the
At this point we need to step back and consider some of the wrong kind of questions, or aggressive interviewing) would
underlying principles of research investigation in general, be disallowed.
and case study research in particular. It should be noted here The first kind of evidence is natural-sciences style. The
that case study research has only recently come into its own, natural sciences are those mainly concerned with the ma-
not being part of the natural-sciences style positivist philoso- terial aspects of our world, and the core disciplines are
phy which in dilute form has dominated the human sciences physics and chemistry ..Physics is the study of motion and
for so long. In its extreme, original form, positivist philoso- the interactions of matter. Chemistry is the discipline which
phers asserted that only observable, and verifiable, phenom- investigates the properties of natural (and artificial) sub-
ena could be the subject matter of science; this excluded stances. Both disciplines are concerned with the development
subjective phenomena or 'unverifiable' theories. It will be of evidence and of generalizable theory or laws as to how these
argued below that the naturalistic style of case study research natural phenomena work. With inanimate substances and
makes it particularly appropriate to study human phenom- their 'activities' scientists can do what they like. Holding
ena, and what it means to be human in the real world 'as it some conditions (called 'variables') constant, they manip-
happens'. ulate others, to see what results emerge. They may study,
Research is about creating new knowledge, whatever the for example, the effects of very low temperatures on the
disciplines - history, medicine, physics, social work. The raw molecular structure of steel subjected to different levels of
material of research is evidence, which then has to be made vibration.
sense of. Provided there are no dangerous effects, scientists can do
In everyday understanding the word 'evidence' is used as they will with their 'natural' materials. There are no
in two main ways. First, to refer to the findings of ethical problems and the materials won't complain or try to
'scientific' research, usually carried out in the form of make sense of what's g~ing on.
2 3
Case Study Research Methods Underlying Principles
With human beings there are great limits to what one can type. The recall of isolated words, or groups of digits, has
do in manipulating conditions that might affect human been tested under various conditions to see how the memory.
behaviour. But even if there were no ethical barriers there works. One apparently durable finding was that the human
are other problems in being objectively 'scientific'. memory can hold, at anyone time, no more than seven 'bits'
For example, medical scientists carrying out clinical of information plus or minus two. However, this is in the
trials of a new drug have to deal with the placebo effect: laboratory. Human beings are inextricably part of their
the fact that if people believe they are being given a new environment: they may behave differently (not typically or
drug, their symptoms may improve for reasons nothing to 'normally') in a controlled 'laboratory' setting. For example,
do with what they have been given. Medical scientists go an American researcher investigated the phenomenon of
to great pains to get round this, at least using double-blind short-term memory in cocktail waitresses. He found that
trials, i.e, where the patients don't know what (drug or these young women, working at speed, could easily
placebo) they are being given, and the scientists who are remember thirty to forty separate drinks orders at any
actually administering the doses also don't know which one time. This is memory working in context with other
one they are giving (so that they can't unwittingly com- elements to help recall. (Henry L. Bennett (1983)
municate which is which). In a triple-blind study, those 'Remembering drinks orders: the memory skills of a cocktail
scientists who assess the improvement in patients from both waitress', Human Learning, 2, 157-69.)
groups don't know which patients have been given the Over the past fifteen years there has been a major growth
drug and which not (so that they can't bias the evalua- in ecological psychology (the study of humans in interaction
tion) . with their environments): a growth which corresponds with
This is an extreme example where the subjects' hopes and the increased importance of non-experimental case study
expectations are clearly high. But the point is that all subjects research. Ecological psychology embodies many of the prin-'-
in experiments are going to have some kind of understanding ciples expounded here.
or expectation of what the experiments are about - and this The foregoing has not been an attempt to 'rubbish'
can affect the results. Experiments work best where the experimental research, but rather an attempt to redress the
activity is not seen as intrinsically important to the subjects, balance. Novice researchers are often obsessed with being
or where subjects can be treated as partners in the investiga- 'scientific'; insofar as this means getting good quality evi-
tion. For example, if as a designer you wanted to develop dence and interpreting and checking it in legitimate ways
safety bottle caps which young children couldn't operate but this is laudable. But it is often taken to mean 'hard science'
weren't too difficult for older people with weak wrists or methods - which are often misunderstood and used inappro-
arthritic hands, you could 'share' your purposes without priately.
significantly affecting the results.
But there is a second problem with experiments: in an
attempt to get to the fundamental research questions, Quasi-judicial or 'naturalistic' research
scientists may 'strip down' the procedures so that they can
be carried out under controlled 'laboratory' conditions. Both 'natural-sciences style' and 'naturalistic' research are
Research on memory, until quite recently, has been of this legi tima te methods of enquiry. You use the methods (and
4 5
Case Study Research Methods Underlying Principles
therefore the underlying philosophy) which are best-suited theorizing, i.e. making sense of what you find after you've
to what you are trying to find out The key question is: how found it.
appropriate is the method to the phenomenon you are But perhaps the major distinction is the greater concern of
dealing with? In other words does the method used mean naturalistic, case study research with subjectivity: with phe-
that important elements are missed out or constrained? nomenological meaning. This doesn't mean that you ignore
Our argument here is that 'experimental science' type the objective (what people do, what records show, and so on)
approaches are ill-suited to the complexity, embedded but that you are after the qualitative element: how people
character, and specificity of real-life phenomena. understand themselves, or their setting - what lies behind the
This last point (specificity) is a key one. Natural sciences more objective evidence. Nor does it mean that you ignore
research is aimed at generalizable findings (which may have 'results' (reading standards in a school, staff turnover in a
general implications for theory). But in human behaviour, children's home, after-care of hospital patients) but that you
generalization from one group of people to others, or one seek to find the underlying reasons - in people's feelings or
institution to another, is often suspect - because there are too perceptions, or their experiences of what is going on. This
many elements that are specific to that group or institution. concern with process (leading to the outcomes or 'results') can
For example, what is true about one school (e.g. the causes of be key to understanding what needs to be done to change
bullying, or low achievement, or high delinquency rates) things.
may well not be true of another. All of this means that the naturalistic researcher is not a
Because of this unknown degree of specificity, and the detached 'scientist' but a participant observer who acknowl-
uniqueness of what are likely to be the facts (and how they edges (and looks out for) their role in what they discover. A
are to be explained), the naturalistic researcher differs from research investigation is not neutral; it has its own dynamic
the experimental investigator in another important way. In and there will be effects (on individuals, on institutions)
the natural-sciences style you study the literature and work precisely because there is someone there asking questions,
out whether existing findings and theories are adequate. If clarifying procedures, collecting data. Recognizing this is
you feel that certain data are not there or that existing part of doing good research. Ignoring it is bad 'science'.
theories need testing or challenging, you set up an experi- A lot has been covered in this chapter and it may not be
mental procedure to yield new data to test existing theory. easy to grasp as a whole. The following table summarizes (in
This is the deductive model, using a predetermined procedure what has to be acknowledged as rather artificial opposition)
of investigation. the key differences between the two approaches.
The naturalistic researcher cannot work like this: the data
and theories in the literature may have little bearing upon
the 'case' under investigation. The researcher needs to know
what others have done (and their explanations) but cannot
be sure they're relevant. The first stage is to review the context
from which the research questions, the means of investigat-
ing them, and likely explanations will emerge. An emergent
design is characteristic of this style along with inductive
6 7
Case Study Research Methods
Quasi-natural sciences!
positivist
Quasi-judicial!
naturalistic
2
Emphasis on: Emphasis on:
• experimental methods • non-experimental methods
• deductive theorizing, i.e. • inductive theorizing, i.e.
Case Study Research:
hypothesis testing .
• preordained research design
hypothesis seeking
• emergent research design
The Qualitative Dimension
• objectivity • subjectivity
• detachment • participation
• quantitative data to • qualitative data to give
determine significance of meaning to results
results
• significance or otherwise of • meaning of processes that 'Qualitative' is one of those words that tend to faze people.
outcomes lead to outcomes Students writing in exams sometimes use the words 'qualita-
• demonstration of changes • meaning of changes that have tive' and 'quantitative' with what is evidently only a hazy
that have occurred occurred notion of their meaning.
• generalizable data sought • generalization regarded as Q,uantitative methods are those which involve counting
suspect: the context and measuring: the much-dreaded subject of statistics.
specificity of data is Tackling one's anxieties about this is an important task for
recognized the novice researcher. Like most objects of anxiety the
• isolating the elements of • the importance of context in reality is a good deal less troubling than the anticipation;
behaviour for investigation shaping behaviour
and once you get to grips with it (even if you are no
• constructing evidence • searching for evidence in
mathematician) you will find much of it relatively easy
context
and - more important - practically useful. Statistics are of
two kinds: descriptive and iriferential. Descriptive statistics are
Like all oppositional comparisons the contrasts here are too
things like averages (usually called 'means') which
simple and too strong, but the dimensions of comparison are
'describe' data in a summary fashion. Inferential statistics
essentially correct. However, in the chapter that follows the
are those which enable you to draw potentially meaningful
merging or breakdown of the hypothetical barriers will be
and significant inferences from quantitative data. For
discussed.
example, if boys and girls show different proportions of
success or failure in an exam, inferential statistics will
enable you to say how likely it is that the difference is a
'chance' one, i.e. whether or not it is 'significant'. The
chapter on chi square in the companion volume in this
series Developing a Q,uestionnaire illustrates this kind of tech-
nique in some detail; and the later chapter in this book on
quantitative methods, in case study research takes a wider
8 9
Case Study Research Methods The Qualitative Dimension
view. That is all we need to say here about quantitative What qualitative tnethods enable you to do
methods.
Q,ualitative methods are essentially descriptive and inferen- 1. To carry out an investigation where other methods - such
tial in character and, for this reason, are often seen as 'soft'. as experiments - are either not practicable or not ethically
But description and inference are also necessary in 'scientific' justifiable.
research. You may have significant statistical results, i. but 2. To investigate situations where little is known about what
these have to be described and interpreted: 'facts' do not speak is there or what is going on. More formal research may
for themselves - someone has to speak for them. And it is here come later.
that the quantitative/qualitative distinction starts to break 3. To explore complexities that are beyond the scope of
down. more 'controlled' approaches.
, Qualitative methods focus primarily on the kind of evi- 4. To 'get under the skin' of a group or organization to find ~
dence (what people tell you, what they do) that will enable out what really happens - the informal reality which can
you to understand the meaning of what is going on. Their only be perceived from the inside.
great strength is that they can illuminate issues and turn up 5. To view the case from the inside out: to see it from the 1/
possible explanations: essentially a search for meaning - as is perspective of those involved.
all research. 'Scientific' researchers, before they run their 6. To carry out research into the processes leading to results 1
experiments, will often engage in qualitative-style investiga- (for example how reading standards were improved in a .
tions which lead to hunches about what modifications could school) rather than into the 'significance' of the results
be made to existing theory, or how different results from themselves.
those in the existing literature could be obtained: another
These are strong characteristics and represent a powerful
part of the blurring of the distinction.
argument for the use 6f qualitative methods to answer some
However, before we go further down this road it needs to
questions in some settings. But we need to be clear about the
be emphasized that case study research is not exclusively
philosophical base. There are three main points:
concerned with qualitative methods: all evidence is pulled
into the case study researcher's data collection. However, 1. Human behaviour, thoughts and feelings are partly
qualitative methods (and what they enable you to do) are determined by their context. If you want to understand .
pnmary. people in real life, you have to study them in their context
Thus for the case study researcher all evidence is of some and in the way they operate.
value, and this value (trustworthiness) has to be carefully v 2. 'Objective' research techniques - abstracted, controlling
appraised. Reality (and the truth) is not tidy. A judge - can produce results that are artifacts of the methods
presiding over a judicial inquiry (as distinct from a court of used. An artefact is something that only arises because of
law) turns no evidence away but assesses what faith can be the method that has been used (like controlled memory
placed in it, and relates it to other evidence to hand. Broadly experiments in a laboratory or 'opinions' given in a
speaking, this is the approach of the case study researcher. questionnaire). You get results, but are they 'true' for
the people concerned in the practice of real life?
, 3. How people beha~e, feel, think, can only be understood if
10 11
Case Study Research Methods The Qualitative Dimension
you get to know their world and what they are trying to tion. One must bide one's time; other evidence may qualify,
do in it. 'Objectivity' can ignore data important for an or complicate, or contradict what came early on.
adequate understanding. Very broadly, the case study researcher must strive to keep
an open mind, to go on looking for data, deferring analysis
until the array is comprehensive (and you don't stop com-
What you are looking for in qualitative research pletely, even then).
12 13
Case Study Research Methods
14 15
Case Study Research Methods Research Preliminaries
• deciding, in a not too focused fashion, what your broad patients taking the drugs but not complying with other
aims are; aspects of medical advice (diet, exercise)?
• making a start on getting your research questions into The point is that the questions emerge, and may change
shape. radically as you get to know the context at first hand.
The importance of framing your research direction in
You don't rush into carrying out your actual investigation, the form of questions is that you are then driven to consider
because if you do so without adequate time given to these your methods: How would I answer those questions? What
first steps your research will be inadequately based, almost information do I need and how would I go about getting
certainly a waste of time. it?
Everything follows from this first essential stage. Framing good questions is the most important part of
research procedure. In real-world research they are not
easily achieved and you will have to spend quite a long time
developing or modifying them. 'Good' research questions are
Ahn.s and questions those which will enable you to achieve your aim and which
are capable oj being answered in the research setting. It's no use
Most people start their research with a broad aim in mind. asking questions that can't be answered.
They want to find out why more boys than girls truant from In the real world there are always constraints on what the
secondary school; or why staff on a company training scheme researcher can do - ethical, practical (in particular avoiding
often fail to complete training; or - to pick up a theme from making life difficult for people who have many other things
the end ofthe last chapter - to find out, in a medical general to contend with). So you have to perform a balancing act
practice, what proportion of patients don't take prescribed between what you want to find out and what the setting will
drugs properly, and why that is so. allow you to do. There is a positive side to this: there may be
In research, broad aims often remain the same. What unrealized potentialities. A real example occurred in a
changes and evolves is the set of research questions. These research project led by the author, which was looking at the
emerge in response to asking yourself: what do I need to find use of volunteer tutors with failing readers in the first two
out in order to achieve my aim? To take the last example years of the primary school. It is easy to demonstrate short-
given above, research questions might be: What proportion term effects with this kind of intervention - but does it last?
of patients don't comply with medical advice on drugs? Are In the course of the research (which was to run for several
there differences (i.e. age, social class) between different years) we found that the school had their own longitudinal
categories of patient? Is age a factor? Is the medical condi- da ta on reading trends in the third year of the school. Thus we
tion a factor? were able to plot changes, more than a year after the
Now you may start off with questions like these but, as you learning support ended, with a sequence of pre-intervention
get into the research, as you get to talk to patients, doctors baseline scores, i.e. those third years who hadn't had the
and practice nurses, other questions might emerge, e.g., how intervention, with those in successive years who had. Because
clearly are patients told about drug use and the need for of that potentiality we were able to ask questions about
compliance? Would follow-up improve compliance? Are follow-up effects that we hadn't thought possible.
16 17
Case Study Research Methods Research Preliminaries
Keeping an open lDind that we feel impelled to understand, to make sense of what
we are experiencing. New knowledge is mainly interpreted in
The broad strategy in case study research is to start by terms of what we already know, until that proves so
collecting data (and looking for it) with as open a mind as inadequate that our 'knowledge framework' undergoes a
possible. A particular difficulty here is that we all carry a lot radical reorganization. In research this is sometimes known
of conceptual baggage with us. We 'know' what it's like to be as a paradigm shift - a complete change in the way we
an office worker or a teacher, or a GP, or whatever. Well, we understand or theorize about what we are studying.
know what it was like for us and, being human, we can easily The American social anthropologist Clifford Geertz
assume that that gives us privileged understanding of others emphasizes the importance of beginning research into any
in similar contexts. This very familiarity can blind us and culture by describing what you find in some detail. He calls
close our minds. this thick description: a process which makes you pay attention
You need to take the stance that you are going into a to the fine grain of what you are observing, and reflecting on
foreign country. In the 1920s, a cultured Swiss, L. S. Renier, it. We are now getting to the central concern of the case
recorded his impressions of the English in a famous book, The study method: the collection and study of multiple forms of
English: Are They Human? It is a highly instructive read. evidence, in sufficient detail to achieve understanding.
Renier was able to see the English with uncomfortable
clarity because he knew quite well they were different from
himself. Accordingly he didn't make assumptions - of famil-
iarity or similarity that another Englishman might. So even
if you 'know' the setting you have to act as if you didn't:
because you don't.
When you move your job (within the same profession) it
can be very surprising how an ostensibly similar school,
hospital, office or factory is profoundly different from what
you've experienced, once you get to know it. Indeed some of
your initial settling-in difficulties may be because of assump-
tions you've made (almost unconsciously).
The researcher needs to be alert to this from the begin-
ning. The stance needs to be that of social anthropologists
going to study a culture quite different from their own. Tht;.
framework they take with them is a strategy for gathering
evidence - detailed evidence, the significance of which (or
otherwise) will only gradually emerge. Analysis, sorting,
categorizing and theorizing must be deferred for the
moment.
That is not easy. A basic limitation of human cognition is
18 19
Evidence: The Primary Concern
20 21
Case Study Research Methods Evidence: The Primary Concern
records of, for example, sketches, mock-ups, materials maintain a 'manual' one: a flip-over secretarial pad is ideal.
used, prototypes, and so on. This goes with you everywhere. In it you put down two main
kinds of things:
Why this evidence is so illlportant • Evidence of one kind or another: a discussion you've
heard in a staffroom; a comment made to you; some-
The basic way of presenting a case study report is a narrative thing you've observed (e.g. a worker by-passing a
following the logic and chronology of your investigation and required safety procedure). These may be fragmentary,
reasoning. In a sense that is true of all scientific papers: they unrepresentative, but they occurred: so they're evidence.
have a 'narrative' sequence and 'tell a story', although they You need to follow this up: are they typical, are they
are rarely a compelling read. But the case study researcher, true, etc.?
who is seeking to recreate the context and sequence of • Personal notes: questions you need to reflect on; insights,
evidence in a way that enables the reader to see and under- hunches or ideas; a report you hear mentioned that you
stand the meaning of what is recounted, has to use a more need to get a copy of; the name of someone you need to
overtly narrative format. Well done, these can be compel- consult; statistics you need to check; and so on.
lingly readable; but they are open to the criticism (sour
A log book is necessarily summary in the main (not always:
grapes?) that they are nothing much more than a good story.
sometimes you will feel the need to record observations in
Well, the truth can be a good story, though not usually a
depth and at length). Entries are best made immediately: a
very tidy one, and presenting this well does call for some
habit that will save you a lot of trouble. In any case, writing
literary skill - of a plain, straightforward kind. But at each
things down, even very summarily, has a curiously clarifying
key point in the narrative - and continuously in an inter-
and focusing effect 0!1 the mind. It is part of research
woven fashion - evidence needs to be presented for the
discipline.
development and direction of the narrative. This must be
Your log is in a sense a private thing, but it is more than
much more than impressionistic: impressions and assertions
that. You need to take the stance that all of your research
must be substantiated in some way.
records (and what they Show) are open for inspection;
Here we are anticipating the outcome of the study, but it is
organized but not 'tidied' so that someone else could follow
to make the point that the ability to do that depends upon
the process of the investigation. This notion of ' open account-
the maintenance of organized and detailed records through-
ing' is partly an ethical stance - demonstrating your reason-
out the project. Record-keeping is of various kinds, but there
ing and the chain of evidence on which it is based. But it is
is one element that is crucial.
also of great practical value when you come to write up your
report. When you are busily 'in' the research activity and
Maintaining a research log things are moving fast (research questions and explanations
developing rapidly) it is easy to lose sight of where you've
Here is where we take a leaffrom the social anthropologist's come from: yet this is a key dimension of the research process.
book. The log book is more than a set of rough notes: it is a
A research log can be electronic but it is still more usual to fundamental part of your database, and needs to be treated
22 23
Case Study Research Methods Evidence: The Primary Concern
with respect. It is also the beginning of what Guba and perhaps for others) of what you've achieved and how your
Lincoln, pioneers in naturalistic research, call the 'audit research design and theory have developed.
trail' - something that an 'auditor' could follow to under- As the data collection progresses (and accumulates) you
stand your procedures (see Effective Evaluation (1981), San will move from gathering data to making more focused,
Francisco: J ossey Bass). selective decisions about what you are going to concentrate
So, you carefully preserve your log books. But you do more on. For example, you may have started out looking at the
than that. Because they are often hastily written by hand it is whole picture of bullying in a school but, because it has
good discipline to do a daily and exact transcription onto a emerged as a neglected dimension, decide to focus on
word processor taking care to date and number each day. bullying that occurs on the way to school and which appears
You could if you wish then colour code the contents, particu- to be a factor in school absences.
larly the 'evidence' and 'personal' notes, and separate them Decisions of this kind need to be recorded (and justified) in
into distinct manual files - especially important for the your review records.
'evidence' notes. Data analysis procedures (how you're going to order and
You will then have your 'field notes' available to yourself present your findings) become increasingly important as the
and others in legible, readily accessible format. investigation moves on. At some point data collection has to
At the end of each week, or whatever interval seems virtually stop and will have been winding down for some
appropriate, you need to read through your notes and time before that. Somehow you have to reduce this mass of
review what you've turned up. What you then have to do is data. There are standard techniques for analysing different
to prepare a summary of: kinds of data - these are described in the chapters that follow
dealing with particular kinds of evidence. But in the same
• the different types of evidence you've uncovered; way that case study evidence may be specific or peculiar to
• what your immediate priorities for action are; the case in question, analysis also has to be appropriate and
• any reworking of your research aims and questions; not a straightjacket that deforms your findings. This may
• your 'theorizing': what you think this might be about, mean that you have to be creative about your methods of
how it might be explained. analysis. Bear in mind that the purpose of analysis is to
If you do regular summaries like this you will be able to plot faithfully reflect in summary and organized form what you
the progression of your thinking when you come to write have found.
your report. A key term in naturalistic research is 'trustworthiness'.
Reviewing and summarizing is essential to the discipline of Although there have been fraudulent researchers, there is
case study research: a kind of intellectual stocktaking. It little real satisfaction in methods and 'findings' that lack
needs to be done regularly and systematically - only in that integrity. But interpreting research data is more than a
way will you maintain the necessary level of control over the matter of good intentions. It requires discipline and
process. Case study research can easily 'lose shape' because of concentration to present a 'true' picture: anything that
the complexity of the material. gets in the way of that threatens the validity of your
Once a month you should do a major review of progress research.
and write a progress report (for your own records and Reality - a reflection of the real world that your case
24 25
Case Study Research Methods
26 27
Case Study Research Methods Evidence: What to Look Out For
When you read research papers you can often see that teachers, for example, may get a better view of how a school
people have found what they wanted to find. Which is hardly works than a visiting inspector.
surprising. But researchers of integrity are constantly chal-
lenging and scrutinizing themselves. And the first stage in
that process is to get your expectations and preferences out Looking for discrepant data
into the open. You acknowledge them. You write them
down. They are as much a part of the research process as As you proceed you acquire a lot of information and you
anything else. 'Objectivity' in the absolute sense may be an develop provisional explanations. But are there data that
impossibility but that doesn't mean that you immerse don't fit these 'theories' that you are developing? Looking for
yourself in an uncritical subjectivity. You strive for a level of negative, i.e. opposite or contradictory, evidence, or evi-
detached honesty which acknowledges your own place in the dence that qualifies or complicates your emerging under-
scheme of things. In a sense, you decentre from yourself. standing, is basic to research integrity. The temptation is to
There are a number of ways in which you go about this. close your mind, to think that you've 'got it'. For example,
you may have decided that the management in a factory are
not to be blamed for violation of safety procedures because
they hold regular safety training programmes for workers, so
Absorbing the culture that it is a matter of individual worker responsibility or
irresponsibility. But as you continue to watch what people on
II. This is the heavy emphasis at the beginning of the research the shop floor do you may see that the training is not suited
process; and it never entirely stops. to the practical requirements of doing the job - so that
You go in with your eyes and your ears open: you look and workers take short-cuts to keep to schedule - and that shop-
you listen. What's it like to be a pensioner subsisting on the floor supervision largely disregards safety procedures except
basic state pension? What's it like to work in a fast-food for a temporary increase in concern and rigour following an
outlet? What's it like to work in an open-plan office the size of accident.
a factory floor? What's it like to be a probationer teacher in
an inner-city school? You can only find out by spending time I
with people in their setting. In a sense each location has its Triangulation: taking different bearings
own culture: the conventions by which it works. It also has its
own values and 'language' - ways of judging and thinking , A constant theme in this book is that different kinds of data
and talking about the living experience. (or different sources) bearing on the same issue commonly
It takes time to penetrate that. Not least because we all ~ yield contradictory or 'discrepant' results. If every kind of
have a 'front' - which we may believe in - that we present to evidence agrees then you have simple, confirmatory triangu-
outsiders. The value of being a participant observer, perhaps lation. If what people say, and what they do, and what
becoming a temporary member ofthe setting, is that you are records show all concur then you have a straightforward
more likely to get to the informal reality. Outsiders of a picture.
perceived high or official status may never get there. Trainee Often you don't get that. The straightforward 'fix' doesn't
28 29
Case Study Research Methods Evidence: What to Look Out For
apply. It doesn't mean that one set of data is 'untrue', rather disproportionately apparent. This is one of the weaknesses of
that the presumed relationship with the triangulation point journalism where (usually) the most quickly accessible is
either doesn't exist or has to be understood differently. what gets published. But that doesn't mean that the picture
A lot of things that make straightforward sense, super- presented is comprehensive or representative.
ficially, can be shown by research not to be or not to operate Whatever the kind of evidence (documents and records,
as you expected. To take the work safety example: if a what people say, what they do, the physical or social context
company was concerned about the rising number of shop- they inhabit) an adequate picture involves a lot of digging
floor accidents, they might set up training programmes as a away. Even published documents, for example, go largely
means of combating the problem. There are archival records unread, or are read without an appreciation of their sig-
of accident rates over several years. Do rates change after nificance. Not all publications are high profile. The stuff is
training is i~itiated? This kind of analysis is called time-series there but it isn't known. Important data are not necessarily
or pattern analysis. If rates do go down is that due to the readily accessible, lying there on the surface, so to speak.
training? Maybe. This is true even with the people who are eager to talk to
But you would have to look at other possible causes. you. The psychiatrist Harry Stack Sullivan, in one of his
:;"
Perhaps shop-floor supervisors have got the message that published lectures on the psychiatric interview, commented
this is a high priority for the management, so they are more that the most important thing to note is not what people tell
vigilant. Workers may feel that carelessness could lead to you about, but what they avoid telling you about. I
dismissal and so on. Different 'bearings' may qualify the So there may be parts of an organization that tend to be
picture. overlooked, aspects of professional practice that are not
acknowledged, and so on. You have to be alert to these
limits, and to the signs ()fwhat is there but not visible.
The representativeness of data
Representativeness is different from triangulation. You listen Asking yourself how you know things
to what people tell you, but are you listening only to some
people? Or, to put it another way, is it that some people - This topic links into the preceding one, but focuses on how
knowing you are the 'observer' - are keen to get their case researchers come to know their evidence. We know what we
across to you? If people want to 'help' you, you need to ask know. But knowledge is either tacit or explicit. If it's explicit
yourself: why? Those who are more cautious may have a we can explain how and why we know it, i.e. we can cite the
quite different picture to present - either because they are evidence. Tacit knowledge (sometimes called intuition) is
more private in their habits or because they know that what where we sense or feel something, often very strongly, but are
they think does not fit the party line. So, is what you're hard put to explain or justify. There is nothing wrong with
getting representative of all shades of opinion? intuitive knowledge: your antennae may have picked up
This is the issue of 'accessibility' . In any area some kinds of something important which needs exploring.
information, some people, are more accessible than 'others. For example, a head teacher may be telling you about the
So that kind of data, the opinions of those people, are school policy and practice in relation to racism in her school,
30 31
Case Study Research Methods Evidence: What to Look Out For
how racist incidents are dealt with, and so on. She is fluent sensitized individuals to what you're about. Some of those,
and coherent; but somehow you don't entirely believe it. on reflection, will perhaps come to you with their further
You need to ask yourself two questions: thoughts.
• why? i.e. what doesn't ring true? It may be that you need to set up a regular consultation
• how would I check out what I've been told? group - particularly if what you are doing has, or could
have, policy implications: again it serves the dual purpose of
You neither accept your intuition as some magical divina- keeping your 'hosts' informed; and getting feedback from
tion of the truth nor reject it because it isn't hard-and-fast them.
evidence. Finally, there may be other key individuals (not neces-
Case study research is very much like detective work. sarily in formal positions of power) whom you could consult:
Nothing is disregarded: everything is weighed and sifted; perhaps thoughtful, experienced or committed individuals
and checked or corroborated. who are experts on the context you are investigating. It is
more likely that you will have to seek them out than that
they seek you ou t.
Checking your ideas and explanations with those in
the culture
Peer consultation
You can do this in various ways. It may be that you give a
presentation to the people in the setting where you are Your peers are other people like you: other students, other
working. This can be important for various reasons. A lot of researchers, your supervisor. These are the ones who are
people will have given you help of one kind or another. doing the same kind of thing as yourself, are experts in the
Perhaps others are curious as to just what you are up to. Of kind of research you are undertaking, or are experts in
course, you should be explaining the purpose of your research methods. They can save you a lot of trouble. Good
research, informally, whenever the opportunity presents supervision, in particular, is of paramount importance as
itself. But that can be a scrappy and incoherent process. research can be a curiously lonely business. A formal
Also a specially prepared presentation signals your wish to research degree is different from an undergraduate degree
communicate, inform - and consult. because you are not following a formal taught course at the
Telling them what you're doing is one thing; asking them level of subject content - although you will normally have
what they think is another. You can turn such a presentation methods lectures or seminars. In a sense you have to write your
into something like a focus group discussion. There is much own curriculum and you are the only one studying it.
to benefit you in the business of presenting a summary If you are following a structured research degree pro-
report; but most important is to test out your provisional gramme then you will have the support of fellow students,
understanding or explanations of what it's all about. And you though each is following his or her own research topic. But
present this by saying: this is what it means to me, this is how methods may be in common and, of course, your exact peers
I make sense of it. What doyou think? Can you put me right? are at the same level as yourself. Their feedback and shared
You'll get immediate feedback but you will also have understanding has a special quality of its own. That is why
32 33
Case Study Research Methods Evidence: What to Look Out For
course tutors regularly organize sessions for each student to confirmatory evidence: evidence that confirms what you
present a summary of their research to their peers. believe or understand. You can see that process in everyday
Finding others who have done the same kind of research as life: people being selective about what they come across, to
yourself, i.e. on the same topic, is more troublesome but can reinforce their beliefs.
be uniquely rewarding - particularly if they are prepared to This is the obverse of being a good researcher and is why,
talk to you. From them you can get specific guidance and on page 29, we suggested that you challenge yourself as to
insights into the informal side of researching the topic of the what you expect or hope to find. Research that is ideologi-
kind that doesn't usually get published. cally driven (e.g. by moral or political beliefs however worthy)
Your supervisor (if you have one) is likely to know a lot is always at risk of being bad research, because contrary
about the broad area of research you are engaged in - evidence may either be discounted or interpreted in a way
without being expert on your specific topic. Perhaps more that fits the belief system.
importantly, they will be expert on the style of research that In naturalistic case study research, theorizing emerges.
you are doing. The tutorial approach is to get you to That is because you cannot usefully theorize in the absence
summarize, explain, justify, question what you are doing. of evidence, or on very little. The evidence you look at is
Over time this has a powerfully disciplining effect on your initially dictated by your broad aims. But increasingly it is
thinking. You will not usually be told exactly what to do, directed by your successively revised theories or explana-
although your supervisor will indicate acceptable and unac- tions. And it is negative or complicating evidence that
ceptable procedures; rather your research programme will precipitates these revisions.
be questioned into shape. You may not think of yourself as a theorist. But we
construct theories (understandings, explanations) in every-
day life about other people. For example, when someone we
Theory-building and the analysis of negative know well behaves in an unexpected way that we don't
evidence understand, what they have done is to give uS evidence that
challenges our 'theory' about them. And it is the new
It is an axiom of scientific philosophy that theories cannot evidence that makes uS realize what we understood them to
be proved - in a definitive sense - only disproved. That is, be like. That is a fortuitous incident: research is a more
you've got a theory that 'explains' all the evidence, and deliberate process. Let uS take a practical but hypothetical
then something comes along - another piece of unarguable example.
evidence - which doesn't fit the theory: which has to In an FE college there is concern about the high drop-out
change. rate from a computing course; in particular, that female
That is what was meant when we said earlier that theory is students leave at a higher rate than male students. All the
not primary, evidence is primary; and that applies to all tutors are male: is sexism rearing its. ugly head? A tentative
kinds of research. 'theory' there.
Good scientists, good researchers, are always testing their Looking at records you find that female students have
assumptions, positively looking out for evidence that chal- lower grades for maths in their entry qualifications. Perhaps
lenges their understanding. The alternative is only to look for they lack the necessary ability? But when you do an analysis
34 35
Case Study Research Methods
36 37
Case Study Research Methods Written and Electronically Stored Material
The two processes (getting to know the literature and would get thousands of citations. But if your interest was in
getting to know your case) should go along simultaneously so bullying by girls then you could key in 'girls' or 'females' or
that your reading and what you are turning up in your case 'gender' which would cut the range, and focus more on your
study interact: they feed into each other. topic.
To a great extent you won't know what you're looking for What the computer doesn't do is think for you, and you
in the literature until you do get into the real context. And may get some very strange titles coming up just because your
what you find in the literature will sensitize your perceptions. keywords appear in them.
This progressive influence is one dimension of the emergent Most computer systems will first tell you how many titles
character of case study research. fit your keywords. There may be so many that you have to
think round ways of pruning the number. This is where
specialist librarians can be a great help: it may not be their
COD1puter-based searches subject area, but they know how the system works and how
to render it manageable. When the number looks feasible 1,1
I
Literature-searching is as much art as science. But computer- you can instruct the computer to print out. This will usually
based searches have made the researcher's life immeasurably give you title and publication details and a summary.
easier. They work best when your research topic fits one or Go through these, highlighting the papers you would like
two of the main databases. These are usually available in a copy of (be sparing: you have to read them!).
university or college libraries that subscribe to the service (at What you are doing is a progressive focusing. A computer
enormous cost); and in technologically advanced institutions search gives you a first quick fix. Ifwe assume that you select
they are available campus-wide or even to distance-learners. ' half-a-dozen papers you then have something to start on.
Some databases are only available on CD-ROM, but these However, finding the references is one thing; getting hold
have declined in importance over the past five years. In any of the actual papers is quite another. If the journals or books
case, some special training in use is required. Research are in the library you're using then they are immediately
librarians can be a great help here, especially in university accessible. But there are an enormous number of journals,
libraries. There is a particular mind-set for this kind of thing, and library subscriptions to them are expensive. A university
apart from 'knowing the ropes'. library's serials budget is usually much more than its books
The major databases involve reviewers reading and budget. Even then they have to be highly selective.
abstracting from an enormous range of journals within one In the UK almost all journals are held at the British
discipline or sub-discipline. This is expensive, which is why Lending Library at Wetherby Spa. To get a copy ofa paper
subscriptions to these systems are correspondingly costly. To you have to put in an inter-library loan request: and these
access them you use keywords (in the database's 'dictionary') are expensive. An intermediate step, however, is to find out
either alone or usually in combination. The computer then whether another university or college or public library
trawls either titles or summaries to see where these keywords holds the journal you're interested in. In which case you
co-occur. The more keywords you use the more 'selective' the can get a copy directly from there. There are sometimes
computer is. For example, if you just put the keyword local databases which will tell you which libraries hold
'bullying' into the main educational database (ERIC) you which journals.
38 39
Case Study Research Methods Written and Electronically Stored Material
Manual databases and other sources Analysing and organizing the literature
Computer-based searches- are a phenomenon of the last The journal papers (or whatever) that you've copied, or
twenty years or so. Before that time one had to use those you have sent for, are to hand. How do you proceed?
enormous indexed volumes, usually classified under sub- The first step is to read each paper through quickly (not
headings but also cross-referenced in various ways. These too many at one time or you'll suffer mental indigestion)
can still be useful - especially browsing through the titles highlighting key points or passages. These are the elements
listed under various sub-headings. Indeed all sources turn you feel are most important or which contain something you
up something different. No system or approach is complete have some further use for. In the text, reference will be made
or infallible. A frustrating but typical experience is that you to other publications that you may want to follow up:
can be well into a project before you find an absolutely key highlight these and, when you've read the paper, go through
reference. the reference list at the end highlighting those papers that
Other sources are: you want to get hold of.
This is how searching proceeds - like an inverted tree
• Browsing through the contents lists of the main journals in diagram where one paper (the 'stem') leads into other
your topic area. Always start with the most recent one 'branches' - which then lead you on to the references cited
(because if you find a paper it will give references to there. To a considerable degree the published literature is
earlier ones) and go back about five years. You'll find interlinked so that, to use another metaphor, when you pick
that you can do this quite quickly. up one thread, others are joined to it.
• Looking through the bibliographies in books on your Gradually you accumulate your references. What you
topic. These can be very extensive, with references going will usually find is that a small number of them is much
back a long way. Photocopy the pages and highlight the more central or useful than others. The process is one of
references you want to pursue. You can do the same with sorting.
the occasional, very lengthy review papers which you may And because you're researching the real-life context at the
find in journals. same time, what's important progressively becomes more
• Talking to experienced researchers in your area. In some apparent. The interaction is an interesting one of intellectual
ways this can be the most valuable source of all: a source discovery. What you read makes you look out for things in
that is intelligent, that can answer questions and, most your research context; what you find there makes you re.ad
important, can challenge and direct you. They will have papers with a different eye.
been through the searching process that you are starting - At some point - not too early - you should attempt a
and will have gone a long way beyond it. Write to them, literature review. You'll need this for your final report
explaining what you are doing and ask if you can come to anyway but you should see it also as a means of clarifying
see them. Don)t send detailed questions which will require things for yourself It's a skill all of its own and the only way
a lengthy written reply: these are the bane of specialist to learn it is to do it. A first draft will be unsatisfactory but
academics. In any case you'll find a face-to-face meeting successive revisions will become increasingly balanced and
richer and more helpful. It will be worth the trip. representative. It isn't just the balance of writing that
40 41
Case Study Research Methods Written and Electronically Stored Material
improves: your thinking does. Writing is a primary discipline out to members of the group is important. If people know
for clarifying what's in your mind. what you're after they may be able to guide you. Some
individuals will be key informants. In a sense they have their
own mental referencing system. One invaluable source can
Researching the 'unpublished' literature be secretarial office staff, who will have word-processed and
filed many of these documents. Their understanding and
Published, or publicly available, documents and records helpfulness can be crucial; but remember to check with the
help you to appraise the wider context that your case management before asking staff to copy documents for you.
inhabits. But there is a more local literature context. Institu- Document search and analysis (a main method if you are
tions in particular will have their own literature which is an historian) epitomizes the case study research strategy.
usually neither published nor available to the public. Some These documents were not drawn up to answer your
of it is generally available by its very nature - school hand- research questions: but they're part of the evidence base. ,
They are not of course to be taken as representative of what !
books for parents, for example. And large organizations or l,
professional groups will probably have a range of public actually h<\ppens - the informal reality. But they bear some )
Locating docuDlents
Getting access to records (archival data)
Because in a group or organization or profession, documents
may not be part of a reference system, learning about their Statistics and summary information are maintained for a
existence (and locating them) may not be straightforward. variety of reasons - usually nothing to do with research per se.
This is where communicating what it is you are trying to find These can go back sev~ral years and so provide a dimension
42 43
r!
Case Study Research Methods
that you could not hope to create for yourself. Getting into
these records may present some problems. For example:
7
• formal/ethical: permission and approval for access and use;
• technical: they may be part of a computerized retrieval Observation: Looking and
system that you can't (or aren't allowed to) operate;
• data format: the data may not be in a form you can easily Listening
use or make sense of;
• data quality: accuracy and completeness.
You can spend much time and effort overcoming these
difficulties, so you need to be clear that it's going to be
worth the effort. It can be. Interesting findings often emerge When we go into a new social situation, one where we don't
from a researcher's painstaking analysis of official or other know the people or the mores, it is the sensible thing to keep a
statistics; and getting round the constraints of the way the low profile, to watch how other people behave before
statistics are organized so as to extract answers of value is plunging in with our own contribution. Not to do so may
often testimony to a researcher's ingenuity. The key thing to mean that one gets off on the wrong foot, perhaps commit-
bear in mind is that the database was not designed to answer ting an irretrievable blunder.
your questions. But then that is true of much of the evidence This is even more important for the naturalistic
that is dealt with in case study research. researcher. He or she also has to gain social acceptance
'Relevance' comes from weighing and assessing and select- through sensitivity to and awareness of what is appropriate
ing the evidence that does have a bearing on the research behaviour. But, unlike_ in the normal social process, the
Issues. researcher remains more in the social background even
when they are 'participant'. They have to earn their place
(like anyone else), but it is not necessarily or usually in the
foreground. It doesn't mean being a 'fly on the wall'.
Detached observation (see below) has its place, but usually
a minor one in naturalistic research.
What is observation?
Observation is of two main kinds: because you've 'seen' it). Unfortunately research on 'witness-
ing' - what people have seen and what they report - shows
• participant: being involved - mainly descriptive, i.e.
that observation is both fallible and highly selective. Becom-
qualitative;
ing anything like an accurate and balanced observer requires
• detached/structured: watching from 'outside' in a carefully
discipline and effort. And because for the case study
timed and specified way - counting and classifying what
researcher the technique is primary - where you start, and
you see, i.e. quantitative.
where the clues for using other techniques are often turned
These two forms of observation are quite different - opposite up - it may be the starting point for error or selective bias.
ends of the observation dimension - and should be seen as From the point of view of positivist 'objectivity' a major
essentially different techniques, yielding different kinds of objection to unstructured participant observation is the
data. But you don't have to choose one or the other. In the effect of your presence on those you are observing. If, as a
same way that there are different ways of getting people to researcher, you work as part ofthe team in a children's home
tell you things - and you can use more than one of them - so you are bound to make a difference: it would be an indict-
you can mix participant/detached observation. ment of you if you didn't. You don't deal with the 'observer
In a case study they constitute just part of the data- effect' by denying it: you look out for the probable influence
collecting techniques. But low-key participant observation ! of your presence. In real-world research as we have men-
is the one you use first: the getting-to-know phase of the tioned before, the researcher is the research instrument, and
research. Structured observation comes later, when you have any instrument used makes some contribution, has some
the research issues well in focus. At that point you'll be able effect on what is found. You have to make a consistent effort
to see where the clarifying (but time-consuming and trouble- to observe yourself and the effects you might be having. You
some) function of structured observation could yield useful can also ask members of the group or institution whether
data. It is a technique not to be undertaken lightly. they think that what happens when you are there is charac-
teristic. A conscious attempt at rigour can usually lead to a
reasonable judgement: we can expect no more.
Observation: pros and cons The observer effect can be judged a little more precisely in
structured observation because you can ask someone else to
The overpowering validity of observation is that it is the check on your observations - using the same schedule and
most direct way of obtaining data. It is not what people have perhaps even at the same time - so as to calculate inter-
written on the topic (what they intend to do, or ShOUld.dO). It observer reliability. More on that later.
is not what they say they do. It is what they actually do (which A major problem with observation of whatever kind is that
v \ may also be reflected to some extent in records). Its very it is time-consuming. Getting to know your case - whether
reality can be overwhelming so that you may easily suspend individual or institutional - is necessarily a slow process.
your critical faculties (like asking: is what I am seeing/ Observing people is slower than asking them about what
hearing typical or representative of this person or this role they do. The point at issue is the cross-validity of different
or situation?). For the novice researcher observation can kinds of evidence (fundamental to case study method) and
seem a very simple business (because it is 'obvious' and the primary validity ~f observation as a technique. In short,
46 47
Case Study Research Methods Observation: Looking and Listening
if you want to do case study research you have to be prepared battery of methods, or comes to be seen as inappropriate
to commit the time to it. or too time-consuming. This is an important decision, and.
The data from observation are also troublesome to collate one to be weighed carefully. It may be that this
and analyse, and difficult to write up adequately, but this is assumption has to be modified later when its potentialities
where your research discipline will help. If you write up your become more apparent.
observations as soon as possible they will be easier to recall and • As a supplementary technique to give the illustrative
also more accurately recorded. Memory 'improves' - simpli- dimension. This is not as superficial a usage as it might
fies and selects - the more time it is given to work. Which is seem - no more superficial than well-chosen visual
why a police officer's contemporaneous notes are seen as illustrations in a text. This can be particularly potent
more valid than a later, 'tidied' version. when the case study is used as a complement to survey
techniques, i.e. involving hundreds or thousands of
individuals or groups or institutions, data that are wide-
scale and representative but hard to translate into specific
The uses of observation
human terms.
• As part of a multi-method approach. This is at the heart of
The use of observation as a technique varies according to the
the case study method (although multi-methods don't just
kind of case you are dealing with and the kind of research
apply to case studies). This is the notion of convergence:
questions you are asking. If your study involves young
different kinds of evidence, gathered in different ways, but
children or older people with severe communication difficul-
bearing on the same point.
ties then observation of them is going to be more productive
• As the main technique when the primary purpose is
than trying to interview them. If you are researching ethical
explanatory description. Describing what you see and
standards in a profession then talking to members of that
explaining it: what could be simpler than that? In fact this
group is going to be primary. In other words, the balance of
is one of the most difficult kinds of observation of all. Like
methods within a case study will always vary - according to
informal interviewing it can seem natural and easy, but it is
what is feasible and according to what you want to find out.
a sophisticated business. Being a good observer, like being a
Observation can be used in these various ways:
good witness, is not a normal, natural activity. It requires
• As an exploratory technique. This is the low-profile discipline and concentration - without which you won't
beginning we mentioned earlier. Experimental psycholo- 'see'. If we are dealing with individ uals, video can be a great
gists may well start in this way having versed themselves help because the same observation can be repeated many
in the literature. But they will be looking for the times: and each time you will see more. For example, if your
constraints and potentialities for the methods they prefer. case study involves mothers and their deaf babies it may
In a broad sense that is entirely analogous with the case only be by replaying video sequences several times that you
study researcher: which directions and methods are see how they communicate with each other. Watching
appropriate or possible here? precisely what people do can be very illuminating. It can
• As an initial phase where other methods will take over. also help you to understand better what people are about:
Here observation is not viewed as part of the potential as we said earlier, particularly important if people can't
48 49
Case Study Research Methods Observation: Looking and Listening
express themselves in other ways. A good example of this appeared to be totally engrossed when working and was
which also shows the particular power of the precise visibly excited by the process as well as the results. She would
description of what is observed is given in the following only release the brush and stop painting once she had
finished and was always eager to start again once a new piece
boxed quotation.
of paper was selected. Although Joginder was unable to
speak, and her understanding of English was reportedly
The researcher Sarah Hall was investigating the value of limited, she definitely responded to encouraging comments
artists/designers working with hospital patients. In this and discussion concerning her work from the other patients.
instance, long-stay patients in a hospital for the elderly in
Glasgow. It would be easy to write generalities to justify this The next stage of developl11ent
sort of activity, although these might not be particularly
convincing. Hall focused on exactly what happened when After several weeks of painting she began to take an interest
patients with sometimes severe medical conditions were in some of the other materials that were available. Although
exposed to this kind of opportunity. Joginder's paintings had developed considerably in a rela-
The particular patient to be described was an elderly tively short time, her abilities to control and manipulate a
paintbrush appeared to be holding her back and she seemed
Asian woman who had suffered a stroke leaving her para-
to sense this. Having begun to realise her abilities and
lysed throughout the right side of her body. She was confined
potential she was keen to experiment with another medium.
to a wheelchair and had completely lost the power of speech. Oil bars seemed a viable alternative to the acrylic paints and
brushes she had begun with as these sticks of oil paint were
After examining the materials and observing the other relatively easy to handle and there were a multitude of
patients' tentative beginnings, she chose a paintbrush and colours to choose from. Joginder happily swapped the paint
moved a palette of acrylic paints into position. Her first and brushes for the oil bars and it was immediately apparent
paintings were very basic and crude circular blobs of a single that she was more adept and comfortable in using them.
colour, apparently randomly spaced and covering the paper's Initially she drew directly onto paper and then embarked on
surface. Joginder had difficulty holding the paintbrush in her a series of mono prints by applying the oil bars onto glass,
left hand, clutching it firmly in an awkward clenched grip, spraying a fine mist of turps onto the image and then
which restricted the range of movements she was able to printing it onto paper.
employ. Despite a water pot beside her, she made no attempt The process is relatively simple yet the results can be
to dip the brush into the water before or after applying the spectacular and Joginder positively delighted in this. When
paint to the brush and painted until the brush was dry before offered paint and brushes again she flatly refused and reached
applying more paint. Other brushes in varying sizes were out for the oil bars indignantly. It was strangely impressive that
placed beside her, but once she had made her initial choice she had become so self-assured and assertive in such a relatively
the others were ignored. Similarly with the paint selection: short time. (From: Hall (1996), in Gillham, Bill (ed.) The
throughout the first few workshops Joginder continued to Challenge oj Age, Glasgow: The Foulis Press, pp. 28-9.)
paint using only one brush and one colour per painting
although she did vary the colour selection from one work to
another. This account has been quoted at moderate length because it
I t was a basic but very encouraging beginning as she conveys the quality .of careful observation and description
50 51
Case Study Research Methods Observation: Looking and Listening
and the power of the evidenced argument of which it forms a Having made that simple oppositional comparison we can
part. It is more convincing than generalizations.and rhetoric. go on to look at the methods in more detail.
52 53
Case Study Research Methods Observation: Looking and Listening
more detail, together with provisional explanations that seem to behaviours: and you need to be clear exactly what it is you are
fit (the inductive method). observing. For example:
The maintenance offield notes is essential. We've referred to
these on p. 24, but key points are worth reiterating and • in a day nursery how long babies are left crying before
someone attends to them;
amplifying here. You include:
• how many children are 'isolated' - on their own - in a
• running descriptions: your basic material; school playground;
• things you remembered later (your notebook should be • how often workers on a shop floor disregard machine
always to hand); safety procedures;
• ideas and provisional explanations (their emergent char- • how long people have to wait in an Accident and
acter is a process not to be lost); Emergency department;
• personal impressions and feelings - even if you can't • how many people stop to look at a particular painting in
explain them: they may be the first hints of more a gallery.
important things;
Note that all these involve some kind of counting or measur-
• things to check up or find out about - star or highlight
ing - of time in relation to behaviour; of particular be-
these so that you don't overlook them. haviours.
As mentioned earlier you 'write up' your rough notes - Structured observation is about sampling these behaviours,
without tidying the content - promptly and regularly. This and there are two mam approaches to it. First, interval
has the effect of running them through your mind again. It is sampling: taking a look every so often - perhaps only for a
also good practice to read through your recent notes before moment (how many people are looking at this painting
you next go on site: this will help you to focus and plan your NOW? how many children are playing on their own
time. NOW?). Second, event sampling: a form of continuous
This is also a process where you should be challenging observation where you note how often or when things
yourself. Am I going off course? Have I got the wrong end of happen (the baby has started crying - how long before
the stick? Do my aims and research questions still stand? someone comes to comfort her? how often have workers fed
What do I need to be doing differently? ma-terial into the machine without putting up the hand
guard? and so on).
You use interval sampling when the behaviours happen at
high frequency so that continuous observation is unnecessary
Structured observation: data collection to achieve a representative picture. You can fix the intervals
so that your samples are quite frequent (every five minutes
This is a technique to be used sparingly: it takes time in counting the number of isolated children). If the frequency is
planning, is very time-consuming, and yields limited infor- moderate, e.g. people stopping to look at a picture in a
mation. But this information can be highly specific on key gallery, you may do your count for a five-minute period
points of evidence. every half-hour (which means you could target other pic-
It is behavioural in the sense that you focus on specific tures for the same period/interval) in between times.
54 55
Case Study Research Methods Observation: Looking and Listening
If behaviours are very low frequency (e.g. how many tions would be particularly useful. You could do this by
disruptive incidents - however defined - in this class with counting the number of isolated children at five-minute
this teacher?) then you should use 'event' sampling. Simi- intervals - perhaps three times on each occasion. So over a
larly if the events are causally related over time then you period of a week you would have five sets of observations.
would observe all the time, recording the events as they The actual schedule might look like this:
occurred (patient reporting to reception in Accident and
Emergency; patient called for medical attention). Date: 2.3.00 Observer: E.G.
There is considerable flexibility here and you can see Time: 10.40-11.00 Interval: 5 minutes
that the interval/event distinctions can become blurred.
Remember that the method is there to serve the purposes Observation No. of children isolated,
of the research. 'Method' is not some sacred cow to be i.e. not in social play with others
worshipped regardless.
I .4-I-t-f .4-I-t-f
The essential point is that the picture that emerges should
be representative: typical of the total reality. You need to get
2 .4-I-t-f I
enough data from sufficient observations to be reasonably
3 .4-I-t-f I I I
sure of that.
Two things you need to be organized on: There are several steps in getting up to this point:
• a clear specification of what is to be observed (and a clear 1. Unstructured observation. This maybe when the need for
grasp of why it's important); structured, focused observation became apparent. You
• a clear procedure for recording your observations. get a 'feel' of how it might be possible to observe in a more
structured way.
2. Specifying what is to be observed. This is something you need
Preparing an observation schedule to get precisely clear in your own mind. In the case of
'socially isolated' children how exactly would you define
You don't sit down and write out an observation schedule that? Distance from other children? No apparent com-
right out of your head, any more than you just sit down and munication or interaction? Both? It's not a matter of what
write out a questionnaire. By 'don't' is meant 'shouldn't' is correct but of you being consistent. Observation recording
because, of course, people often do exactly that: with is impossible if you have to do too much making your
disastrous results which they richly deserve. mind up on the spot.
So that we're not talking in abstractions let us look at an 3. Finding a good vantage point. Can you see all you need to see
actual schedule. adequately? Too far? Too close? Visibility obscured in
If you are involved in a project which aims to increase some way?
opportunities for social play in a primary school break-time, 4. Is the schedule workable? Have you given yourself an
then observation of the numbers of children on their own impossible task? Intervals too frequent or not frequent
would be one kind of evidence. A 'base-line' set of observa- enough? Too much to record at a time?
56 57
Case Study Research Methods
58 59
Interviewing·
are less demanding than interviews, people are much less before you ask questions. And your first questions should be
likely to respond to them - although you can improve on of a naturally occurring kind - referring to the table on page
this (see Developing a Questionnaire). 62, those you can bring into ordinary conversation, which
5. If the material is not particularly subtle or sensitive. can be a research tool of some importance.
However, even apparently straightforward question-
naires can easily be 'misunderstood' - and you can't
correct that; a great strength of interviews is that you Using naturally occurring conversation
can pick up these nuances which are often quite subtle.
As you get to know the setting, and focus your aims and
The overwhelming strength of the face-to-face interview is research questions, you will begin to see what you have to
the 'richness' of the communication that is possible. Ques- find out, and what will best be answered by asking questions;
tionnaireda ta in particular can appear (and usually are) and at a slightly later stage what will best be answered in an
thin, abstract and superficial. The richness comes at a price, interview setting.
of course. It isn't just the time you give to the interview itself, But you can ask questions systematically without setting
it is the time involved in transcription and analysis - a factor up an interview.
of about ten at least is involved here. If you are 'participant' in the setting you can decide on a
In summary you use interview techniques when: small number of questions you want answers to - and ask one
1. Small numbers of people are involved. or two of them of people as the opportunity naturally arises.
2. They are accessible. The people in the setting will know your purpose is one of
3. They are 'key' and you can't afford to lose any. research enquiry so they will expect you to ask questions
4. Your questions (or the most significant ones) are mainly (and find that acceptable once you've 'earned your place').
'open' and require an extended response with prompts And because they're not being formally interviewed they
and probes from you to clarify the answers. may give particularly revealing answers. You won't be
5. If the material is sensitive in character so that trust is recording them - but you should write down their responses
involved: people will disclose things in a face-to-face as soon as possible and as verbatim as possible. The research
interview that they will not disclose in an anonymous log again.
questionnaire.
63
62
I
Case Study Research Methods
1 Interviewing
researching. A hospital administrator, a head teacher, a :elati~nship wi~h this individual and, in any case, a follow-up
safety officer, a social work director, or whatever. Interviews mtervIew later m your investigation - when you have further
with such people are relatively-unstructured, for a number of points to raise - will be productive.
reasons, and have rather special characteristics.
1. They will know more about the topic and the setting than
you do: to a large extent they can tell you what questions The selDi-structured interview
you should be asking, what you need to know.
2. By virtue of their authority and experience they will have This is the most impor~ant form of interviewing in case study
their own structuring of their knowledge. They will not research. Well done, It can be the richest single source of
tamely submit to being interviewed where you direct a data.
series of questions at them. Its apparent simplicity is deceptive. If you are able to see
3. The best you can hope for is that you will raise topics that an experienced interviewer at work (live, or on video) it can
they will respond to. s~em almost 'natural': there is a pace, a fluency, a respon-
4. Where they can be particularly informative is where (and sIv<:ness that seems to have nothing of 'technique' about it.
what) documents or records are to be found; other people ThIS very flexibility is what makes the semi-structured inter-
you should particularly speak to; what you can and view such a productive research tool; and the 'naturalness'
cannot expect to be able to do. rests on ~ clear structure, carefully developed and practised.
5. They will expect to have some control over what you do, All skIlled performance is deceptiVely simple because the
and will usually demand a level of accountability and per ormer , h as subsumed technique - is, in ' fact, hardly
'f(
reporting back. If you can accept that, they, in return, aware of it.
can be important 'facilitators'. So you don't use an interview for your actual research
purposes until you are confident in its use. But, even more
The elite interview is something you will usually report
Important, you don't use it until you are clear about two
very fully in your write-up. This is partly because it is 'key' - things:
of central importance vis-a-vis other elements of your pattern
of evidence; partly because it will lead you in to different • what t~e key issues are in your research investigation;
areas of investigation (a direction which has to be acknowl- • what wIll best be answered in a face-to-face interview.
edged); and partly because it won't be susceptible to content
analysis according to a common framework as with other
'semi-structured'interviews (see below). The need for econolDY in interviewing
Of course, some elements of it will be more relevant or
important than others, and editing and summarizing will be Even one interview generates a huge amount of work for the
necessary, but there should also be fairly extensive direct researcher. As a simple rule-of-thumb, a one-hour interview
quotation, particularly of those elements that you want to (assuming you've tape-recorded it - recommended) is ten
cross-reference to other kinds of evidence. hours of transcription and almost as many hours of analysis.
Sometimes you will develop a kind of 'consultative' So you have to control the number of interviews and their
64
65
Case Study Research Methods
1 Interviewing
length: the latter is particularly important. With practice and • identifying key topics (you may have more than one
careful preparation you can get a gr.eat deal out of an question for some of these) ;
interview lasting no more than 30 minutes: but you need to • framing questions (around five to ten is about right);
be firmly, though unobtrusive4r, in control. • checking that these questions are genuinely open, i.e. that
You need to prune your list of question topics to those that they let the interviewee determine the answer and don't
are really essential for your research project and which cannot indicate a preferred answer;
be answered satisfactorily in any other way. The questions you ask • deciding on prompts: things you may need to remind the
will be open, i.e. where the answer is open. For example: interviewee about (e.g. 'what about financial support?');
'What do you think about teaching in an open-plan school?' • the use of probes: getting the interviewee to tell you more
The kind of answers you get to that sort of question are up to about a particular topic ('I'm not quite clear about that',
the person you're interviewing and largely unpredictable. etc.);
That's ~here the element of discovery comes in. You are • recording the interview (taking verbatim notes stalls the
asking the interviewee to tell you: and they may do so at whole thing and involves on-the-spot selection that may
some length, not all of it on the topic. Being able to move be doubtful; and writing up afterwards can also miss key
people on when they have said what is to the point is a key elements);
skill in interviewing ('That's very interesting. Another thing • keeping the thing moving: which means having all the above
I wanted to ask you was ... ', and so on). working efficiently.
We'll return to interviewing technique a little later.
A 'practice' interview
Interview preparation
If you are part of a research course it is of great value to
This is no quick process. There are three main elements: practise on your fellow students. What follows on p. 70 is an
• practising interviewing per se; interview outline used by the author with some of his
• developing and focusing the interview topics and ques- postgraduate students. It is largely self-explanatory.
tions; In your actual research interview the prompts will have
• rehearsing the actual research interview itself. been derived from other data and/or previous interviews, i.e .
pilots for the real thing. You will know that certain elements
You can practise interviewing as part of developing the have to come up. If a particular interviewee omits to
questions and format that you'll use in the actual research mention one you simply 'prompt' them by saying 'what
interview which you'll need to rehearse (or pilot) once about ... ?' . You're not leading them because you're not
you've got it into shape. determining or indicating the answer. What you are doing is
However, it's useful to practise interviewing 'off' your ensuring that all the interviews have comparable coverage:
topic, and an example of this is given below. Before we go on this is of central importance when you reach the stage rif content
to that we need to run over the main elements in organizing analysis.
an interview. These are: Prompts are impmtant, but so also are probes. In the same
66
67
Interviewing
68 69
1
Case Study Research Methods In terviewing
pause button and half a key interview is lost. You also need Transcription and content analysis
to make sure that you have spare batteries and, above all,
that the sound recording quality is good enough. There is a special Carrying out the interview and recording it is one thing;
kind of stress involved in trying to transcribe or analyse a transcribing and analysing it is quite another. Transcription
badly recorded interview. This 'technical' side is just one should be carried out as soon as possible after the actual
dimension of what has to be practised. interview: your memory will help you in hearing what is on
In an interview you are, critically, the research instru- the tape.
ment - a phrase we've used before but more generally. In You cannot analyse an interview just by listening to it,
an interview the fine detail of how you handle yourself is although you can get a feel and an impression, and certain
important. As part of your practice sessions with your peers parts will stand out. But transcription - writing it down -
you should arrange to be videotaped. Most people react is terribly time-consuming. It is at this point that the
with horror to the idea. It's a curiously emotional business, importance of moving the interview briskly along becomes
but you just have to steel yourself to it. In short you have to apparent.
learn to look at yourself in a detached manner. You will get You can't really study an interview's content except in
a lot of information which is not really accessible in any written form. But verbatim transcription brings home to you
other way. that much of what people say is redundant or repetitive.
Watch the playback on your own: there's a special kind of S The essence of content analysis is identifying substantive
ignominy in watching 'your' video as part of a group. ~ statements - statements that really say something. One short
Briefly you need to look out for these points: cut is to play the tape and stop it when one of these
statements come up. Even so you'll have to do a good deal
• How you come across visually - in particular what you of to-ing and fro-ing and what you will have written down is
are communicating non-verbally. Do you appear to be necessarily discontinuous. If you do this you should ask
listening? Do you appear interested? Are you attending to someone else to double-check by doing the same kind of
the non-verbal signals from the interviewee? stop-go analysis to see whether they identify essentially the
• Do you probe and prompt appropriately? same statements. You do something similar with a complete
• Do you explain the purpose and uses of the interview transcription - as a check on your judgements.
research adequately? In summary, this is how to do a content analysis (but see
• Do you move the interview along at the right pace? Chapter 8 in The Research Interview).
• Do you focus and direct the interviewee in a clear but
unobtrusive manner, i.e. have you control? 1. Take each transcript in turn.
• Do you close the interview in a socially acceptable way? 2. Go through each one highlighting substantive state-
ments (those that really make a point). Ignore repeti-
There is a lot here, and what is presented in this chapter on tions, digressions and other clearly irrelevant material.
this important technique is a bare summary, but sufficient 3. Some statements will be similar but if you feel they add
for you to appreciate the character and complexity of this something mark them up.
kind of interviewing. 4. Take a break. If you try to do transcripts one after
70 71
1
another your concentration will become dulled. Two a assign to any category. ModifY the wording of the
day, well spaced, is a maximum. But don't space them category headings (or revise them entirely) so that they
too much or you'll lose the categories that will be forming fit the statements better or can include 'query' state-
in your mind. ments. It may be that you will need to add new
5. When you've been through all the transcripts go back to categories. If there are a lot of 'query' statements then
the first one and read them through again. Are there any you should deal with them at a separate stage: too many
statements you've failed to highlight? Have you high- of these may indicate that your list of category headings
lighted some that aren't really 'substantive'? It may be is inadequate or that you have a lot of 'unique' state-
useful to ask someone else to go through a set of ments that necessarily resist classification: see below.
unmarked transcripts, highlighting what they see as 9. Enter your categories on an analysis grid like the one in
'substantive' statements as a check on your judgement. Figure 8.1. If you have a large number of categories for
Make any changes necessary. each question, make up a grid/spreadsheet for each of
6. Now comes the difficult, intellectually creative stage. them rather than for the transcripts as a whole. The
You go back to the beginning again (after an interval!) category headings go along the top, the names or codes
and, going through the highlighted statements, try to of the respondents down the side. If you make the
derive a set of categories for the responses to each analysis sheets A3 size (or even larger) you'll be able to
question. Give these a simple heading ('Safety training enter in the cells what the respondents actually said, or
procedures', 'Playground bullying', 'Written guidance part of it. This is very useful when you come to write up.
on prescription use', and so on). At this point all you are Category headings, remember, are simply a way of
trying to do is get a list of categories. You'll get a lot from classifying the kinds of statements people have made:
the first transcript, more from the next, but progressively they don't tell you much on their own.
fewer as you work through them all - because indivi- 10. Go through the transcripts, assigning each substantive
duals will be making essen tially similar points. Depend- statement (where possible) to a category. Statements
ing on the number of categories you are deriving, you you can't assign have to be dealt with separately:
may find it easier to go from one transcript to another 'unclassifiable' but not unimportant. Sometimes just
dealing with one question at a time. one individual makes a key point. Put the number of the
7. You then look at your list of categories and ask yourself category against the statement on the transcript: this
whether some of them could be combined or, alterna- tells you that you've entered it and where it's gone; if
tively, split up. As you are compiling the list you will you can't classify a statement mark it 'u.c.' (unclassifi-
sense that some of the headings you've noted down are able). On the analysis grid you can either tick the
not adequate or necessary. There is more work to be relevant box (this person made a statement which fits
done here. this category) or write in the actual statement or do both on
8. Go through the transcripts, with your list of categories separate sheets: one for a count analysis (how many
beside you. Check each substantive (highlighted) state- people said this kind of thing) and one for a meaning
ment against the category list to see if it has somewhere analysis. Sometimes a count analysis is all that is
to go. Mark'?' to those statements you cannot readily required, but tabulating the actual statements has a
72 73
Interviewing
0 """ ~ ~ ~
Interviewing
Case Study Research Methods
I
statistics which paints a graphic picture, but they sup- This excerpt makes clear why interviewing is a central
plemented this with an interview ~tudy of 31 s~ch ~others I technique in case study research (and an important sup-
plementary technique in survey research).
in an approximate 'quota' sample, I.e. an approxImatIon to a
national representation. How representative this was is open But it needs to be emphasized again that what is given
to question, but the inclusion here is because the report I here is only an introduction and should not be treated as a
illustrates the writing up of a content analysis, a sample of I comprehensive guide to practice. Aspects of interviewing
which are dealt with here in a page or two are given a
which is given in the box below. I chapter each in the parallel volume The Research Interview.
What follows is an even more summary review of other
interviewing techniques.
Experience of Illotherhood
Most of the single lone mothers said they had found
parenthood difficult. There were, however, nuances to the~r Telephone interviewing
experiences. Most notably, many also talked about theIr
positive enjoyment of bringing up children, so that for most Telephone interviewing - at one level or another - has come
of them motherhood was neither 'all good' or 'all bad'. to the fore in the past decade, largely because it offers some of
'I think you get a lot ofjoy out of having a child but . .. they are hard the virtues of the face-to-face interview, e.g. its responsive-
ness and reflexivity, but without the cost (in time and
work.'
money) of setting up individual meetings. It is widely used
' ... it's hard 'cause you haven't got anyone - although I've got close in market research, often 'cold', i.e. without preparation or
family . .. you've got to take on the role offather as well.'
the prior agreement of the interviewee. I t is extremely
For some, high expectations of motherhood had mostly been difficult to do even moderately well, but it has emerged as a
fulfilled: widely used research tool especially in the USA. The
'I loved being pregnant. I thought it was brilliant ... had this feeling technique of 'random digit dialling' can even be used to get
of being worthy of something and I just felt ... radiant all the time. a probability (random) sample of respondents. However,
And I was looking forward to having the baby . .. I couldn't wait for 'cold-calling' in a culture where telephone selling is often
this little thing to look after and love.' seen ~s a contemporary nuisance, can be a punishing
expenence.
But not so for others:
It works best in small-scale research either if you know the
'I didn't realise it would feel like it did . .. I was so tired . .. I wasn't respondents and arrange a telephone interview at a time that
prepared . .. I don't think anybody is.' suits them or if you telephone (or write) first to explain what
' ... it always seemed easy when it was somebody else's and you could the interview is about and arrange a convenient time for you
give it back . .. I just thought it'd be the same.' to call for the interview. It is, of course, perfectly feasible to
record telephone interviews, which enables you to focus on
From L. Burghes and M. Brown, Single Lone Motkers:
Problems, Prospects and Policies (1995), p. 50. responding to the person you are interviewing (keeping it
going is a particular difficulty) .
77
76
Case Study Research Methods Interviewing
Group interviewing are primary concerns - but they have their niche in case
studies.
This is one source of information, particularly useful for Questionnaires have to be filled in by the respondent
getting an early orientation on your research topic - asking without any assistance - which is why questionnaire design
simple open questions and then noting the range and kind of and development is so crucial.
responses you get. Issues of conflict or disagreement may Recording schedules are also questionnaires but are given
alert you to hidden complexities. face-to-face - the sort of thing used by the market researchers
Group dynamics, however, can be a powerful distorting who haunt our main shopping streets.
force. If there are marked differences in status within the Both have some value in case studies as a way of getting
group then those who are 'high status' will either dominate straightforward, fairly factual information. It may be that
proceedings or inhibit others. This information itself is, of factual data on employees in available records is either
course, of some value. But attention to group composition is incomplete or lacking in important respects (educational
important, e.g. are all people of more or less the same status?; qualifications, previous employment experience, for ex-
are women or men in a minority? Even if one 'equalizes', ample). These are factual matters most easily pulled in by a
some individuals will tend to dominate, though skilful very short questionnaire. The trouble with questionnaires is
'chairing' on your part will control this to some extent. Your that people often ignore them or don't complete them
attention and interest given to 'minority' members of the properly (even if they're 'simple' and 'obvious'). In other
group may act as a strong, encouraging signal. words, data quality or completeness suffers.
The group dynamic's potential for conflict is, in a sense, With a recording schedule you can pin people down. It
one of its strengths in that it may bring out tensions and can be time-consuming, but you get your data and you are
reveal groupings not apparent in an individual interview or able to deal with 'misunderstandings'. It can also give you a
the routine process of everyday. chance to gather some data on summary opinions and
judgements from a wider sample than you could hope to
involve in a semi-structured interview.
How useful they would be in any particular research
Questionnaires and recording schedules project depends on the value of the extra data they would
provide. It is not difficult to keep a recording schedule of,
These are at the structured end of the verbal-information- say, a dozen to twenty questions down to about ten minutes.
getting continuum and usually have a minor place in case Analysis is simpler because the structured presentation and
studies (ifthey are used at all). They are, however, central to the 'closed-questions' format mean that answers are stan-
the survey main method. Questionnaires are a much over- dardized. Closed questions are those where there is only one
used research technique because they are assumed to be easy answer or the choice of answers is given (Which of these
to construct. That is a fallacy (see Developing a Questionnaire in newspapers do you read?) - the opposite of the open-
this series). What is true is that they are easy to construct question format of the semi -structured interview.
badly.
Questionnaires are of little use if meaning and understanding
78 79
Quantitative Data in Case Study Research
been new initiatives, e.g. safety education for children, safety (a) , An 'effects' pattern
training for staff, it makes sense to ask whether there is any 8:<:,
~,
evidence that safety has improved: i.e. what inferences can <D'
£1
you draw?
All of these patterns reqUIre interpretation. Are they Figure 9.1 Time series analysis.
82 83
Case Study Research Methods Quantitative Data in Case Study Research
actually saying what they appear to say? In each case you Graph (b) - the apparent 'no effects' pattern - presents
need to challenge the surface explanation. Maybe the true another set of questions. Are we comparing like with like?
explanation is not the obvious one. The number of reported incidents remains the same and the
Graph (c) would appear to be the most baffling, so let us criteria for inclusion haven't changed. However, perhaps
deal with that first. Going back to our raw data - the actual when you sort accidents requiring medical attention from
incidents that were recorded as accidents - we may find those that don't you find a marked decline in the former -
that, post-intervention, teachers have been recording many masked by a modest increase in the latter; and perhaps here
trivial accidents that would not usually have been recorded teachers are being more scrupulous in recording incidents
before. In other words the criteria for inclusion have 'on the margin' of concern.
changed. Interventions commonly have side-effects of this Now these examples are hypothetical, but they illustrate
kind. It may be also that new legal requirements or the attitude of mind one has to bring to them.
professio~al procedures advised by unions have also had an Time series analysis can be used with successive observa-
effect (you may have picked these elements up from other tions (of the structured kind), usually where some changes
sources) . are going to be introduced after a period of 'base-line'
If the data are re-analysed maintaining the same criteria observations. But it may be that the technique is used
for inclusion as operated earlier, the rates are shown to have simply to display 'normal' variations over time (times of
declined. day, days of the week, etc.).
It is not being asserted that this is always the case, but
rather that you need to be alert to alternative, rather more
complex explanations. If you are not, this is where statistics Categorical analysis
may lie.
The researcher needs to be rigorous in the search for Quantitative data in case studies quite often come in
alternative or supplementary explanations for any apparent categories or can be put into them, usually to make compar-
consequence (or lack of it) in a time series analysis. It is isons between different groups (by age, gender, occupation,
rarely only the intervention that is new. This is where the ed uca tionallevel, etc.).
multiple sources of evidence in a case study may help to For example:
provide a more valid picture.
Let us take graph (a): an apparent 'effects' pattern where • visits to the GP in the past six months (none or one or
accident rates have dropped steadily. Did anything else more) according to age or gender;
happen at the time of the safety education and training? • employment or unemployment in school-Ieavers accord-
Could that be more important as a 'cause'? Perhaps super- ing to level of qualifications;
vision in the playground was increased? Perhaps play equip- • negative or positive statements about computer use by age
ment was improved? Perhaps break-times were staggered so or gender.
that the playground was less crowded? What do the people
most directly involved think contributed to the improve-
ment?
84 85
Case Study Research Methods Quantitative Data in Case Study Research
You can put these into a 2 x 2 matrix: e.g. circumstances of the men and women involved really
similar?
Visits to GP No visits For case study research operating in the real world,
quantitative data analysis has to be subjected to the scrutiny
M 8 20
of what it might mean - whether or not it is statistically
F 15 7 significant. Even if the numbers involved don't yield a
statistically significant difference, there may be important
Is there a significant difference here? By 'significant' is differences in reality and in more subtle human terms.
meant: how often could this apparent difference occur by Statistical significance is heavily dependent on the size of
chance? The statistic known as chi square enables you to put a the numbers involved: if the numbers in the table on p. 89
precise 'probability' value on the answer to this question - a were ten times bigger then differences would be much more
one in ten chance, or one in twenty, or whatever. Ifit could likely to be 'significant'. But a qualitative analysis of the
only have occurred, say, once in a hundred by chance, then interviews with the individuals concerned would probably
it would be considered very significant. Being able to make have demonstrated this in a different way.
this sort of statement can be used to complement the
qualitative analysis of interview data, or sImple questIOnnaIre
data. The calculation of chLsquar:e is described in a chapter
in Developing a Q,uestionnaire, or will be found in any standard
statistical text.
Quantitative data has a special place in case study
research in so far as it extends the range of evidence on the
topics under investigation - and qualifies what we have
learnt from other sources. This kind of cross-referencing is
part of the internal validity of a case study: it all has to fit
together - and theorizing (explanation) has to account for all
of it.
Whatever kind of statistics we use (descriptive or inferen-
tial) the special rigour we have to bring to the outcome
(using other evidence) is to ask: does it say what it appears to
say? what might lie behind it? what other explanations are
there?
In the 2 x 2 table above, which could be subjected to chi
square, it looks as if there were a real difference between
men and women and GP visits; is the difference 'signifi-
cant'? Even if it is we need to ask: why? what lies behind it?
In particular, are we comparing like with like? Are the
86 87
Physical Artefacts
10 report there? Skill in writing helps, but words are often not
adequate: they can guide and can interpret but are often a
poor substitute for the physical reality. Indeed there is a clue
to their limitations here: that they can get in the way.
Ph ysical Artefacts Physical evidence has a direct quality: it is first-hand. You
can read or listen to what others have to say about it but the
actual thing is there for you to make up your own mind.
The visual dimension is uniquely powerful. In research it
can bring your report to life - enable people to 'see' in the
cognitive as well as the visual sense.
This is all very well but how can you get what may be
'Physical artefacts' is a rather clumsy term referring to large or bulky or heavy objects into a research report?
anything that is made: and in variants of case studies this Photographs, of good quality, are a minimum. But objects
kind of evidence may be the most important of all. But even and materials can be stored (and their accessibility referred
in case studies where this is not primary (and perhaps not to). In some contexts (for example in art and design) an
even considered) it is a category to be conscious of. To the exhibition might be linked to the research report. You have
real-world researcher all evidence is of some potential value to break out of the bounds of what is conventionally seen as a
ifit has a bearing on the aims of the project and the questions way of evidencing your research. The overriding question
being asked. has to be: how adequate is the report to the range and
There is an assumption that research evidence (and its { quality of evidence that is relevant?
presentation) has to be in written form. Apart from summary For example, in the chapter dealing with observation we
'figures' of one kind or another, papers in research journals talked about the value of video as a tool for observation
rarely include illustrations; they certainly don't include analysis but also as a way of providing primary evidence for
physical objects. others to see the raw material of your descriptive analysis.
But even if you can't include them in your write-up, Video can also be of value in presenting physical artefacts - it
physical objects may be part of the database you have to can move round them, 'look' at them from different angles,
maintain. And good quality photographs of these at least show them in use, and so on. Video is of special value when
need to be included in your report. you are researching the process of making objects, where a
Some kinds of evidence cannot be described or measured, static illustration cannot communicate - even with many
only shown. There is an exact analogy here in the judicial successive illustrations - the activity of making and creating.
process. In courts oflaw evidence is often of physical objects in
considerable variety. More than that, judges will sometimes
insist that jurors go to visit the physical setting of the crime as Physical artefacts as prim.ary evidence
an aid to fuller understanding and better judgement.
Real-world researchers may find themselves in an exactly We've already talked about the 'primacy' of physical evi-
analogous position. How can they take the 'reader' of their dence in the appeal of its directness to our senses. But it is
88 89
Case Study Research Methods Physical Artefacts
primary when the process of making is the essence of the make you hear, to make you feel. It is, before all, to make you
research project. A burgeoning area of research is that see. That - and no more. And it is everything.'
concerned with the design process. Conrad here meant 'see' in the sense of gaining insight into
There are different kinds of design research (for example, the motives and actions of his characters. And written
of users trying out prototype products) but a fundamental language is uniquely powerful in that. But seeing in the
one is to follow a designer through from first conception of a visual sense gives you a different kind of insight.
design idea to finished product. You end up with something For example, Sarah Hall's account of the developing
attractive, novel and useful (we assume): it has a polished artistic skill of an elderly stroke patient is quoted in the
perfection. But a designer has to work through from first chapter on observation (pp. 52-3). This is a good example of
beginnings - at least, if it is a product of particular origin- a detailed account of what the writer had witnessed. But a
ality. How did they get there? The process has been largely fuller appreciation of what this elderly woman was about can
neglected because it's the finished product that counts. But only be achieved by seeing the works of art she produced.
any understanding of design has to be rooted in what The exigencies of economic book production preclude illus-
designers do. Moreover, if you are training designers, being trations in the text; but if you could see one of Joginder's
able to give examples of the stages in the process (including, wall-hangings you would read Hall's account with new eyes.
most importantly, the dead-ends and the things that go You would see more, and understand better.
wrong) can be helpful, and encouraging to intending Most artefacts won't have quite that dramatic quality; nor
designers who lack confidence in achieving the sophisticated will most cases have that kind of emotional appeal. But case
outcomes they see emerging from the workshops and offices studies quite often focus on exceptional groups or indivi-
of practising designers. duals.
An influential use of case study process analysis is when the Supposing, for example, your case study is of students with
researcher researches their own creative activity. Such a severe literacy difficulties in a secondary school. You can
researcher is in a privileged position: for one thing they're collate schools records; you can describe the setting; analyse
'on hand' at all times. He or she also has privileged access to the curriculum demands; you can interview the students
thoughts, insights, mental 'discoveries' which an external themselves. But samples of work the students produced,
researcher could only achieve with difficulty - through unaided, would add to the reader's understanding, a better
interviewing, asking the designer to keep diaries, etc. It is appreciation of the problems these students might have in
equally valuable as a learning process, making young coping with curriculum demands and the requirements of
designers more reflective and more aware of what is the employment market.
involved. In the write-up of a case study (which is dealt with in the
next chapter) these examples or illustrations are often con-
signed to an appendix, so that they don't interrupt the flow
Enabling the reader to 'see' of the text. In quantity, and together, that makes sense. But
they may be very much part of the narrative, even though
In the introduction to one of his books, Joseph Conrad they are not textual. Illustrations and samples are as much
wrote: 'My task is, by the power of the written word, to part of the chain of evidence as material that can be formed
90 91
Case Study Research Methods
92 93
Case Study Research Methods Analysing and Presenting Your Findings
Writing a case study research report is a demanding task reminding yourself of material that is partly forgotten, you
because of: are also making new discoveries in your understanding: you
may have been partly conscious of these but the reviewing
• the variety of different kinds of evidence obtained in process brings them into sharp focus so that the whole
different ways; pattern shifts - sometimes radically.
• the skill required in weaving this evidence into a coherent Research data analysis is an absorbing, demanding pro-
narrative; cess requiring clear, undisturbed time. It helps to set out
• the need to maintain the focus and direction determined your different kinds of data in piles on a table. There is
by the overall aims and the specific research questions; something about that simple device of external organization
• the need to plot the successive revisions of the explana- which helps your internal organization.
tions or 'theories'. You move from one form of evidence to another, reading,
studying and thinking. What you are looking for, in parti-
cular, are different kinds of evidence bearing on the same
The groundwork issues in your research. These multiple sources of evidence-
which have to be related to each other - have to be woven
It is here that the task of writing up is made more difficult by into your narrative which itself represents what Yin calls a
a lack of progressive organization in the collation of the 'chain of evidence'. And alongside that chain of evidence
research evidence and developments in the research design comes your interpretation of it.
and theorizing. The task requires a grasp and degree of
concentration which stretches one's capacity. As Yin (1989)
rightly remarks, case study research, traditionally seen as The structure of the research report
'soft', is extremely hard to do well; and that extends to the
writing-up stage. The structure of a traditional natural-sciences style journal
Your provisional, interim, summaries and your log book paper follows a long~established format. A review of the
will be a great help at this stage. Essentially what you have to literature from which issues that require further research are
do is to review all the evidence, and your procedures. You drawn; a clear specification of the research questions them-
will necessarily have developed an overall picture of what selves; the methodology used to answer the questions; the
you've found and what it means. But parts of it will be hazy results obtained; a discussion and analysis of these; the
(and will largely remain so until you start writing). The theoretical gains or developments appraised for their short-
weakness of this accumulated mental overview is that it will comings; modification of theory suggested; the need for
be selective and, in some respects, superficial. further research outlined. There is a beautiful logic to this
You read your summaries, you carefully sift through the deductive, hypothesis-testing model and it is easy to under-
contents of your log book; and you review all ofthe evidence stand its intellectual appeal.
you have accumulated. Hidden from view, however, is the less tidy, more intui-
This detached, reflective overview is a key stage in the tive, trial-and-error reality of intellectual discovery. The
development of your study. In carrying it out you are not just logic of the internal st~ucture of a research paper does not
94 95
Case Study Research Methods Analysing and Presenting Your Findings
usually reflect the untidy chronology of how it emerged. In a failure a factor? If so, why does it affect some boys and not
sense, part of the creative process, and meaning of the others? And so on. Theorizing drives your research ques-
discovery is lost. Indeed, it is often only in the autobiogra- tions, and both act to structure your narrative.
phies of scientists that you learn how discoveries really take These structural elements of a naturalistic, case study
place. The scientific papers are not 'untrue', but they are not report are a challenge to one's intellectual capacity: in
adequate as a representation of how knowledge is achieved. particular to one's writing ability.
The naturalistic researcher is more concerned to give an
account of the reality of the research process. This has its own
structure which has broad similarities to a traditional
The business of writing
research report but has a quite different quality. How is this
structure characterized? What are its essential components?
Thought is not merely expressed In words; it comes into
First; chronology: the order in which things happened. This existence through them.
is not quite the same as the logic of the research process: you Vygotsky, Speech and Thought
will discover evidence that relates back to discoveries you
made earlier; later insights will cause you to revise your Most of what we say and do is not necessary, and its omission
understanding of what happened before. But in your write- would save both time and trouble. At every step, therefore, a
man should ask himself, 'Is this one of the things that are
up you can switch back and forth to link these up.
superfluous?' .
Second, logical coherence: chronology is not always going to
Marcus Aurelius, Meditations
be adequate as a way of presenting common themes or issues.
You may need to bring these together and lay them side by Surprisingly little attention is given in research methods texts
side: especially important when you are cross-referring or to the adual business of writing. There will be guidance on
transplanting different sources of data on the same issue. conventions and organization but almost nothing on how
Third, the aim of your research: which acts as a kind of one writes or how this process interacts with one's thinking.
retrospective effect on the structure. That's where you're Good writing is not just a matter of having a style or
going and that direction has to be kept in sight, whatever avoiding jargon. Writing up a case study report is a skill of
digressions may arise. high order: a reduction to sufficiently detailed essentials for
Fourth, your research questions: the development of these, the reader to follow the reasoning-from-evidence process.
in response to your increasingly clear grasp of the issues, is a This empirical stance (where evidence is primary and inter-
continuous strand. These are the sub-plots of your narrative, pretation is firmly grounded in it) can result in an accessible
the answers to which will (we hope) allow you to achieve account, where logic and meaning are transparent. This very
your overall aim. transparency meets an essential requirement of research
Fifth, your emergent theorizing or explanation of the issues reporting; not just that one can follow the reasons and
you are dealing with. This is what gives meaning or under- reasoning but that one can also see the limits: what could
standing to what you are about. It is not enough to be have been done differently and what should be done addi-
descriptive, you have to be able to explain what you find. tionally.
Why do boys truant more often than girls? Is academic And it isn't just the external reader that sees this, but also
96 97
Case Study Research Methods Analysing and Presenting Your Findings
the writer. When you are preparing to write, when you have starting again - this assists in the carryover, which will
reviewed your material, there comes a point where you have be going on, half-consciously, in between.
to make a start on getting it down-on paper. There is a point 3. Write in the way where the actual mechanics of getting
of discomfort where you can't contain it in your head any things down is least obtrusive - in longhand if your
longer. In a sense, your existing mental grasp is no longer word-processing is not fluent.
adequate. 4. At the weekend, assuming you've written longhand,
But when you do start to write, apart from easing the word-process your week's output.
mental burden, freeing up mental space, so to speak, you 5. Go through this manually, improving the continuity
discover an additional benefit. And that is that you are able and expression, making insertions where you feel you've
to think about the products of your thinking. You will missed something out.
immediately perceive some inadequacies there; but you will 6. Word-process your revision and print out a 'clean' hard
also find that simply putting your thoughts into words acts copy.
back on your thinking to improve it. 7. Maintain this kind of discipline until you have a com-
This clarification process is perhaps the most important plete first draft.
function of writing for the scholar. It harks back to the 8. Then leave it alone for a fortnight - put it out of your
'visibility' factor cited by Conrad and quoted earlier. Visual conscious mind as far as possible, i.e. dealing with the
metaphors are commonly applied to good or bad writing: other things you've had to neglect.
clarity, opacity, obscurity. Clear writing equates clear 9. This interval is to allow for the 'rotting down' process:
thinking, and that is what you have to work towards. It giving the unconscious time to work.
isn't just a matter of whether it reads well; fluent writing 10. When you return to your draft you will find that a lot of
can have a glibness which means that you glide over the work has been going on. The need for alterations and
surface, engaging with nothing. additions will show up as 'obvious'. Having made these,
Words can run away with you. Good writing is clear prepare a new draft.
writing where the reader can see your thinking, and that 11. At this point give it to someone else to read and write
means stripping away non-essentials. comments on: not someone specialist in the topic you are
writing on but someone who can appraise your writing
and put detailed comments on your manuscript (most
usefully: 'I don't understand this').
The specifics of writing
12. Getting your wri ting clearer shows up the deficiencies in
Writing a research report requires a disciplined momentum, your reasoning. It is at this point that 'specialist' feed-
back - perhaps from a supervisor - is appropriate: when
and a high level of concentration. It is a process not to be
your evidence and argument are clear enough for the
interrupted, if at all possible: clear (mental) space is
inadequacies to be apparent.
required. Here is a basic procedure:
1. Write a regular amount each day (1,000-1,500 words). By now you are almost there, although you will go on with
2. Read through the previous day's production before fine-tuning for a while. However, this is mainly the stage
98 99
Case Study Research Methods
100 101
Case Study Research Methods
102 103
Index
lO4 105
Index
106