Middle Range Theory and The Study of Organizations Middle Range Theory and The Study of Organizations

MIDDLE RANGE THEORY
AND THE STUDY

OF ORGANIZATIONS
MIDDLE RANGE THEORY
AND THE STUDY
OF ORGANIZATIONS
Edited by:
CRAIG C. PINDER
LARRY F. MOORE
The University of British Columbia
GMartinus GJVijhoff Publishing

Boston/The Hague/London
ISBN-13: 978-94-009-8735-7 e-ISBN-13: 978-94-009-8733-3
DOl: 10.10007/978-94-009-8733-3
This book is dedicated with love to our wives, Pat and Margaret.
PREFACE
Late one afternoon in the fall of 1976, we were sipping Sanka and speculating
on the possible directions towards which research and theory in organizational
science might lead. One of us had just re-read Walter Nord's Marxist critique of
Human Resource Management, and the discussion evolved into an enumeration
of the many articles that had appeared in the recent literature attacking the
discipline, its mission, and its methods. In no time the list was long enough to
suggest that a number of scholars, both young and established, were dissatisfied
with the rate of progress begin made in the accumulation of knowledge about
organizations. The critics we identified were located at many different schools,
and they were associated with diverse research traditions and biases. The causes
they identified as underlying the problems they cited varied, as did the solutions
they offered. We decided to pursue these polemics with a view to seeking com-
monalities among them, hoping that if there were any dominant common
themes, it might be possible to anticipate the directions the field could take.
Our reading and thinking led us to the conclusion that many of the issues
being raised by the critics of the discipline could be seen as disagreements over
some implicit (or ignored) metaphysical and epistemological assumptions about
organizations. We hypothesized that much of the controversy resulted from a
lack of consensus regarding what organizations are and how knowledge about
them can be developed.
In an attempt to clarify our own perception of the situation, we wrote two
vii
viii PREFACE
papers summarizing our understanding of the problem and outlining our pro-
posal for its resolution. (Both of these papers appear in this volume.) The essence
of our analysis is that, while research and theory have been, in many ways,
dominated by the systems metaphor, different observers see, record, and write
about different things when they study organizations. We believe that reliance on
single, overarching images and paradigms in organizational science is premature
at present, and probably dysfunctional. We agree with Karl Weick's 1974 sug-
gestion that scholars should entertain theories of the middle range when con-
sidering organizations, and we suspect that organizational science may never
achieve a situation in which one paradigm truly dominates and serves well.
To explore these issues, we called a conference to which we invited those
people whose attacks on the field seemed most fundamental and whose solutions
seemed most promising. We also invited one or two people who have reputations
for fostering creative exchange at conferences. For the most part, the invited
participants were requested to serve as discussants and panelists. Original papers
on the theme of middle range theories of organizational science were solicited in
journals and professional newsletters, and in two mass mailings of fliers to all
members of the Academy of Management. In response to our calls for papers, we
received forty-three submissions. From these we selected seventeen papers, to be
presented at the conference, held August 1-3,1978, at the University of British
Columbia in Vancouver. This book reports the proceedings of the conference.
All of the papers submitted were reviewed anonymously by at least two re-
viewers at other universities, as well as by ourselves. Most papers received three
or four reviews in total. Manuscripts submitted by our own University of British
Columbia colleagues were subjected to the same scrutiny given to outside papers.
The selected papers sorted themselves neatly into the five major topic areas
found in this volume.
Larry Moore wrote the first draft of the paper we had planned to use to open
the conference. By coincidence, Gary Johns (of Concordia University) submitted
a manuscript that touched on many of the themes we had planned to address.
Johns' paper was too good to ignore, so we incorporated it into our own and
asked him to help set the stage accordingly. He graciously consented to do so.
One of the papers in this book, "The Resurrection of Taxonomy to Aid the
Development of Middle Range Theories of Organizational Behavior," was not
actually presented at the conference. As indicated earlier, however, it was
written about the time the conference was being organized, and has subse-
quently been published in the Administrative Science Quarterly (March 1979).
We are grateful to Cornell University for permission to include it in this volume.
We have chosen to include it here because it is germane to the theme of the
conference and relates closely to the papers presented by John Bigelow and
Bill McKelvey.
PREFACE ix
Enterprises such as this conference may originate with two people over a cup
of decaffeinated coffee, but such enterprises require the devotion and assis-
tance of a large number of organizations and people. We wish to acknowledge
those who made the conference a success. Funding was provided by the Canada
Council (now the Social Sciences and Humanities Research Council of Canada),
The Leon and Thea Koerner Foundation, The Province of British Columbia, Im-
perial Oil Co. Ltd., McGraw-Hill Ryerson Co. Ltd., West Publishing Company,
and the Office of the President and the Faculty of Commerce and Business
Administration of The University of British Columbia. These sponsors provided
the necessary financial support to make the conference possible. We are es-
pecially grateful to Deans Peter Lusztig and Stanley Hamilton and President
Douglas Kenny of the University of British Columbia for their support.
Our manuscript review panel did a superb job. We are grateful to Mary-
Elizabeth Beres (Temple), Richard Daft (Queen's), Craig Lundberg (Oregon
State), Vance Mitchell (British Columbia), Donald Nightingale (Queen's),
Rodney Schneck (Alberta), and Barry Straw (Northwestern) for the time and
energy they spent helping to select the best papers from those submitted.
Considerable credit for the smooth organization of the conference itself goes
to Joan Harrison, who anticipated, dealt with, and subsequently told us about
countless problems (some of which we ourselves created) and considerations, in
a sensitive and efficient style that made us look as though we knew all along
what was happening. She made the conference fun and intellectually viable by
cheerfully mastering the myriad details that can interfere and spoil things.
Finally, we are grateful to our colleagues in the Industrial Relations Manage-
ment Division of the University of British Columbia Faculty of Commerce and
Business Administration, and to our wives (to whom the book is dedicated) for
the constant support and encouragement they provided us.
The book is organized in a structure parallel to that used at the conference.
In each section, three or four papers are presented, followed by the remarks of
one of the discussants and then the rejoinders of those authors who elected to
prepare rebuttals after the conference. At the end of the book are the remarks
of Les Roos, Gerald Salancik, and Karl Weick, who had been asked to provide
integrative summary statements based on all they had heard at the conference,
both on and off the record.
The order of editorship was determined by a toss of a coin. We are equally
responsible for the conference as well as the contents of this volume.
Craig C. Pinder
Larry F. Moore
Vancouver, B.C.
February 1979
CONTENTS
Preface xi
1. Toward Middle Range Theory: An Overview and Perspective 1

Larry F. Moore, University of British Columbia; Gary
Johns, Concordia University; and Craig C. Pinder,
University of British Columbia
I. WHAT IS MIDDLE RANGE THEORY? 17

2. Middle Range Theory and the Strategies of Theory
Construction 19
Allen C. Bluedorn, The Pennsylvania State University;
and Roger Evered, University of fllinois
3. Middle Range Theory: An Overview and Assessment for
Organizational Research 33
Paula C. Morrow and Paul M Muchinsky, Iowa State
University
4. Characteristics of Middle Range Organizational Theories and
Their Implications for Operationalization and Testing: A
Conceptual Analysis with Empirical illustrations 45
David P. Gilfillan, Temple University
xi
xii CONTENTS
5. Commentary on Gilfillan, Morrow-Muchinsky, and Bluedorn-

Evered: The Circle ofInquiry 61
Louis R. Pondy, University of Illinois
6. Metaphors, Theories, and the Processes of Scientific
Inquiry: A Reply to Pondy 72
II. WHY AND HOW DOES MIDDLE RANGE THEORY DEVELOP? 85

7. The Inevitability of Multiple Paradigms and the Resultant
Need for Middle Range Analysis in Organization Theory 87
Craig C. Pinder and Larry F. Moore, University of
British Columbia
8. Middle Range Theories of Organizational Behavior:
Some Implications for the Development of Theory 101
Philip Bobko, University of Maryland
9. Middle Range Theories: Clusters of Clusters of Organi-
zational Phenomenon 113
Randall S. Schuler, The Ohio State University
10. Nomothetical Nets and Higher Order Factor Analysis in
Middle Range Theory Development 127
Uma Sekaran, Thomas N. Martin, Richard S. Trafton,
and Richard Osborn, Southern Illinois University
at Carbondale
11. On the Potentialities of Middle Range Theory 140
J. Kenneth Benson, University of Missouri - Columbia
12. Reply to Benson's Comments in "On the Potentialities of
Middle Range Theory" 148
Randall S. Schuler, The Ohio State University
13. Zen and Science 151
III. GENERAL FRAMEWORKS FOR MIDDLE RANGE

THEORIZING 155
14. Evolution and Middle Range Theories: Toward a Matrix
of Organizational Modes 157
John D. Bigelow, Oregon State University
15. Organizational Speciation 169
William McKelvey, University of California, Los Angeles
CONTENTS xiii
16. The Resurrection of Taxonomy to Aid the Development of

Middle Range Theories of Organizational Behavior 187
Craig C. Pinder and Larry F. Moore, University of
British Columbia
17. Functionalism as a Base for Midrange Theory in Organiza-
tional Behavior/Theory 212
Orlando Behling, The Ohio State University
18. Organizational Implications of Exchange Theory: Is It
Time for a General Middle Range Theory? 225
Laurie Larwood, Claremont Men's College and
Claremont Graduate School
19. A Step out from the Middle: Thoughts Stimulated by
Papers of McKelvey, Bigelow, Behling, and Larwood 239
Walter Nord, Washington University
20. A Rejoinder to Nord 247
21. Some Other Parts of Exchange Theory 250
Qaremont Graduate School
IV. EXAMPLES OF MIDDLE RANGE THEORY 255

22. Middle Range Organization Theorizing: Role Theory
as an Example 257
Mary Elizabeth Beres and Karl F. Price, Temple
University
23. limited Domain Theories of Organizational Energy 273
John B. Miner, Georgia State University
24. Relationships between Modes of Social Interaction 287
E. Allen Slusher and Kenneth 1. Roering, University
of Missouri - Columbia
25. The Applicability of Middle Range Theories to the Study
of Organizational Effectiveness 304
Paul M Muchinsky and Paula C. Morrow, Iowa Stdte
University
26. On Research in Organizational Socialization: The Case for
Middle Range Theory 315
Daniel C. Feldman, Northwestern University
27. Examples of Middle Range Theory: Discussion 326
Thomas A. Mahoney, University of Minnesota
xiv CONTENTS
28. A Rationale for the Limited Domain Approach to the

Study of Motivation 334
V. COUNTERPOINTS AND ALTERNATIVES 337

29. Theory Development in Organization Behavior:
A Systems' Perspective 339
David Cooper, University of British Columbia and
University of Manchester; and Frank Walt,
30. Having One's Cake and Eating It Too: Middle Range
Content and Generalized Process as Ways of Understanding
Organization 354
Peter J. Frost and David C. Hayes, University of
British Columbia
31. Discussion of "Theory Development in Organization
Behavior" by Cooper and Wolf and "Having One's Cake
and Eating It Too" by Frost and Hayes 369
William McKelvey, University of California, Los Angeles
32. A Piece of Cake: A Response to McKelvey 379
David C. Hayes and Peter J. Frost, University of
British Columbia
VI. CONTEMPLATIVE PANEL DISCUSSION 383

33. Exploring Big Fierce Theories 385
Leslie L. Roos, Jr., University of Manitoba
34. Middle Range Themes in Organizational Theorizing 392
Karl E. Weick, Cornell University
35. Science as a Social Reality 408
Gerald R. Salancik, University of Illinois
MIDDLE RANGE THEORY
AND THE STUDY
OF ORGANIZATIONS
1 TOWARD MIDDLE RANGE
THEORY
An Overview and
Perspective
Larry F. Moore, University
of British Columbia; Gary Johns,
Concordia University; and Craig
C. Pinder, University of
British Columbia
This conference was called for the purpose of exploring the meaning of Merton's
(l968) concept of middle range theory, and for examining the relevance of the
concept for the scientific study of organizations and organizational behavior
(O.B.). This paper will attempt to set the stage for our deliberations by providing
a brief interpretation of the history of theoretical development in O.B., and by
highlighting the five major themes of the conference.
The distinguished sociologist Robert Merton (1968) has suggested that middle
range theories of social phenomena lie somewhere between casually generated
working hypotheses and grandiose attempts to explain all social behavior with a
given systematic set of assumptions. But middle range theories are not derived
from the grand theories, and they are more than summary statements of ob-
served relationships (empirical generalizations). As such, middle range theories
should provide for the efficient generation, consolidation, and evaluation of
empirical evidence and be liable to modification, refutation, and replacement.
More specifically, Merton (l968) indicates that such theories have the following
three major characteristics:
1. They do not attempt to deal with all social phenomena; rather, they are
each concerned with one or a few phenomena.
2 MIDDLE RANGE THEORY AND THE STUDY OF ORGANIZATIONS
2. They tend to be (ultimately) linkable to each other.

3. They are abstract enough to transcend simple description but concrete
enough to generate testable hypotheses.
Given the delimited subject matter of middle range theories, it follows that the
problems encountered and the methods used in generating data may be unique
to the particular theory in question. Using Stinchcombe's (1968) terminology,
middle range theory should consist of theoretical statements from which empiri-
cal statements can be logically deduced and tested by observations.
HISTORY OF THEORY AND RESEARCH IN

ORGANIZATIONAL BEHAVIOR
A brief historical summary of the development of O.B. helps explain the current
interest in middle range theorizing. Risking oversimplification, it can be argued
that our field has witnessed three historical phases:
1. A period of grand theory, ranging approximately from 1920 to 1950;

2. A period of search for empirical generalizations, from about 1940 to
1960;
3. A period of development of limited theory, beginning about 1960 and
continuing to the present.
Obviously, these stages do not represent distinct and independent eras, and it is
unlikely that their existence can be validated merely by counting the books and
journal pages devoted to each emphasis over the time period cited. Rather, each
stage represents an implicit orientation in the literature of a given period, and
each stage can be viewed as a partial reaction to the previous stages. Let us now
examine each period a little more closely.
The Grand Theory Period
Until around 1950, thinking about organizations was dominated by classical

organization theory with its three main sub streams: scientfic management
(Taylor, 1911), bureaucratic theory (Weber, 1947), and management principles
(Fayol, 1916); and by the human relations movement (cf. Roethlisberger and
Dickson, 1939). It is our position that these two main schools of thought can be
considered grand theories of organizational behavior for the following reasons:
TOWARD MIDDLE RANGE THEORY 3
1. They contained an implicit view of the basic nature of people, which can
be traced to certain other grand theories in philosophy and psychology.
Individual differences were downplayed.
2. They suggested that human energies can be directed toward organiza-
tional purposes by the application of universal principles.
3. They were stated at a level of abstraction such that it was logically impos-
sible to deduce hypotheses amenable to disproof.
The third characteristic of the classical and human relations theories is

especially crucial to our labelling them as grand theories. While grand theories
are sometimes stereotyped as theories without research, there is a fairly large
body of empirical work loosely associated with each of these theories. However,
the observations gathered to test these theories were not guided by logically
deduced networks of empirical statements (hypotheses). This was clearly ineffi-
cient and was in part responsible for the inability of the universal theories to
permit the deduction of hypotheses concerning potential moderators of the
main effects they proposed. Many other weaknesses of the classical and human
relations movements have been cited, including the failure to deal with mutual
dependency among variables, reliance on incomplete or erroneous assumptions,
time-boundedness, and logical inconsistency (Massie, 1965). Further, it might be
argued that these positions failed to account for important aspects of organiza-
tional reality, either considering organizations pretty much without people or
people without organizations. Finally, the classical and human relations theories
failed to meet all of the traditional standards of scientific theory, such as the use
of careful iteration between induction and deduction, replication, and so on.
However, at the broadest level, theories are simply statements of interrelation-
ships among concepts, and both the classical and human relations positions made
these types of statements. And they were convincing. There was enough intuitive
wisdom embodied in these theories to permit much evangelistic fervor and many
cookbook solutions based largely on childlike faith. Unfortunately, too much of
this continues to pervade our field.
The Search for Empirical Generalizations
The period from around 1940 to the early 1960s can reasonably accurately be
described as a period of research without much theory. During this time span
the empirical groundwork of the discipline we now call organizational behavior
was laid, with its emphasis upon individual and interpersonal behavior in an
organizational context. As Landsberger (1967) points out, researchers were so
busy generating observations, searching for generalizations in these observations,

and trying out more elegant designs and methods that they exhibited little inter-
est in theorizing. This period is important because it was a partial reaction to the
inadequacies of the grand theorizing of the previous stage, and because it pro-
vided an impetus for the subsequent interest in limited range theory develop-
ment. As Landsberger (1967) noted:
Obviously, a price has had to be paid occasionally for ... rigor in terms of
masses of uninterpretable data and factors, loss of imagination, etc. But errors
in this particular direction are presumably no worse for science than errors in
the opposite direction, i.e., too much unconfirmed impressionism.
It is probably safe to say that the widespread search for empirical generaliza-
tions derived fairly equally from academic reactions favorable to the grand
theories, academic discouragement with the grand theories, and pragmatic reac-
tions resulting from the drive to facilitate productivity during World War II.
The work of Likert (1961) would characterize the first reaction, the classic
studies by Gouldner (1954) and Blau (1955), which take issue with Weber's
bureaucratic formulations reflect the second reaction, and the studies of pro-
ductivity under piece rate pay plans (Viteles, 1953) are examples of the third.
But while the grand theories stimulated most of this research, they did not give
it much direction.
The research of the empirical generalization stage had several distinguishing
characteristics that resulted from the absence of logical guidelines provided by
carefully derived empirical statements. While these characteristics are not typical
of all research efforts of the period, and while they are not independent of each
other, they do provide a fair picture of the state of the art and help to explain
the subsequent interest in limited range theory development.
Empirical Shotgunning. There was a tendency to correlate a large number of

variables with a phenomenon of interest in hope of isolating predictors of the
phenomenon. Examples of this are the search for leadership traits among psy-
chological, physical, intellectual, and biographical characteristics (Gibb, 1954;
Stogdill, 1948) and studies of the correlates of tenure (Schuh, 1967).
The Assumption of Consequences. Researchers often made assumptions about

the effects of phenomena without actually measuring these effects. One example
was research demonstrating that managers and their subordinates disagree about
work-related matters (e.g., Likert, 1961) but failing to demonstrate empirically
the importance of such disagreement. Another example is found in the studies of
the importance of pay (Lawler, 1971), which failed to examine whether judg-
ments about pay affect behavior on the job. Such research was essentially de-
scriptive and lacked external validity. Much of the work of this nature was
stimulated by the human relations movement. As observed by Whyte (1965):
The survey researchers have been concerned with measuring attitudes and
perceptions and with relating these both to the demographic characteristics
of their informants and to "hard criterion variables" such as productivity,
absenteeism and turnover. They have not generally been concerned with re-
lating attitudes and perceptions, on the one hand, or hard criterion variables,
on the other, to the social processes in which their informants are involved
(p. 308).
Neglect of Moderators. Study after study correlated measures of job satisfac-

tion or morale with productivity, absence, or turnover (Brayfield and Crockett,
1955; Vroom, 1964) with little consideration for variables that might influence
the strength of these relationships. Hulin and Blood (1968) have discussed the
failure of early studies of the relationship between job scope and employee
responses to include potential moderators. A classic example of this problem
involves the plethora of research concerning the leader behaviors of considera-
tion and initiating structure. Although this research was carried out after per-
ceptive reviews of the trait approach suggested the importance of situational
moderators (Gibb, 1954; Stogdill, 1948), serious attention was seldom given to
the issue (Korman, 1966).
In summary, the flavor of much of the research carried out during the empiri-
cal generalization stage is nicely captured in Lawler's (1971) summary of the
work on perceptions of the importance of pay: "It is an excellent example of
what often happens when research is not guided by a desire to develop or test
some kind of theoretical statement or model. Most of the research is fragmented,
noncumulative, and poorly designed .... [It represents] a great expenditure of
effort that contributes virtually nothing to our understanding (p. 59)."
For the sake of comparison and historical accuracy, it is worth noting that
there was research during this period, which stemmed from middle range theory
and had relevance for O.B. (In social psychology, theories of conformity, co-
hesiveness, reference groups, and cognitive dissonance efficiently guided the
collection of much useful information (Merton, 1968), but this work did not
usually have as its central focus the study of organizations, per se).
Recent Trends in Theory Development
By the 1960s there were two almost separate approaches to the study of organi-
zation, corresponding closely to developments in the two major parent disci-
plines: psychology and sociology.
The Individual Behavior Approach. On the psychological side, there were signs
of increasing disenchantment, both with the universally prescriptive but empiri-
cally unsupportable theory of the principles era and with the simplistic and
disjointed attempts at empirical generalization characterizing the 1950s (Strauss,
1970). For example, serious doubts were cast about the universality of Mc-
Gregor's Theory Y assumptions; Herzberg et al., (1960) challenged the prevailing
assumptions about job satisfaction, pay, and performance; and Vroom's (1959,
1964) literature review and research demonstrated the importance of moderating
variables in the relationship between leadership style and job attitudes.
Prompted by the large-scale infusion of methodological rigor in O.B. by young
researchers possessing the academic training of psychology, and the wider avail-
ability of more sophisticated and easy to use computer programs, the search was
on for more complete, if complex, explanations of individual behavior in the
work setting. As a result of the research generated, a host of multivariate models
dealing with aspects of motivation (porter and Lawler, 1968), leadership (Fiedler,
1967), job design (Hackman and Oldham, 1975), and most other individual vari-
ables of interest to the field have been developed.
The Complex Organization Approach. A second major research thrust was

emerging in the early 1960s among sociologists whose focus of attention was
directed more upon the characteristics of organizations and their contexts,
rather than upon the behavior of individuals. The work of Blau and his col-
leagues (cf. Blau, 1955; Blau et aI., 1966; Blau and Schoenherr, 1971), Wood-
ward (1965), and Lawrence and Lorsch (1967) illustrate this vein of study. The
complex interdependencies revealed in these studies contributed to the popularity
of the systems approach to the field (Katz and Kahn, 1966; Thompson, 1967).
The Contingency Way of Thinking. Growing out of both the individual behav-
ior and the complex organization branches of research, one can see a rejection of
the search for grand theory in favor of theory building on more modest dimen-
sions. Students of individual behavior have developed several so-calJed contin-
gency theories, e.g., Vroom and Yetton's (1973) work on decision making, the
Fiedler (1967) theory of leadership, and the Hackman-Oldham (1975) theory of
job characteristics. Each involves a fairly specific organizational process (e.g.,
decision making, leadership, motivation) or the relationship between a limited
set of specific, distinctive variables (e.g., leadership and performance, job charac-
teristics, and employee responses). Each of these examples has generated testable
hypotheses and has encouraged systematic data collection. At the complex
organization level several studies have been designed to reveal or to test situation-
dependent or contingent relationships. For example, Woodward (1965) demon-
strated contingent relationships between technological, structural, and managerial
variables and performance, while Lawrence and Lorsch (1967) and Rage and
Aiken (1967) found several organization variables to be contingent on environ-
ment. Contingency models now pervade management textbooks, supplanting the
universalistic models of the past.
In spite of the recent popularity of contingency theorizing in organizations
and management, a number of serious problems still plague our field and are
curtailing its advancement.
CURRENT PROBLEMS IN ORGANIZATIONAL BEHAVIOR
Overemphasis on Hypothesis Testing at the Expense of

Hypothesis Construction
We previously mentioned that the past ten years have been characterized by a
large influx to faculties of administrative studies by researchers already trained
in foundation disciplines such as SOciology and psychology. As pointed out
recently by Lundberg (1976) and Bass (1974), this shift in composition of
scholars interested in O.B. problems has resulted in more emphasis being placed
on methodological rigor at the hypothesis testing stage than was previously the
case. We have become preoccupied with measurement precision, with the cre-
ation and internal validation of new instruments, and with trying out newly
available analytical models in the context of tight, formal designs based on hy-
potheses carefully derived from previous studies of a similar nature. White
(1978) provides an excellent illustration of such preoccupation in his review of
the investigations of individual difference moderators of the job characteristic-
employee response relationship. The danger in such research lies in the tendency
to extend intellectual exploration mainly along avenues that have previously
received attention-that is, by extending existing hypotheses and relying on
previously accepted assumptions. This approach, carried to the extreme, can
result in theory construction which, systematic though it may be, is narrowly
based and not capable of rendering accurate explanations and predictions of
reality. What must be sought is an approach to theory building that ensures
adequate emphasis on creative idea and hypothesis generation.
Worship of Moderator Variables at the Expense of

Dependent Variables
The recent use of limited range theories in specifying the operation of moderator
variables has led to an understandable increase in the interest devoted to contin-
gent relationships. This may be a good thing. However, the search for moderators
has not been accompanied by equally careful consideration of the outcomes or
dependent variables specified by the theories. This issue transcends the tradi-
tional criterion problem. It also extends beyond the obvious recommendation
that researchers predict and then demonstrate that variables account for more
criterion variance as moderators than as simple predictors (Zedeck, 1971). In
an incisive critique of the utility of systems theory as a guide for organizational
research, Weick (1974) has pointed out that the existence of simple direct causal
relationships may be obscured by inadequate attention to effects or outcomes.
By the same token, gross summary criterion measures may artifactually indicate
the operation of moderators. For example, if a particular measure is a reliable
indicator of voluntary absenteeism for males but not for females, sex may serve
as an apparent moderator of the relationship between job satisfaction and ab-
sence, when it is actually better conceived as a predictor of different types of
absence. Devoting attention to moderator variables at the expense of outcome
variables could lead to an inability to accomplish important linkages between
emerging theories as they develop.
A related problem, but not necessarily traceable to a concern with moder-
ators, is the failure of limited range theorists to specify how conceptually distinct
criteria are impacted by the process delineated in their theories. For instance,
there has been an implicit tendency to equate absence, turnover, satisfaction,
and performance as equally predictable effects, or to ignore certain outcomes
in the specification of a given theory. In reviewing the research on task design,
Pierce and Dunham (1976) have stated that investigators ... "have generally not
attempted to differentiate conceptually or empirically the role of the dependent
variables in associations with task design ... these conceptually distinct responses
should be treated as such (p. 94)." Thus, it is unclear whether researchers' con-
clusions that task characteristics predict affective responses better than they pre-
dict behavior indicates a theoretically relevant phenomenon of a problem of
measurement and design. In a similar vein, as originally proposed by Vroom
(1964), expectancy theory was designed to account for intra-individual decisions
both to participate in organizations and to produce in them. While it seems pos-
sible that an ongoing, continuous decision process (whether or not to work hard)
might differ from a one-shot, discontinuous decision (to join or to quit), such
concern with the nature of the dependent variable is not evident in the theory.
Thus, it would be difficult to decide whether differential success in predicting
production-related variables versus organizational entry or turnover would re-
flect upon the theory itself or the measurement properties of the criteria. To sug-
gest that this issue represents a simple empirical question is to revert to empirical
generalization and to forsake the guidance that should be provided by theory.
Disappointment over the Results of Present Research Efforts
The knowledge we have gained about organizational phenomena over the past
fifteen years has been far from commensurate with the energy expended. During
that time the incremental gain in knowledge about how organizations should be
structured and designed, how leaders should lead and motivate people, how jobs
should be designed, and how organizations should be evaluated in terms of effec-
tiveness, has been modest. Our sophisticated correlational designs typically pro-
duce omega squares around .10, leaving large amounts of variance unexplained
(cf. Dewar and Hage, 1978; White, 1978). What accounts for the inability to
explain the rest of the variance? Perhaps faulty measurement, poor sample selec-
tion, poor theory, or poor hypothesis construction, or maybe a combination of
these aspects are deficient - we rarely find out.
A second area of disappointment lies in the poor record of replication that
seems to characterize O. B. Far too much of the research conducted in our field
simply is not replicated at all, causing us to rely quite heavily on the theoretical
relationships established in one study or a set of closely related studies using a
common research design. Where replication has been attempted using other
methods (as in the case of the Herzberg and Fiedler models), the results have
proved inconsistent and have resulted in great confusion. One wonders just how
great this confusion might be if systematic replications of most of our other
major theoretical constructs were undertaken. Furthermore, we are tempted to
speCUlate with White (1978) that the real situation is probably worse than indi-
cated because studies reporting null results tend not to be published in the
journals.
The Loss of Parsimony
The increasing tendency of O.B. researchers to make use of multivariate methods

indicates more than ever that we should be concerned with standardization in
defining the variables we include in our hypotheses and in the methods by which
the variables may be appropriately measured. For example, how many measures
do we have of job satisfaction and how many ways do we defme organization
effectiveness? Our theoretical language seems full of fuzzy concepts and of simi-
lar but not identical definitions of terms, resulting in an inability to achieve par-
simony (Behling, 1978). We agree with Argyris (l976) that semantic clarity as
well as rigorous research methods are essential if behavioral knowledge is to
be advanced.
It is our view that the dominant mode of theory-building currently charac-

terizing the field of O.B., the contingency approach, requires modification. The
four problem areas cited above appear to be related to the pattern by which
contingency theory usually evolves. Typically a contingency theory is born
when, by accident, a previously generalized relationship (e.g., participation in
decision making and job performance) is empirically shown not to apply to all
situations. The model begins to be elaborated more or less in trial and error
fashion as moderators are sought and tested in various settings. It is precisely this
piecemeal mode of research that we believe leads to overemphasis on hypothesis
testing and the worship of moderator variables. The absence of an adequately
formulated theoretical foundation from the very beginning of the research effort
almost necessarily results in inadequate parsimony because there is not an over-
riding preoccupation with relating various research undertakings with an overall
theoretical framework. Concomittantly, because of lack of standardization of
variables, measures, and sample frames we should not be surprised when we
obtain inconsistent results and lack of replicability. Although we believe the
field of O.B. has made only limited progress in theory development when viewed
historically, the current focus on limited ranges of phenomena may be entirely
appropriate. We do, however, question the adequacy of the present contingency
approach.
As its title implies, this conference aims at an intensive exploration of various
facets of theory development in the middle range of Merton's (1968) continuum
with grand theory at one polar extreme and routine empirical hypothesis genera-
tion at the other. We hope that the output of the conference (its papers and dis-
cussion) will help foster a greater understanding of the role of theory in O.B.
and suggest avenues by which better theory may be developed.
OVERVIEW OF THE CONFERENCE
This section will identify five issue areas that seemed to materialize as the con-
ference committee examined the entire group of papers selected as a result of
the multiple-blind review process. Under each issue area, a few overview com-
ments will be made about the related papers and their orientations. Obviously,
much of the intellecuta1 richness of a conference grows out of the comments of
the discussants and the subsequent scholarly interchange that takes place. The
comments of each discussant follow the papers in each section, and the edited
transcription of the final contemplative panel session will be included as a sepa-
rate section in the proceedings.
What is Middle Range Theory?
Although this paper began with a summary of Merton's (1968) conception of

middle range theory in order to offer a backdrop against which we could frame
the historical development of organizational theory, we intentionally avoided a
complete definition and analysis of the concept. Thus, a number of questions
concerning the nature of middle range theory remain unanswered. For example,
how can a middle range theory be distinguished in precise fashion? Gilfillan's
paper, based on a literature survey, reveals that at least ten characteristics have
been proposed as components of the definition of middle range theory. Gilfillan
indicates that the ten characteristics may be used to assess the extent to which a
given theory may be considered middle range, but he notes some particular prob-
lems that may arise in testing a theory for these characteristics. Thus, for Gilfil-
lan, not every limited theory is a middle range theory.
Bluedorn and Evered examine the Merton definition of middle range theory
and conclude that middle rangeness is not an absolute but a relative quality (a
point also emphasized by Beres and Price and by Miner) and that middle range
theory construction is justified as a means to the end of creating theories of
larger scope. They then identify twelve general theory construction strategies
and raise questions about the necessity of a middle range imperative for success-
ful theory construction. Bobko argues that organizational research does not
necessarily progress towards a single unifying theory, contrary to common belief.
The paper by Morrow and Muchinsky on the other hand, after discussing sev-
eral advantages and limitations of the middle range approach to theory construc-
tion, offers some insights regarding how middle range theories may be integrated
or consolidated, in keeping with the assumption that middle range theories do
evolve toward higher levels of abstraction. Of course, from a purely logical
standpoint, as Bobko and Miner point out, all organization theories are middle
range theories since none can account for the infmity of variables impinging
onO.B.
Why Does Middle Range Theory Develop?
For Bobko, middle range theories arise because of different sets of assumptions
and unique sets of imagery about what constitutes the field of O.B., resulting
in a need for multiple explanations. Further, he believes there may be at least
as many ways of structuring middle range theories as there are theories, and
calls for humor and creativity as we strive to comprehend alternative, simul-
taneous images.
Pinder and Moore concur that multiple imagery is inevitable in the study of
organizations, basing their arguments on two different philosophical perspec-
tives: Realism and Phenomenalism. They also expect emerging middle range
theories to compete with one another and/or to be melded together as com-
munalities between theories are identified, but predict that general theory will
not be justified for some time to come.
The process of integration of theories is the subject of Schuler's paper. He
discusses how clusters of organizational phenomena might be recognized and
defined so as to suggest trans-middle range theories, which can in turn act as
integrators for other new or existing middle range theories.
The paper by Sekaran, Martin, Trafton, and Osborn tackles the problem of
clustering phenomena for theory building by illustrating that nomothetical nets
can be constructed using higher order factor analysis. Their empirical approach is
one that recognizes both the measurement of concepts and patterns of relation-
ships, and represents a step beyond the purely philosophical perspective regard-
ing the "how" of middle range theorizing.
General Frameworks for Middle Range Theorizing
Continuing with the notion of a multiple-paradigms perspective suggested by

Pinder and Moore, Bigelow explores the implications of an evolutionary per-
spective. He stresses the importance of time-related dynamics in understanding
how organizations develop, maintain, and transform themselves. A four-
dimensional matrix of middle range adaptive modes is presented.
Building on the biological theory of natural selection, McKelvey argues for an
evolutionary theory of classification based on the view that the prime cause of
organizational differences is isomorphic differences in the niches of the respec-
tive organizational forms. Several constraints are mentioned for a theory of
organizational speciation, the process by which different organizational forms
come into being. McKelvey outlines the major stages in the organizational spe-
ciation process, and he suggests that middle range theories tied to different
species of organizations need to be developed.
Behling suggests functionalism as an alternate base for attempting to describe
(theorize about?) organizational phenomena. Functionalism permits the develop-
ment of individual sets of constructs for analysis of phenomena, yet it is free
from several of the limitations of other frameworks (e.g., systems theory).
Behling illustrates his thinking with an example of functional analysis applied
to leadership.
The paper by Larwood addresses a problem in middle range theory also recog-
nized by several of the other authors: the need for an underlying theoretical
network or method of fostering integration of theories that may have developed

independently and/or at different time periods. For Larwood, such a network
must be flexible enough to integrate middle range theories without displacing
them. An exchange theory is proposed as providing such a network.
Examples of Middle Range Theory
The next series of papers either provides examples of what the authors perceive
to be middle range theories in operation or discusses some aspect of middle range
theorizing using a specific theory as a case-in·point illustration. Beres and Price
use role theory to focus on the scope and limits of middle range theory. They
also offer an interesting set of guidelines for developing middle range theory.
Miner elucidates his concept of middle range theory by presenting four lim·
ited domain theories of the processes by which organizations induce human
effort and create energy. Slusher and Roering use catastrophe models of social
interaction to illustrate their notion of middle range theory and to show how
linkages can be built between theories-again, a recurrent concern.
Muchinsky and Morrow explore the potential merit of using a middle range
approach to study organizational effectiveness. Their analysis indicates that
problems of global theoretical orientation, construct validity, and inductive ver-
sus deductive approaches to organizational effectiveness might be mitigated by
the construction of middle range theories.
Finally, Feldman reviews the nature and meaning of middle range theory,
then weighs the merits of middle range theorizing drawing on his own study of
socialization in organizations for illustrative material.
Counterpoints and Alternatives
Let's hear it for the systems model! Cooper and Wolf believe that recent discon-
tent with the state of theory development in O.B., which has led some scholars
to call for a change in paradigms, should be more appropriately viewed as a need
for further elaboration on the core of our existing paradigm (which they argue is
the system paradigm). Their view is that there may be no need for new para-
digms if we would strive for more careful and precise language in theory building
and if we would clearly specify the domain of organization theory and its in-
tended set of applications. An elaborate defense of system theory is presented to
demonstrate its applicability as the major theoretical paradigm for organizational
study and to call for more efforts to carefully elaborate this paradigm.
Finally, in keeping with the spirit of the conference, Frost and Hayes propose
a middle ground for middle range theory. Their paper argues for a dual approach
to the study of organizational phenomena. It is suggested that if the focal con-
cern of research is the content (substance perceived and created by participants),
then a middle range approach is appropriate. If, however, the focus is on the
process of organizing (the mechanics and associated behaviors through which or-
ganization takes place), then a general theory approach may be the best strategy.
The paper includes speculation on some potential constructs in a general theory
of organizing.
In conclusion, the diversity of thought expressed in the conference papers
promises to be highly conducive to the sort of cross-fertilization necessary to
enable us to understand better the potential role of middle range theorizing in
advancing the state of knowledge about organizations. We invite you to read on!
REFERENCES
Argyris, C. Problems and new directions for industrial psychology. In M.D. Dun-
nette (Ed.), Handbook of Industrial and Organizational Psychology. Chicago:
Rand McNally, 1976.
Bass, B.M. The substance and the shadow. American Psychologist, 1974 29,
870-886.
Behling, O. Some problems in the philosophy of science of organizations. Acad-
emy of Management Review, 1978,3, 193-201.
Blau, P.M. The dynamics of bureaucracy. Chicago: University of Chicago Press,
1955.
Blau, P.M.; Heydebrand, W.; and Stauffer, R. The structure of small bureaucra-
cies. American Sociological Review, 1966,31, 179-191.
Blau, P.M., and Schoenherr, R.A. The Structure of Organizations. New York:
Basic Books, 1971.
Brayfield, A. H., and Crockett, W. H. Employee attitudes and employee responses.
Psychological Bulletin, 1955,52,396-424.
Dewar, R., and Hage, J. Size, technology, complexity, and structural differentia-
tion: Toward a theoretical synthesis. Administrative Science Quarterly, 1978,
111-136.
Fayol, H. Industrial and General Administration. New York: Pitman, 1916.
Fiedler, F. A Theory of Leadership Effectiveness. New York: McGraw-Hill,
1967.
Gibb, C.A. Leadership. In G. Lindzey (Ed.), Handbook of Social Psychology
(Vol. 2). Cambridge, Mass.: Addison-Wesley, 1954.
Gouldner, A.W. Patterns of Industrial Bureaucracy. Glencoe, Ill.: Free Press,
1954.
Hackman, J.R., and Oldham, G.R. Development of the job diagnostic survey.
Journal of Applied Psychology, 1975,60, 159-170.
Hage, J., and Aiken, M. Relationship of centralization to other structural prop-

erties. Administrative Science Quarterly, 1967, 12, 72-92.
Herzberg, F.; Mausner, B.; and Snyderman, B. The Motivation to Work. New
York: Wiley, 1960.
Hickson, D.J.; Rinings, C.R.; Lee, C.A.; Schneck, R.E.; and Pennings, J.M. A
strategic contingencies theory of intraorganizational power. Administrative
Science Quarterly, 1971,16,216-229.
Hulin, C.L., and Blood, M.R. Job enlargement, individual differences, and
worker responses. Psychological Bulletin, 1968,69,41-55.
Katz, D., and Kahn, R.L. The Social Psychology of Organizations. New York:
Wiley, 1966.
Korman, A.K. Consideration, initiating structure, and organizational criteria-A
review. Personnel Psychology, 1966,19,349-361.
Landsberger, H. A. The behavioral sciences in industry. industrial Relations,
1967,7(1),1-19.
Lawler, E.E., III. Pay and Organizational Effectiveness: A Psychological View.
New York: McGraw-Hill, 1971.
Lawrence, P.R., and Lorsch, J. W. Organization and Environment: Managing Dif-
ferentiation and Integration. Cambridge: Harvard Graduate School of Busi-
ness Administration, 1967.
Likert, R. New Patterns of Management. New York: McGraw-Hill, 1961.
Lundberg, C.C. Hypothesis creation in organizational behavior research. Acad-
emy of Management Review, 1976,1(2),5-12.
Massie, J.L. Management theory. In J.G. March (Ed.), Handbook of Organiza-
tions. Chicago: Rand McNally, 1965.
Merton, R.K. Social Theory and Social Structure. New York: The Free Press,
1968.
Pierce, J.L., and Dunham, R.B. Task design: A literature review. Academy of
Management Review, 1976, 1, 83-97.
Porter, L., and Lawler, E.E., III. Managerial Attitudes and Performance. Home-
wood, Ill.: Irwin, 1968.
Pugh, D.S.; Hickson, D.J.; Hinings, C.R.; and Turner, C. The context of organi-
zation structures. Administrative Science Quarterly, 1969, 14, 91-114.
Roethlisberger, F.J., and Dickson, W.J. Management and The Worker. Cam-
bridge: Harvard University Press, 1939.
Schuh, A. The predictability of employee tenure: A review of the literature.
Personnel Psychology, 1967,20, 133-152.
Stinchcombe, A.L. Constructing social theories. New York: Harcourt, Brace, &
World, 1968.
Stogdill, R.M. Personal factors associated with leadership: A survey of the litera-
ture. Journal of Psychology, 1948,25,35-71.
Strauss, G. Organizational behavior and personnel relations. In A Review of
Industrial Relations Research (Vol. 1). Madison: Industrial Relations Research
Association, 1970.
Taylor, F.W. Tf.e Principles of Scientific Management. New York: Harper, 1911.
Thompson, J.D. Organizations in Action. New York: McGraw-Hill, 1967.
Turner, A.N., and Lawrence, P. Industrial Jobs and the Worker. Cambridge,
Mass.: Harvard Graduate School of Business Administration, 1965.
Viteles, M.S. Motivation and Morale in Industry. New York: Norton, 1953.
Vroom, V. Some personality determinants of the effects of participation. Jour-
nal of Abnormal Social Psychology, 1959,59,322-327.
- - . Work and Motivation. New York: Wiley, 1964.
Vroom, V., and Yetton, P. W. Leadership and Decision Making. Pittsburgh:
University of Pittsburgh Press, 1973.
Weber, M. Theory of Social and Economic Organization, tr. and ed. by A.M.
Henderson and T. Parsons. Oxford University Press, 1947.
Weick, K.E. Middle range theories of social systems. Behavioral Science, 1974,
19,357-367.
White, J.K. Individual differences and the job quality-worker response relation-
ship. Academy of Management Review, 1978,3, 263-280.
Whyte, W.F. A field in search of a focus. Industrial and Labor Relations Review,
1965,18,305-322.
Woodward, J. Industrial Organization. London: Oxford University Press, 1965.
Zedeck, S. Problems with the use of moderator variables. Psychological Bulletin,
1971,76,295-310.
I WHAT IS MIDDLE
RANGE THEORY?
2 MIDDLE RANGE THEORY
AND THE STRATEGIES OF
THEORY CONSTRUCTION
Allen C. Bluedorn, The Pennsylvania
State University; and Roger Evered,
University of Illinois
I am not one of those who hold that there is a scientific method as such. The
scientific method, as far as it is method, is nothing more than doing one's
damnedest with one's mind, no holds barred. -P.W. Bridgman (1950, p. 351)
In 1949 Robert Merton published the first edition of Social Theory and Social
Structure (STSS). This volume crystallized a theory construction imperative for
the social sciences: a strategy of building theories of the middle range (Merton,
1949, pp. 5-11). Although Merton (1967, 1968, pp. 56-58) would later trace
the call for middle range theories (MRTs) to such eminent philosophers as Plato,
Francis Bacon, John Stuart Mill, and Karl Mannheim, contemporary concern
about the middle range imperative is due almost entirely to Merton's discussion
in the three editions of STSS (1949,1957, and 1968V
Despite the discussion of MRT by many distinguished scholars, several ques-
tions about it still inhibit theorists who attempt to adopt the middle range
strategy. Two of these questions will be examined in the following analysis. First,
what is a middle range theory? Second, what role in the larger theory construc-
tion enterprise are MRTs intended to play? After these two questions have been
discussed, the final portion of this paper will examine the importance of socio-
logical and psychological factors on the choice of theory construction strategies
by individual theorists.
19
THE MEANING OF MIDDLE RANGE THEORY (MRT)
Merton has provided the most comprehensive defmition of MRT in his editions
of STSS. In the most recent edition (Merton, 1968), the following definition of
MRT is provided.
Throughout ... theories of the middle range: theories that lie between [italics
ours] the minor but necessary working hypotheses that evolve in abundance
during day-to-day research and the all-inclusive systematic efforts to develop
a unified theory that will explain all the observed uniformities of social be-
havior, social organization and social change (Merton, 1968, p. 39).
Merton also indicates that the MRT concept does not involve the distinction
between the "nomothetic and idiothetic [idiographic]" (1968, p. 44), nor does
it refer to either an empirical generalization or a derivation from a single all-
embracing theory of social systems (1968, p. 41V Aside from some examples of
MRTs from the physical, biological, and social sciences, few additional clues to
the meaning ofMRT are provided by Merton. 3
Through his definition and examples, Merton has implicitly defined a contin-
uum along the dimension of theoretical scope. Scope is the quality of a theory
that refers to how many of the basic problems in a discipline are handled by the
same theory (Hage, 1972, p. 178). The more problems handled by a given theory,
the greater its scope.
Empirical generalizations are located at the small-scope end of the continuum,
and general or grand theories are located at the large-scope end. MRTs are lo-
cated in the area between the two extremes. Because Merton has not specified
the location of MRTs with greater precision, any theory with greater scope than
an empirical generalization, but of less scope than the most general theory
in the field, would qualify as an MRT. Merton, however, has failed to discuss this
point explicitly and, therefore, a great deal of confusion and misunderstanding
has resulted.
The source of the confusion and misunderstanding is the way Merton defmed
MRT. His intentional definition is inadequate and readers are forced to rely
upon the extentional meaning he provides. The extentional meaning of a term is
composed of the specific entities to which the term applies (Copi, 1967,
pp. 107-108). In the case of MRT, the extentional meaning has been partially
given in the examples Merton provided. A complete extentional definition for
MRT could be an infmite set of referents, and, in any case, would be an ex-
tremely inefficient and tedious way to define the concept.
A more efficient approach is to defme MRT intentionally. The intentional
meaning of a concept is the list of characteristics, qualities, attributes, and so on,
which qualify an entity as a member of the class to which the term or concept
MIDDLE RANGE THEORY AND THE STRATEGIES OF THEORY CONSTRUCTION 21
refers (Copi, 1967, pp. 107-108). Merton left the work on MRT's intentional
meaning incomplete, and it is this failing that is primarily responsible for the
confusion over its meaning.
If Merton's definition of MRT reproduced in this analysis is examined, very
few intrinsic qualities or attributes can be found. The major attribute, and even
this is not explicitly stated in the defmition, is that middle rangeness is a relative
concept; that MRTs lie on a continuum somewhere between a single empirical
generalization and a theory of infmite scope. Merton is reasonably clear in dis-
tinguishing an MRT from an empirical generalization, because his definition of
an empirical generalization is precise enough to differentiate the two (Merton,
1968,p.41).
Confusion has resulted from problems involving the other end of the contin-
uum; problems that arise in attempts to distinguish between MRTs and general
theories. Put another way, how middling must a theory be to be an MRT? Un-
like the case with empirical generalizations, there are no absolute standards at
the large-scope end of the continuum. The general theory end will be occupied
by the theory with the greatest scope in the discipline. Any other theory with
greater scope than an empirical generalization is an MRT. The relative nature
of the middle range versus general theory classification is highlighted by the
possibility of future theory construction producing a theory of greater scope
than any that may have existed at some prior time. The creation of the theory
of greater scope will transform the theory that previously had the greatest scope
into an MRT. And so forth ad infinitum.4
Unfortunately, much of the discussion of MRT has proceeded as if that desig-
nation were an absolute quality rather than a relative property. This, in turn,
has resulted in many of the wrong questions being asked. For example, a com-
mon quandary for investigators attempting to classify a given theory as middle
range or not is the existence of two or more theories of clearly lesser scope than
some other theory. The problem arises when it becomes obvious that the theories
under examination also differ in scope. But if middle rangeness were an absolute
quality, only one of the theories could possibly be an MRT. This conundrum re-
sults from considering middle rangeness as an absolute property of a theory. By
recognizing the relative nature of the middle range concept, the solution to the
dilemma is straightforward: Both theories are MRTs.
MIDDLE RANGE THEORY AS PROCESS
The basic reason so much concern has been given to MRTs over the last three
decades is the belief that not only will MRTs successfully explain delimited sets
of phenomena, but they will also serve as the raw materials for the construction
of more general theories. It is this belief which serves as one of the major justifi-
cations of the middle range imperative.
Sociological theory, if it is to advance significantly, must proceed on these
interconnected planes: (1) by developing special theories [MRTs] from which
to derive hypotheses that can be empirically investigated and (2) by evolving,
not suddenly revealing a progressively more general conceptual scheme that is
adequate to consolidate groups of special theories [MRTs] .... theory which,
instead of proceeding from the head of one man, gradually consolidates
theories of the middle range, so that these become special cases of more gen-
eral formulations (Merton, 1968, p. 51).
MRT construction is justified as a means to the ends of creating theories of
ever-increasing scope. This imperative can be stated as a proposition (which is
implicit in Merton's discuSSions): The optimal method of generating acceptable
(criteria of acceptability will be discussed later) theories of relatively large scope
is to: (1) generate MRTs and (2) synthesize the MRTs into theories of larger
scope. This proposition will be examined in two W\lYS. First, it will be compared
with possible alternative strategies. Second, the development of several success-
ful theories of relatively large scope will be examined to determine whether or
not they were produced in the manner directed by the middle range imperative.
THE STRATEGIES OF THEORY CONSTRUCTION
Many of the great systems of sociological theory of the nineteenth century (e.g.,
Spencer's and Marx's) are theories with the degree of scope Merton had in mind
as general theories. These theories were produced by single individuals without
following Merton's middle range imperative. Instead, they were produced by a
single individual based on empirical data, some generalizations, and insight. Mer-
ton was critical of these theories and described them as "all-inclusive specula-
tions," possessing "scientific sterility," and being "general orientations toward
data, suggesting types of variables which need somehow to be taken into account,
rather than clear, verifiable statements of relationships between specific vari-
ables" {Merton, 1949, pp. 9_10).5
What the grand theories lack is the necessary conceptual and propositional
integration to produce testable hypotheses that would ground them in the em-
pirical world. Merton felt that these shortcomings could be overcome if the
initial scope of a theory was restricted. Merton's thesis is that premature at-
tempts to initially construct theories of large scope directly from empirical data
will produce systems of theory lacking both empirical grounding and conceptual-
propositional integration.
2 i
I
I
E
REFINEMENT JJ
X
I
P 12
L
iI
I I
C REFINEMENT
I
T
N
E
6
S
S
11
EMPIRICAL . MIDDLE RANGE

G ENERALIZATION~MIODLE
EMRPAI~:~!~IO~ RANGE THEORY
_______ 77'
T'EO~ V
GENE --,. 8 ___
5/ ~
lr
If ------------------------________8______ ~ ...........
10.........
GENERAL THEORY
__ .......
- -... 00
00
LOW HIGH
SCOPE
tv
W . of
Figure 11.. The Strategies Construction
0 f Theory Construct ion
The absence of these characteristics is what Dumont and Wilson (1967, pp.
988-989) have called the explicitness of a theory. They have constructed an
explicitness continuum ranging from a low end (implicit theories), through a
moderate range (theory sketches), to a high end (explicit theories). At the high
end, concepts are tightly (often mathematically) integrated, and measurement
and testing procedures are tightly linked to the theory as well. Merton's criti-
cisms of the large scope (grand, general) theories amount to criticisms of their
explicitness. It follows from his prescription that theories of great initial scope
will be less explicit than theories of lesser initial scope.
If the previously discussed scope continuum is combined with Dumont and
Wilson's explicitness continuum (1967, p. 989), the diagram in Figure I is pro-
duced. Twelve possible theory construction strategies are depicted within the
axes. Although there are actually an infmite number of strategies, the 12 strat-
egies, depicted in Figure 1 can be thought of as the core strategies. The remain-
ing infinity of possibilities represents differences in degree rather than kind.
Table 1. Theory Construction Strategies.
Strategy Number Strategy Description

1 Develop empirical generalizations.
2 Refine empirical organizations.
3 Synthesize empirical generalizations into a
middle range theory.
4 Synthesize refined empirical generalizations
into a middle range theory.
5 Develop a middle range theory without using
strategies 3 or 4.
6 Refine middle range theories.
7 Synthesize middle range theories into a gen-
eral theory.
8 Synthesize empirical generalizations into a
general theory.
9 Synthesize refined middle range theories into
a general theory.
10 Develop a general theory without using
strategies 7, 8, 9, or 11 or the compound
strategies (see Note).
11 Synthesize refined empirical generalizations
into general theories.
12 Refine general theories.
NOTE: Although they have not been listed, it is possible to create compound strategies
by combining individual strategies in a serial fashion (e.g., strategies 3 and 7, and 4 and 9).
Merton's analysis of MRTs and his imperative to use them to construct more
general theories involve strategies 5, 6, 7, and 9. The remaining strategies, in-
cluding the possible compound strategies, were either not discussed by Merton
or were rejected by him. The individual strategies are described in Table 1.
TESTING THE MIDDLE RANGE IMPERATIVE
The proposition concerning the optimal technique for developing relatively gen-
eral theories is completely undocumented by Merton. He fails to cite even a
single specific example from the physical, biological, or social sciences of the
type of synthetic process he prescribes. His reference to Darwin's theory of coral
atoll formation (Merton, 1968, p. 40) as an example of an MRT is particularly
insightful on this point. It reveals that Merton did not consider Darwin's theory
of organic evolution to be an example of an MRT. He probably regarded it as a
theory equivalent in scope to the work of Spencer, Marx, and so forth. Other-
wise, why pick the relatively obscure coral atoll theory rather than the famous
evolutionary theory for use as an example?
If Darwin's theory of evolution is of greater scope than Merton envisioned
for MRTs, three disconfirmations of the middle range imperative are revealed.
The first two disconfirmations involve the independent development of the
theory of evolution by Darwin and Wallace. Neither man produced the evolu-
tionary theory by synthesizing theories of lesser scope (MRTs). Instead, both
developed the theory from an examination of the data. In fact, Wallace dis-
covered the principle of natural selection while he was experiencing an attack of
fever (Eiseley, 1961, p. 291). Since both Darwin and Wallace independently de-
veloped the theory of evolution by techniques different from Merton's prescrip-
tion, they represent two disconfirmations of Merton's middle range imperative.
Examining the history of the evolutionary theory leads to a third disconfir-
mation of the middle range imperative. One of the few great theoretical syntheses
in the history of science is the modem synthetic theory of evolution (Dobzhan-
sky, 1970, pp. 1-29; Huxley, 1942; Mayr, 1970, pp. 1-9). The synthetic theory
combines Darwinian evolutionary theory with the theory of genetics. But since
Merton did not consider Darwin's theory of evolution an example of an MRT,
its synthesis with genetics cannot be regarded as an example of the application
of the middle range imperative. 6
The development of genetics reveals four additional disconfirmations. Mendel
originally discovered the laws of inheritance in the 1850s and 1860s working
with empirical data, not MRTs (Asimov, 1972a, p. 367). In 1900, three biolo-
gists (De Vries, Corens, and von Tschermak) independ~nt1y rediscovered the laws
of inheritance and discovered Mendel's original papers (Asimov, 1972b, p. 585).
And none of the three synthesized MRTs to produce their results.
An eighth disconfirmation can be found in the physical sciences: Newtonian

mechanics. While it is known that both Kepler's laws of planetary motion and
Galileo's laws of freely falling bodies are deducible from Newton's theory (Berg-
mann, 1966, p. 90), Merton does not mention Newton's theory as an example of
the successful use of the middle range imperative. The reason is Newton did not
build his theory by synthesizing the theories of Galileo and Kepler. Instead, he
built his own theory and then empirically grounded it by demonstrating that the
laws of Kepler and Galileo were derivable from it. This grounded Newton's
theory, because the laws of Kepler and Galileo had previously been empirically
confirmed. Newton's theory, too, was the result of work with empirical data.
Since not only did Merton fail to provide even a single example of a suc-
cessful application of his prescription, but eight discon/irming examples were
located from the history of science, a history Merton knows very well (e.g.,
Merton, 1938), the validity of the middle range imperative is highly questionable.
It is thereby fair to ask: Should it be rejected altogether? Although his prescrip-
tion is based on a priori conjecture rather than on a demonstrated record of prior
success, the pragmatic position of following several strategies and evaluating
what works, if anything, seems much more likely to produce an acceptable
theory of large scope than any single strategy.
The disconfirmations do not demonstrate that theories of relatively large
scope cannot be produced by synthesizing MRTs. Rather, the disconfirmations
demonstrate that they can be (and have been) produced by other means. Choos-
ing a single theory construction strategy to the exclusion of all others is fool-
hardy in light of the history of science. It is particularly so, given the dearth of
evidence for the middle range imperative suggested by Merton. Singling out one
procedure for a concentration of effort is more likely to maximize the chances
of arriving at a dead end than it is to produce good theory.
The middle range imperative, like most theory construction prescriptions,
suffers by making a false assumption. This assumption is that the logic of dis-
covery is the same as the logic of demonstration, and as such, is subject to rules
and prescriptions. As Robert Nisbet (1976, p. 5) properly points out, the logic
of demonstration is properly subject to rules and prescriptions, but given our
small knowledge of creativity, the logic of discovery is not properly subject to
rules and prescriptions at all.
Most social theorists abandoned the concept of uniform and unilinear cultural
evolution some time ago (Timasheff and Theodorson, 1976, pp. 54-55). It
would be well for theorists to learn from the lessons offered by the history of
thought on social evolution and to likewise reject prescriptions and imperatives
for a uniform and unilinear theory construction strategy. In fact, there may be
several compelling reasons to support a multiplicity of theory-building strategies.
TOWARDS A MULTIPLICITY OF THEORY-BUILDING

STRATEGIES
The foregoing criticism of Merton's formulation of MRT should not be taken as

a criticism of middle range theory construction per se, which we would argue is
an essential feature of any scientific endeavor. Indeed, we need to acknowledge
Merton's valuable role in underscoring the importance of MRT building and in
stimulating three decades of fruitful dialogue on the topic. His ideas were un-
doubtedly a critical catalyst for this conference.
We do, however, object to the way in which Merton relates MRTs to both
empirical generalizations at the limited-scope end, and to grand theorizing at
the broad-scope end of the theory-building scale. While accepting his contribu-
tion regarding the value of MRT, we view his overall formulation of the relation
of MRTs to raw empiricism and grand theorizing as being both historically
unfounded and scientifically impeding. To use Merton's own phrase, "The issue
must be fairly joined" (1968, p. 45)!
Specifically, we reject most of what he writes about the relationship between
MRTs and grand theorizing, which leads him to the position that social theory
can best be developed by avoiding the ''vague and unordered speculations" asso-
ciated with trying to construct comprehensive, all-embracing, unified, wide-
ranging grand theories. In essence, Merton is urging social scientists to avoid
wasting energies on broad-scope conjecture and grand theorizing, which are
described as too vague, too remote, too soon, too inclusive-in short, too much!
His pejorative incantations against grand theorizing activities seem to us both
in error (from the evidence of the development of thought and the sOciology of
knowledge), and dysfunctional for future scientific inquiry (except from the
standpoint of stimulating antithetical discussion from those who do appreciate
the value of "mere philosophical speculation").
Merton writes as if broad-scope, grand theorizing is both an evil to be avoided
and is simultaneously a desirable good. It is an evil since it represents a single,
comprehensive, all-knowing termination of inquiry, and a desirable good, since
it provides the meaningful goal toward which scientists are gradually workinl! via
careful empirical research and middle range theorizing.
We do not believe that these ambivalent viewpoints should be widely shared.
There is no reason why several grand theories cannot simultaneously prevail;
there is no compelling basis for believing in the existence of an ultimate grand
theory; and we would hope that any grand theory is modifiable. We view inquiry
as a process of continuous transformation rather than a journey to some hypo-
thetical ultimate destination. We also see a positive value in broad-scope theo-
rizing, however tentative and temporary it may be.
In Merton's world view, MRTs act as a kind of buffer to separate empirical

research from the invasion of unbridled speculative thought. MRTs, it appears,
protect the purity of empirical inquiry and keep at bay the ''vague and unor-
dered speculations" of imagination and abstract thought. Such a world view is
unacceptable, and no less than four alternative world views suggest themselves as
being less constraining and more fruitful for scientific research.
Alternative 1. Theorizing as Blending. It can simply be assumed that all scien-

tific inquiry necessitates both attention to empirical evidence (concrete sense
data) and to speculative thought (generalized abstractions). Why must a prescrip-
tion for inquiry that fixates on anyone part of the scope scale be accepted?
Creative theory building requires that movement be continuously flexible back
and forth along the scope continuum.
Alternative 2. Theorizing as a Bridge. Kurt Lewin (1951, pp. 77-80) once de-
scribed the scientific process as a bridging or tension-spanning process, a process
spanning the distance between observed facts on the one side and abstract
thought on the other. The observed facts create a need (tension) in the researcher
to provide an explanation, a theory. The more unexplained facts the researcher
can hold, the greater the need (tension) for explanation, and the broader the
tension-spanning of the resulting theory will be.
Empirical generalizations are low-tension spanning theories, while grand
abstractions are high-tension spanning theories. Thus, the scope of a theory, in
Lewin's formulation, is a function of the researcher's capacity to manage the
intellectual tension between unexplained observations and theory.
Alternative 3. Theorizing as Personal Style. Psychological research on individ-

ual differences indicates the existence of a wide range of cognitive and affective
styles. One would therefore suppose that individuals would vary considerably in
their learning styles, and more particularly, in their research styles. The Jungian
typology of personality provides an excellent illustration. Jung's (1975) theory
indicates that individuals differ in their methods of gaining knowledge about
their worlds based on two dichotomous scales: (1) the perception scale of sensing
(S) or intuiting (N) and (2) the judging scale of feeling (F) or thinking (T). Work
on the Myers-Briggs Type Indicator (Briggs-Myers, 1962) provides empirical sup-
port for this theory.
One would expect that ST researchers would do fundamentally different
kinds of research from that done by NT researchers. Specifically, the theory
would predict that ST researchers would tend to work on the empirical general-
ization end of the scope continuum while NT researchers would tend to be
found at the grand theory level.
Alternative 4. Theorizing as Socially Detennined. From the perspective of the

sociology of knowledge, theorizing is a sociopolitical process fundamentally
linked to particular historical contexts. Mannheim (1965) has provided the most
systematic elaboration of the sociology of knowledge perspective. It is the total
perspective of social scientists that determines their world view and hence the
way they will theorize about that world.
Georg Simmel (1950) has made essentially this same point in his discussions
on the nature of reality.
We obtain different pictures of an object when we see it at a distance of two,
or of five, or of ten yards. At each distance, however, the picture is "correct"
in its particular way and only in this way .... All we can say is that a view
gained at any distance whatever has its own justification. It cannot be re-
placed or corrected [italics ours] by any other view emerging at another
distance (1950, pp. 7-8).
Simmel goes on (1950, p. 8) to make the same point about the investigation
of human life, that there are many different "distances" [perspectives] and that
no particular perspective is more justified than any other. The concreteness of
the situation, and the perspective and interests of the researcher are the primary
determinants of both the form and content of the theories that will be generated.
Merton's prescriptions for "the right way to build social theory" are, from
this viewpoint, related to Merton's own total perspective and situation rather
than to any intrinsic universal truth about theory-building. It is apparently our
own perspectives, our own unique views of the social world, which have led to
this critique of the middle range imperative and to the series of recommenda-
tions that have been made.
NOTES
1. The first post-World War II call for MRTs was made by T.H. Marshall in 1946 (1963,
pp. 21-22). This call was then tentatively amplified by Merton (1948) before he stated his
position on the middle range imperative in the three editions of STSS. Recently, Walton
(1972) and Weick (1974) have also argued for the middle range imperative.
2. EXamples of empirical generalizations can be found in Berelson and Steiner (1964)
who have provided 1,045 examples in their codification of findings on human behavior.
The "proverbs of administration" (Simon, 1976, pp. 20-44) would be likely candidates for
empirical generalization status from organization theory. All-embracing theories would be
the works of Marx, Spencer, Parsons, etc. (Merton, 1968, pp. 43, 46).
3. Examples Merton (1968, p. 61) cited, which are familiar to organizational theorists,
include the following: Se1znick (1949), Gouldner (1954), Blau (1963), Lipset, Trow, and
Coleman (1956), Crozier (1964), and Katz and Kahn (1966). Merton (1968, p. 61) con-
sidered the work of March and Simon (1958) too general to qualify as an MRT.
4. It is also possible that the transformation process could operate in the opposite direc-
tion. If a general theory was disconfumed a sufficient number of times-perhaps often
enough to produce the crisis Kuhn (1970) describes as leading to revolutionary science-it
might be abandoned, and some MRT might ascend to the position of most general theory.
5. Some theorists have expressed the opinion that providing general orientations and sug-
gesting which variables are important is, in fact, the major task of social theory. See Blumer's
(I954, pp. 7-10) discussion of "sensitizing concepts" for a statement of this view.
6. Dobzhansky (1970, p. 25) attributes the modern synthesis to the work of Fisher
(1930), Haldane (1932), Tshetverikov (1961), and Wright (1931).
REFERENCES
Asimov, I. Asimov's Biographical Encyclopedia of Science and Technology. Gar-

den City: Doubleday, 1972a.
- - . Asimov 's Guide to Science. New York: Basic Books, 1972b.
Berelson, B., and Steiner, G.A. Human Behavior: An Inventory of Scientific
Findings. New York: Harcourt, Brace, and World, 1964.
Bergmann, G. Philosophy of Science. Madison: University of Wisconsin Press,
1966.
Blau, P.M. The Dynamics of Bureaucracy (2nd ed.). Chicago: University of
Chicago Press, 1963.
Blumer, H. What is wrong with social theory? American SOCiological Review,
1954,19,3-10.
Bridgman, P.W. Reflections of a Physicist. New York: Philosophical Library,
1950.
Briggs-Myers, I. Myers-Briggs Type Indicator Manual. Princeton: Educational
Testing Service, 1962.
Copi, I.M. Introduction to Logic. New York: Macmillan, 1967.
Crozier, M. The Bureaucratic Phenomenon. Chicago: University of Chicago
Press, 1964.
Dobzhansky, T. Genetics of the Evolutionary Process. New York: Columbia Uni-
versity Press, 1970.
Dumont, R.G., and Wilson, W.J. Aspects of concept formation, explication, and
theory construction in sociology. American Sociological Review, 1967,32,
985-995.
Eise1ey, L. Darwin's Century. Garden City: Anchor Books, 1961.
Fisher, R.A. The Genetical Theory of Natural Selection. Oxford: Clarendon
Press, 1930.
Gouldner, A.W. Patterns of Industrial Democracy. Glencoe: The Free Press,
1954.
Hage, J. Techniques' and Problems of Theory Construction in Sociology. New
York: John Wiley & Sons, 1972.
Haldane, J.B.S. The Causes of Evolution. New York: Longmans, Green, & Co.,
1932.
Huxley, J. Evolution, the Modern Synthesis. New York and London: Harper &
Brothers, 1942.
Jung, C.G. C. G. Jung: Collected Works, Vol 6, Psychological Types. Princeton:
Princeton University Press, 1975.
John Wiley & Sons, 1966.
Kuhn, T.S. The Structure of Scientific Revolutions (2nd ed.). Chicago: Univer-
sity of Chicago Press, 1970.
Lewin, Kurt. Comment concerning psychological forces and energies, and the
structure of the psyche. Organization and Pathology of Thought (David
Rapaport, ed.). New York: Columbia University Press, 1951.
Lipset, S.M.; Trow, M.; and Coleman, J. Union Democracy. New York: The Free
Press, 1956.
Mannheim, K. Ideology and Utopia. New York: Harcourt, Brace, and World,
1965.
March, J.G., and Simon, H.A. Organizations. New York: John Wiley & Sons,
1958.
Marshall, T.H. Sociology at the Crossroads and Other Essays. London: Heine-
mann, 1963.
Mayr, E. Population, Species, and Evolution. Cambridge: Harvard University
Press, 1970.
Merton, R.K. Science, technology and society in Seventeenth Century England
(in Osiris: Studies on the History and Philosophy of Science, and on the
History of Learning and Culture, G. Sarton, ed.). Bruges, Belgium: St.
Catherine Press, 1938.
- - . Discussion. American Sociological Review, 1948,13, 164-168.
- - . Social Theory and Social Structure. Glencoe: The Free Press, 1949.
- - . Social Theory and Social Structure (revised and enlarged ed.). Glencoe:
The Free Press, 1957.
- - . On Theoretical Sociology. New York: The Free Press, 1967.
- - . Social Theory and Social Structure (enlarged ed.). New York: The Free
Press, 1968.
Nisbet, R. Sociology as an Art Form. New York: Oxford University Press, 1976.
Selznick, P. TVA and the Grass Roots. Berkeley: University of California Press,
1949.
Simmel, G. The Sociology of Georg Simmel (translated by Kurt H. Wolff). New
York: The Free Press, 1950.
Simon, H.A. Administrative Behavior (3rd ed.). New York: The Free Press,
1976.
Timasheff, N.S., and Theodorson, G.A. Sociological Theory: Its Nature and
Growth (4th ed.). New York: Random House, 1976.
Tshetverikov, S.S. On certain aspects of evolutionary process from the stand-

point of genetics (originally in Zhurnal Eksperimental'noi Biologii, 1926, A2:
3-54. Translated by Malina Barker and edited by I. Michael Lerner). Proceed-
ings o/the American Philosophical Society, 1961,105, 167-195.
Walton, R.E. Advantages and attributes of the case study. Journal 0/ Applied
Behavioral Science, 1972,8,73-78.
Weick, K.E. Middle Range Theories of Social Systems. Behavioral Science, 1974,
19,357-367.
Wright, S. Evolution in Mendelian populations. Genetics, 1931,16,97-159.
3 MIDDLE RANGE THEORY
An Overview and Assessment
for Organizational Research
Paula C. Morrow and Paul M.
Muchinsky, Iowa State University
References to middle range theories, middle range theorizing, and middle range
concepts are increasingly common in organizational research. The growing fre-
quency of their use may indicate the onset of a new trend, perhaps even a new
strategy of inquiry for organizational research. The middle range conceptualiza-
tion, however, like most research tools, has limitations as well as advantages.
These potentially offsetting factors require critical evaluation before the middle
range notion can be further endorsed. The purpose of this paper is to provide
such an overview of the middle range approach, rendering particular attention to
the nature and meaning of middle range notions for understanding organizations
and organizational behavior. In addition, we will offer some pragmatic sugges-
tions for conducting research consonant with the middle range orientation.
The middle range designation is used to refer to two different, yet related
ideas. The first is the notion of a middle range theory. A middle range theory,
according to Merton (1967), consists of a set of assumptions from which specific
hypotheses are logically derived and subsequently confirmed by empirical in-
vestigation. More generally, a middle range theory has corne to be regarded as a
limited set of interrelated propositions aimed at an intermediary level of expla-
nation of a given phenomenon. Such theories are said to be of the middle rather
than ordinary or highly abstract range because the concepts embodied in the
33
constituent propositions are couched at a middling level of generality. They are

specific enough to be of use in the evaluation of observed phenomena, yet gen-
eral enough to be incorporated into broader theoretical statements. Such
generality, in tum, permits many middle range theories to transcend historical
time frames and cultures.
Merton advocates the construction of middle range theories when grand
theoretical schemes are premature, that is, where the theoretical and empirical
groundwork necessary for the promulgation of a comprehensive series of logi-
cally interrelated propositions has not been established. As these situations
characterize many areas of organizational research, a policy that recommends
more middle range theories seems eminently reasonable. Merton, however,
explicitly specified that the formation of middle range theories was not an end
in itself. The middle range designation also implies that researchers will periodi-
cally consolidate their less abstract middle range theories into more general
theories that manifest greater theoretical coverage. In short, the middle range
designation represents a two-step model: (1) the building of middle range theories
and (2) the subsequent integration of these theories into more encompassing
theoretical frameworks. The integration process is thus a kind of theory con-
struction technique that can be termed middle range theorizing. This duality of
meaning in the middle range designation must continually be borne in mind, as
advocacy of middle range theories necessarily reflects a desire to build more
general theories.
THE MIDDLE RANGE LEGACY
While Merton (1949; 1967) is often credited with the development of the middle
range notion, he clearly notes that he is not the originator of the idea. The
writings of Plato, Bacon, J.S. Mill, and T.H. Marshall all advocate middle axioms,
middle principles, limited theories, and stepping stones in the middle distance.
Beyond this, many contemporary social scientists have endorsed the policy: for
example, Rossi, Riesman, Zetterberg, and Weick (see Merton, 1967, pp. 56-62).
Common middle range theories evident in current research include dissonance
theory, theories of collective behavior, reference group theory, and in the orga-
nizational sphere, theories of oligarchy, bureaucracy and contingency. The
middle range concept is thus neither a new idea nor an innovative strategy for
scientific inquiry. It has experienced a sufficient tenure for evaluation in terms
of relative advantages and disadvantages.
Numerous advantages can be cited for theories of the middle range and mid-
dle range theorizing. The primary benefits of this approach, according to Merton
MIDDLE RANGE THEORY 35
(1967), are: (1) its ability to transcend pure description and single empirical
observations; (2) its ability to draw upon divergent units of analysis, perspectives
and even disciplines in order to formulate new theory; (3) its recognition that
basic knowledge must be obtained before complex theoretical questions can be
answered; and (4) its flexibility in that it allows researchers to seek generalizabil-
ity without totally endorsing the belief that a single, unified social science is
achievable. Other advantages to the middle range orientation include that it is a
strategy that legitimates desires to solve practical problems while engaging in
basic research (Le., it seeks to produce generalizations while remaining empiri-
cally grounded) and, because it rests on the integration of theories constructed
by various researchers, it manifests a built-in form of replication. A fmal advan-
tage rests on Weick's (1974) observation that too many theories are not falsi-
fiable. Middle range theories are valuable because, when properly tested, they
permit empirical confirmation or rejection.
LIMITATIONS OF THE MIDDLE RANGE APPROACH
Limitations of the middle range orientation have also been observed. These dis-
advantages can be summarized as: (1) the charge that the emphasis on testable
propositions is merely a ruse for a positivistic conception of science; (2) the fact
that errors in the initial theories or set of assumptions from which a middle
range theory is derived are undetectable; (3) the charge that middle range think-
ing promotes fragmentation and an overemphasis on unrelated, specialized
theories; and (4) the charge that the pursuit of middle range theories stimulates
only low levels of ambition and generally atheoretical research (Bierstedt, 1960).
The first disadvantage is associated with those scholars who contend that
empirical confirmation and logic can be supplanted by other techniques of
knowing (e.g., experiential, phenomenological approaches). This critique reflects
a philosophy of science issue that applies to more than just middle range theo-
rizing and is, accordingly, beyond the scope of this essay. It is, however, a
legitimate criticism of the approach. The second disadvantage stems from a
common failure among theorists and researchers to identify and explicate all of
the assumptions underlying the derived hypotheses formulating a middle range
theory. These assumptions, along with restrictions associated with testing a mid-
dle range theory in a particular research setting, can be labelled as auxiliary
theory (Blalock, 1968). Middle range theories, in comparison to more grandiose
theories, require more auxiliary theory in order to reduce the number of poten-
tially relevant variables to a manageable set and to permit empirical evaluation
of the derived hypotheses. The greater the number of assumptions necessary for
bringing closure to an explanatory system (Le., a theory), the greater the likeli-
hood that some implicit assumptions will be overlooked and that some erroneous
assumptions will unknowingly be accepted.
The charge of fragmentation is perhaps the most serious criticism that can'be
directed against the middle range orientation. Many researchers neglect the con-
solidation component of the middle range approach and thus never seek to go
beyond explanations connected with a limited range of observations. Such
tunnel visioning tactics produce areas of inquiry marked by disparate, sometimes
inconsistent, specialized theories. In addition, such fragmentation can extend
to the point where differential aspects of the same or closely related phenomena
are explained by distinct, unrelated theories. For example, the relationship
between job satisfaction and job behavior has alternately been explained by
attitude-behavior theories, theories of need satisfaction (Alderfer, 1969), and
instrumentality theory (Lawler and Porter, 1967). Ideally, these middle range
theories should ultimately be synthesized into a more general macroscopic the-
ory. The final disadvantage that has been observed concerns the potentially athe-
oretical nature of middle range research. Some regard endorsement of the middle
range approach as "sickly, pale, and singularly unambitious" (Merton, 1967,
p. 63). It is apparent that the degree of scope appropriate to a middle range
theory is a matter of opinion and probably contingent on the extent to which
the specific research problem subsumed within the theory has received empirical
consideration in the past.
In summarizing the last three disadvantages, one common problem can be
identified. The fundamental limitation of the middle range perspective is its
susceptibility to misuse. When the construction of middle range theories be-
comes an end in itself, rather than a means to an end, the ultimate purpose of
the strategy is destroyed and the disadvantages cited above become operational.
Organizational researchers can readily understand this problem as a higher-order
example of goal displacement. In short, the common disadvantages of the middle
range orientation are not endemic to the approach but become evident in the
hands of some researchers.
Finally, the evaluation of the middle range orientation is sometimes prede-
termined by one's stance in regard to other issues. For example, there are
various opinions concerning how science (and organizational research) can best
be advanced. Many argue that science making is facilitated by researchers build-
ing and refming each other's work over time. Viewing science as such a cumula-
tive process engenders norms of cooperation among researchers. Theories and
theorizing of the middle range, with their emphasis on gradual development
through periodic consolidation, are highly consistent with this outlook. Others,
however, feel that genuine scientific advancements occur most often through
the concentrated work of a single individual or through serendipitous fmdings.
In addition, there are those (e.g., Popper, 1959) who believe that science is best
promoted under conditions of competition rather than cooperation. Researchers
who maintain these attitudes toward science fmd little utility in middle range
approaches. Merging theories that are discrepant in some respects may be con-
ceived as compromising, or as a strategy more likely to reduce the probability
of uncovering more useful theory.
In the remainder of this paper some of the less commonly recognized impli-
cations of the middle range approach are discussed. These issues are perhaps
more important for the overall assessment of middle range theory as they im-
pede the implementation of the middle range approach.
ISSUES IN MIDDLE RANGE THEORY CONSTRUCTION:

SOME SUGGESTIONS
The foregoing review of advantages and disadvantages of middle range theorizing

attests, on the whole, to the utility of the approach in organizational theory
development. Certain ambiguities persist, however, which dampen our enthusi-
asm for middle range theories in general, and for middle range theorizing as a
theory construction technique. Specifically, the rationale behind middle range
theorizing is so abstract that we are provided little insight as to how one actually
builds a middle range theory or consolidates middle range theories into more
general ones. Merton's (1967) advice is to follow a functionalist paradigm of
inquiry. While this paradigm may be useful to some organizational researchers,
it incorporates a myriad of assumptions that are unacceptable to others. By
explicating the rationale behind middle range theorizing, we would like to offer
some suggestions for constructing middle range theories.
The major function of a middle range theory is to reduce the scope or cover-
age of a set of interrelated propositions (Le., a theory) to a level that can be
empirically evaluated. The basic idea is to narrow the theoretical domain of the
propositions to a point where they can be refuted or supported on empirical
grounds. In addition, these propositions should be selected and stated in a man-
ner that anticipates subsequent consolidation on a more general plane. This
reduction in scope or coverage is accomplished through the initial set of assump-
tions established prior to the derivation of propositions from the more abstract
sources. The selection, content, and nature of these assumptions is nearly always
a matter of researcher discretion. We feel that when judiciously exercised, this
freedom to assert assumptions provides an excellent vehicle for devising middle
range theories. At least three categories or types of assumptions that would
simplify an area of organizational inquiry and thus facilitate the construction of
middle range theories can be identified.
1. Operating at a Lower Level of Abstraction
Researchers who are confronted with a very complex idea often feel they cannot
adequately capture the essence of their idea in a single concept. The idea, which
is sometimes identified as a construct rather than a concept in order to under-
score its complexity, is judged to be multidimensional (Le., a notion that must
be broken down into other less abstract concepts before it can be operation-
alized). Researchers who cannot incorporate all the dimensions of highly com-
plex constructs into their work, may simply state that they will only deal with
a restricted domain of these constructs. Hage (1972) views the formulation of
such an assumption as a way to build middle range theories, and terms the pro-
cess concept explication. He illustrates this technique of moving to a lower level
of abstraction by discussing how a researcher who is interested in the relation-
ship between bureaucracy and organizational efficiency might proceed. Bureau-
cracy, as a construct, can be defined as the extent to which an organization is
characterized by a rational-legal authority structure. Hage suggests that this
complex construct involves four subsidiary notions: (a) the degree of hierarchy
or centralization, (b) the number of rules or formalization, (c) the number of
formal offices or complexity, and (d) the level of technical competence charac-
terizing the organization. By explicating bureaucracy in this manner, one can
construct propositions linking dimensions of bureaucracy and efficiency at a less
abstract 'level (e.g., the greater the level of centralization, the greater the organi-
zational efficiency). While Hage's example pertains only to the bureaucratic con-
struct, a similar procedure could be Simultaneously followed for efficiency.
Shifting from a more abstract to a less abstract level is a useful assumption to
make in situations where a middle range theory is desired for at least two rea-
sons. First, the lower informative value of the derived propositions increases the
likelihood that the propositions will be empirically supported (Zetterberg,
1966). This occurs because the propositions must only hold true for a smaller
theoretical domain. A second advantage is an increase in confidence in the valid-
ity of the fmdings attributable to higher epistemic correlations between the
explicated notions and their measures (Kaplan, 1964). The offsetting disadvan-
tage of subscription to this and the two forthcoming assumptions is the reduced
generalizability of the resultant theory. Should the level of abstraction become
too low, our ability to extend the fmdings beyond the narrowly defmed con-
cepts in the theory would be markedly reduced. In the worst of situations we
would have as many middle range theories as we have objects of investigation.
2. Restriction of the Sampling Frame
A second technique for reducing scope is to restrict the applicability of the

propositions under examination to one unit of analysis. This technique is tanta-
mount to assuming that the theory must only be confirmed in a limited sample
of units. Organizational researchers frequently use this device in that they build
theories intended to describe limited categories of organizations (e.g., nonprofit
organizations, utilitarian organizations, etc.). Assuming that one's theory is ap-
plicable only to certain units of analysis, reduces the level of coverage required
by the propositions and thus is a means for building middle range theories (Hage,
1972).
The ultimate utility of this technique, however, rests on the outcome of ef-
forts to consolidate middle range theories devised under this assumption. In the
study of organizations, this point is especially problematic. Those attempting to
integrate middle range theories formulated via this assumption are working with
theories generated under an intratypical sampling frame. Intratypical sampling
is a sampling strategy where only one type of organization is used in the analysis
(Kimberly, 1976). The rationale behind intratypical sampling is that theories
supported in one organizational sample will be replicated in other types of orga-
nizations, reconfirmed, and thereby achieve greater generalizability. To this
point, middle range theorizing and intratypical sampling logic are quite com-
patible. The problem emerges when one attempts to discern when a sufficient
number of organizations has been sampled and used to test the theory. As
many have observed (e.g., Blau and Schoenherr, 1971; Hall, Haas, and Johnson,
1967), no one has been able to delineate the universe of organizations. Accord-
ingly, there is no ready formula for ascertaining when a random sample of
organizations has been acquired.
The existence of this problem should not be interpreted as a lack of endorse-
ment for this assumption as a way of building middle range theories. The alter-
native of intertypical sampling (i.e., selection of a heterogeneous sample of
organizations) has also encountered criticism (Kimberly, 1976). But it is impor-
tant to recognize that the issue of what constitutes a universe of organizations
will have to be resolved before middle range theories of organizational types can
be consolidated into a singular theory of organizations.
3. Imposition of Control Conditions
A third assumption one can make to reduce the scope of propositions is to

institute a requirement that certain ancillary conditions exist before the prop-
ositions are evaluated. Control conditions commonly instituted in organizational
research include whether the propositions describe the organization in the short
run or the long run and the frame of reference (constituency) to be used in the
analysis (Katz and Kahn, 1966; Pennings and Goodman, 1977). The most com-
mon use of this assumption, however, is reflected in organizational contingency
theories. The essence of these middle range theories is that organizational struc-
tures are strongly influenced by (contingent on) their environments. Empirical

assessments of contingency theories related to complex organizations are mixed
(Pennings, 1975). Particular middle range theories identify one or more environ-
mental elements as critical (e.g., degree of uncertainty, level of competition),
and then attempt to show how these environmental characteristics are associated
with patterns of organizational structure or behavior. The restriction in scope is
accomplished by the requirement that the theory need only be confirmed for
organizations existing in or under particular environmental conditions. It should
also be noted that contingency approaches have found expression in intraorga-
nizational research. Fiedler's (1967) contingency theory of leadership is a well-
known example.
From an operational perspective, some overlap between this assumption and
the use of a restricted sampling frame can occur. When the designated control
condition coincides with a factor used in the formation of organizational typol-
ogies, the derived middle range theory may reflect both a control condition and
a sample of organizations. Technology, for example, is used as a differentiating
characteristic in organizational typologies and as an environmental dimension by
some researchers.
Once a set of assumptions is specified for a middle range theory, implicit
propositions can be derived and evaluated. These deductive procedures are
well established and are not further discussed here. Instead, we turn to the in-
ductive aspect of the middle range orientation: the consolidation of middle
range theories into more comprehensive theories. This component of the strat-
egy also manifests a number of unresolved issues.
ISSUES IN MIDDLE RANGE THEORY CONSOLIDATION:

SOME OBSERVATIONS AND INSIGHTS
While the incorporation of various types of assumptions is a useful technique for

building middle range theories, the issue of how to recombine them into more
macroscopic theories remains problematic. A strategy we investigated but did
not deem overly useful was an adaptation of the procedures used for combining
two or more ordinary level propositions into higher-order statements. Zetterberg
(1966), for example, suggests that propositions can be synthesized by ensuring
that linkages between concepts are of the same variety, by subsuming specific
terms into more general ones, and by avoiding references to time and space
factors. This advice is, of course, relevant but not especially pragmatic for
the synthesis of middle range theories. Others (Ritzer, 1975; Morrow, 1978)
who have examined the problems associated with the synthesis of theories
have observed that the formation of higher-order theories is often resisted for
philosophical (e.g., incompatible domain assumptions) and political reasons

(e.g., individuals may have vested interests in the continued endorsement of
lower·level theories). This implies that strategies for consolidation will need
to take into account more than just technical or analytical issues of theory
recombination.
Our examination of the consolidation process may, however, provide some in-
sights useful to future researchers attempting to integrate middle range theories.
Several latent functions of the theory consolidation process can be observed.
The first is related to Hage's (1972) concept explication as a technique for mid-
dle range theory building. Once researchers have secured empirical support for
two middle range theories and attempted to integrate them, they sometimes fmd
that concepts that are intended to reflect dimensions of a more abstract con-
struct are not positively interrelated. In terms of Hage's example, one might have
a middle range theory of formalization and a middle range theory of centraliza-
tion, each supported by empirical findings. A logical step might be to synthesize
these two theories to a higher level of abstraction (i.e., formulate a more general
theory of bureaucracy). However, if formalization and centralization should
be unrelated or negatively interrelated empirically, the likelihood of a successful
merger is in doubt. In such a situation the researcher must choose between em-
pirical criteria for defining the concept of bureaucracy (i.e., positive correlations
between formalization and centralization) or theoretical criteria (e.g., Weber's
definition of a bureaucracy). A positive function of the consolidation process is
thus the potential for recognizing such inconsistencies. It is likely that an unrec-
ognized assumption or empirical condition is present in one or both of the
middle range theories that is responsible for the inconsistency. Researchers must
then retrench and critically examine the auxiliary theory associated with each
middle range theory. Turner (1974) thus seems to be correct when he states that
consolidation efforts are more than just knocking the extremes off theories and
pointing to areas of convergence (p. 122). At least two more benefits of the
theory consolidation process can be described.
One of these is the recognition of when a theory has exhausted its explana-
tory capability and cannot be elevated to a higher level of abstraction. The other
is attributable to the systematic nature of the middle range theory consolidation
process. Middle range theorizing can occur only when researchers relinquish their
allegiances to separate theories and attempt to view an entire series of theories
holistically. In adopting such a gestalt orientation, researchers hope to discern
new areas of convergence that were previously unrecognized and thereby gain
insight for synthesizing more general theories. This process may have 3.Qother
benefit as well. Holistic assessments can reveal knowledge gaps and thereby iden-
tify some objectives of future research. This function is analogous to the use of
the periodic table of elements in chemistry. For many years this systematic table
guided chemists' search for new elements by indicating which logical possibilities
had not received empirical confmnation. The middle range approach to organiza-
tional inquiry may similarly be able to guide researchers to areas of investigation
that have a strong potential for theoretical advancement.
Finally, it must be recognized that the recombination of middle range theories
is a creative act contingent on a researcher's own unique analytical abilities. Any
process of induction, including the synthesis of theories, requires a selection of
elements for recombination that cannot readily be replaced with cookbook-type
guidelines. Even though we would contend that the ability to intuit similarities
and resolve discrepancies is a skill that can be developed through concerted
application, those who feel that the talent has escaped them are not without re-
course. Scholars who maintain a highly theoretical orientation are continually
offering suggestions for others to empirically explore. Schein (1975), for exam-
ple, has proposed a linkage between "Theory Y" and the contingency view of
organizations. He argues that both (middle range) theories recognize that a broad
range in individuals and organizations exists and that one must evaluate the total
situation (Le., the environment, technology, aspirations of individual workers)
before implementing a managerial or organizational plan. This analytical link
could provide the impetus for a more general contingency theory of organiza-
tional behavior.
SUMMARY
The middle range strategy, with its dual emphasis on theoretical generalizability
and empirical confirmation is a reasonable approach for organizational re-
searchers to follow. The advantages noted speak for themselves while the limita-
tions reviewed serve as reminders to implement the strategy carefully. It seems
particularly appropriate and a reasonable compromise for organizational scholars
who have traditionally expressed a preference for abstract theory or empirical
research. We feel that research on certain organizational topics (e.g., effective-
ness, climate) has stagnated somewhat because of a failure to move to an inter-
mediary level represented by middle range theory. Perhaps this strategy with its
broader outlook and ability to discern inconsistencies, can extricate these topics
from the myriad of contradictory fmdings that characterizes them. We hope the
suggestions for building middle range theory developed in this paper will be use-
ful in these endeavors. We also hope that once these theories are acquired,
organizational researchers will not stop their efforts but will seriously pursue the
goal of consolidation. To leave middle range theories as disjoint, specialized
theories is to defeat the ultimate purpose of the strategy and to eradicate its
greatest potential contribution. We recognize, however, that this task will

not be easy. The development of techniques and guidelines for consolidating
middle range theories should be among our foremost research priorities.
Success in meeting this objective would extend far beyond the confmes of
organizational research.
REFERENCES
Alderfer, C.P. An empirical test at a new theory of human needs. Organizational

Behavior and Human Performance, 1969,4, 142-175.
Bierstedt, R. Sociology and humane learning. American Sociological Review,
1960,25, 3-9.
Blalock, H.M. The measurement problem: a gap between the languages of theory
and research. In H.M. Blalock (Ed.), Methodology in Social Research. New
York: McGraw-Hill, 1968.
Blau, P.M., and Schoenherr, R. The Structure of Organizations. New York: Basic
Books, 1971.
Fiedler, F.E. A Theory of Leadership Effectiveness. New York: McGraw-Hill,
1967.
Hage, J. Techniques and Problems of Theory Construction in Sociology. New
York: Wiley, 1972.
Hall, R.H.; Haas, J.E.; and Johnson, N.J. Organizational size, complexity and
formalization. A merican Sociological Review, 1967, 32, 903-912.
Kaplan, A. The Conduct of Inquiry. Scranton, Pennsylvania: Chandler, 1964.
Wiley, 1966.
Kimberly, J.R. Organizational size and the structuralist perspective: a review,
critique, and proposal. Administrative Science Quarterly, 1976,2i, 571-597.
Lawler, E.E., and Porter, L.W. The effect of performance on job satisfaction.
industrial Relations, 1967, 7, 20-28.
Merton, R.K. Social Theory and Social Structure. New York: Free Press, 1949.
- - . On Theoretical Sociology. New York: Free Press, 1967.
Morrow, P.C. Functionalism, conflict theory and the synthesis syndrome in
sociology. International Review of Modern Sociology, 1978, 8, 209-226.
Pennings, J.M. The relevance of the structural-contingency model for organiza-
tional effectiveness. Administrative Science Quarterly, 1975, 20, 393-410.
Pennings, J.M., and Goodman, P.S. Towards a workable framework. In P.S.
Goodman and J.M. Pennings (Eds.), New Perspectives on Organizational
Effectiveness. San Francisco: Jossey-Bass, 1977.
Popper, K. R. The Logic of SCientific Discovery. London: Hutchinson, 1959.
Ritzer, G. Sociology: A MUltiple Paradigm Science. Boston: Allyn and Bacon,
1975.
Schein, E. The Hawthorne group studies revisited: a defense of theory Y. In E. L.

Cass and F.G. Zimmer (Eds.), Man and Work in Society. New York: Van
Nostrand Reinhold, 1975.
Turner, J.H. The Structure of Sociological Theory. Homewood, Illinois: Dorsey,
1974.
19, 357-367.
Zetterberg, H. On Theory and Verification in Sociology. Totowa, New Jersey:
Bedminster, 1966.
4
CHARACTERISTICS
OF MIDDLE RANGE
ORGANIZATIONAL THEORIES
AND THEIR IMPLICATIONS
FOR OPERATIONALIZATION
AND TESTING
A Conceptual Analysis
with Empirical Illustrations*
Merton's original defmition of theories of the middle range was relatively

straightforward, and remained almost unchanged from 1949 to 1968. Middle
range theories are ''Theories that lie between the minor but necessary working
hypotheses ... and the all-inclusive systematic ... unified theory .... Each ...
is more than a mere empirical generalization .... (It is) a set of assumptions from
which empirical generalizations have themselves been derived" (Merton, 1968,
pp. 39,41). Despite being one of the most widely cited ideas of the most often
cited American sociologist, however, the concept has had relatively little system-
atic development, and most of that has been by Merton.
While the above defmition tends to emphasize the descriptive nature of the
concept, Merton also had a clear prescriptive intent. The descriptive aspect
·This paper is based in part on a doctoral dissertation submitted to the University of

Minnesota. Preliminary analyses of some of the data were presented at the forty-eighth
annual meeting of the Midwestern Psychological Association, May 8, 1976. This research
was supported in part by Grants GS28353 and GS1927 from the National Science Founda-
tion, Karl E. Weick, principal investigator. Support for the data analysis was provided by
Northwestern University.
The author gratefully acknowledges the useful comments of David Pierson, Jane Rose,
and MaryBeth Beres regarding earlier versions of this paper. Revised June 8, 1978.
45
appeared in the characterization of middle range theories (MRTs) as being con-

cerned with specific, limited classes of phenomena, being of moderate abstract-
ness, and the like. Prescriptively, however, Merton stated quite strongly that the
use of MRTs would be especially productive in advancing science, relative to use
of either grand theory or empirical generalizations alone. This also implied a
truce in the long-standing debate between champions of inductive and deductive
approaches to theory.
This paper, likewise, has both descriptive and prescriptive components. De-
scriptively, an attempt will be made to systematize several aspects of the mean-
ing of middle range theory, and to evaluate that attempt by applying it to several
organizational theories and to a particular attempt to test one such theory.
Prescriptively, conclusions will be drawn from that application, that certain of
the descriptive aspects are particularly crucial to the ope rationalization and
testing of a theory.
CHARACTERISTICS OF MIDDLE RANGE THEORIES
Perhaps the very straightforwardness of Merton's defmition of MRT, cited above,

accounts for the frequency with which other authors have quoted it, in whole
or in part. However, once past the quotation, these others have added a con-
siderable variety of other content to their own defmitions. However, they have
usually treated these additions as synonymous with Merton's defmition, implic-
itly, regardless of the actual degree of relationship between them.
At least ten different characteristics have appeared in the literature as com-
ponents of definitions of MRT. These may, tentatively, be divided into: (I.)
those which are primarily quantitative, and (II.) those which are primarily
conceptual. These categories themselves may each be subdivided into two
subcategories, also tentatively. The quantitative characteristics include: (LA.)
those which are based on counting laws, or theoretical objects, and (LB.) those
which are based on counting real-world, or content, objects. The conceptual
characteristics include: (ILA.) relational characteristics, those involving the
ratio of one count to another, and (ILB) those characteristics which are essen-
tially qualitative. The characteristics are as follows.
LA. Quantitative characteristics based on laws

1. The number of laws, or empirical generalizations, explained by
the theory (Abel, 1970, p. 220); this is implied by the Merton
definition above.
2. The variety of (qualitatively different) laws or empirical general-
izations explained by the theory (Abel, 1970, p. 220; Dubin, 1969,
CHARACTERISTICS OF MIDDLE RANGE ORGANIZATIONAL THEORIES 47
p. 141) referred to homogeneity of the phenomena explained,

while Merton stated the MRTs "deal with delimited aspects of so-
cial phenomena" (1968, pp. 39-40), "limited conceptual ranges"
(1968, p. 51) or "limited ranges of data" (1957, p. 9). Kolb
seemed to go beyond Merton in emphasizing this point: "Middle
range theory, then, must mean not a middle level of abstraction
but rather theory at the highest level of abstraction in dealing
with social systems but concerned only with selected aspects of
those systems" (1958, p. 545).
3. The number and/or types of units of analysis covered by the
theory. At one point, Kaplan referred to this as the field of a
theory, defined as the subject matter, the universe of discourse,
or the elements of the theory, all three being essentially synony-
mous (1964, pp. 94-96). At another point, however, this same
idea, what the theory is about, was referred to as the range of
the theory (p. 301). Dubin's discussion of boundary-determining
criteria, which limit both the units covered and the laws of inter-
action among them (1969, p. 138) supports this defmition.
Gibbs (1972, pp. 67-68) also referred to the number of types of
events or things (units) as the range of the theory.
4. Number of properties of the units of analysis that enter the the-
ory. Merton's statements about MRTs dealing with "delimited
aspects of social phenomena" (1968, pp. 39-40) point to this
characteristic. Kaplan referred to this early in his book as the range
of the theory, and identified it with an attribute space within
which the units are located, and with a set of boundary condi-
tions limiting the range (1964, pp. 94-96). At a later point,
however, Kaplan referred to the size of the set of attributes or
properties as the scope of the theory (1964, p. 301), a usage with
which Gibbs (1972, p. 67) concurred. Hage (1972, pp. 178-179)
considered range, scope, and generality to be synonymous, adding
to the terminological confusion.
5. Size of the set of properties and relationships making up the laws.
Kaplan distinguished this aspect of a theory from the previous
one (1964, p. 95), and later reinforced the distinction by noting
that scope refers, not to "what things the laws are about, but ...
what the laws say about those things" (1964, p. 301). Dubin's
concept of the boundary-defining criteria that define the homo-
geneity of domain of a theory (1969, p. 138) supports this point,
since the criteria limit the number of laws of interaction among
the units. However, this component of the definition clearly
overlaps the previous one, as well as the first on the list. The lat-
ter overlap suggests the difficulty of categorizing these character-
istics unambiguously.
II.A. Conceptual, relational characteristics
6. Generality 1: "How many of the basic problems in the discipline
or specialty are handled by the same theory" (Hage, 1972,
p. 178). While this criterion is somewhat similar to no. 2, above,
it is stated in a manner that seems to imply derivation from, or
reference to, grand theory, rather than derivation from, or refer-
ence to, data.
7. GeneralitY2: Ratio of derived to primitive terms in the theory.
"The higher the ratio of derived to primitive terms, the wider the
scope of the theory ... (or its) generality" (Hage, 1972, pp.178-
179).
[LB. Conceptual, qualitative characteristics
8. The clarity and precision of theoretical terms, postulates, axioms,
and operationalizing rules (Abel, 1970, p. 220). Merton made this
point indirectly, contrasting MRTs with supposed theories which
are, rather, "general orientations," lacking "clearly formulated,
verifiable statements of relationships between specified variables"
(Merton, 1968, p. 52).
9. GeneralitY3: Abstractness. MRTs contain analytical concepts more
general than empirical generalizations (Abel, 1970, p. 220; Dubin,
1969, pp.141-144;Merton, 1968, p. 41), but from which testable
hypotheses, or empirical generalizations, may be derived directly
(Merton, 1968, pp. 41, 68). Kaplan (1964, p. 301) called this ab-
stractness, the length of the reduction chain from theoretical
terms to observable ones.
10. Development from data, rather than from grand or total theories.
Merton emphasized this quite strongly: MRTs "have not been
logically derived from a single all-embracing theory of social sys-
tems, though once developed they may be consistent with one"
(1968, p. 41), or more than one (p. 43). The point was reinforced
by arguing that grand theory in physics developed by integrating
MRTs. Finally, he insisted, theoretical progress can only be made
by developing MRTs and then, by combining them, "evolving,
not suddenly revealing, a progressively more general conceptual
scheme that is adequate to consolidate groups of special theories"
(1968, p. 51). Despite the vigor with which Merton made this
assertion, others blandly ignored it. Dubin, for instance, stated
"In which direction the model building goes probably does not
make any difference" (1969, p. 143).
Merton lists several other characteristics of MRTs that are not included in the
above list, on the grounds that they do not distinguish MRTs from other types
of theory. For instance, while MRTs are used to guide empirical research (Mer-
ton, 1968, p. 39), so are general theories and even empirical laws. Similarly,
consistency with more than one general theory (Merton, 1968, p. 43) is a char-
acteristic shared with empirical laws.
Merton's defmition has the advantage of simplicity, but the disadvantage of
vagueness. It is not entirely clear whether the various supplements to that
definition have clarified matters, although each offers a variety of specificity.
While the ten characteristics do seem to be related to one another, they are
by no means reducible to a single characteristic, or even to a much smaller set.
Thus, while we might reasonably expect the number of types of units of analy-
sis to be related to the number of empirical generalizations, it is not the only
determinant of the number of empirical generalizations. The terminological
confusion involved, extending even to inconsistencies between different chap-
ters of Kaplan, obscures the issue but does not affect the substance. All of
these characteristics seem at least potentially important.
Furthermore, there does not seem to be any obvious principle by which
one of these characteristics may be preferred to another for the purpose of
establishing a working definition of an MRT. Nor is it immediately obvious
how these ten characteristics might be combined to yield a single index of
middle rangedness. Merton asserted that the search for total systems of socio-
logical theory is "premature and apocalyptic ... (we) are not ready" (1968,
p. 45), and that in sociology, as presumably in related disciplines, pursuit of
MRTs will be more productive (p. 51). This may be taken to be an hypothesis,
testable using unobtrusive indicators of scientific productivity in the manner
of other hypotheses in the history and sociology of science. However, it will be
extraordinarily difficult to make such a test without having an operational defi-
nition, or a single index, of middle rangedness. Perhaps such an index can be
developed during the discussions of this conference.
An Example of a Middle Range Theory:

Weick's Information-Theory Approach
Among the multitude of systems-type theories of organization, that of Weick

(1969) was unusual in several ways, not the least of which was that it was a type
VI
o
+
Feedback of information from retention
~---------------------------process to enactment process; may be treated
r----------------------------process
as high or as low in equivocality.
+ Feedback of information from retention process

~to
~otor selection process; may be treated as high]
1
1 or as low in equivocality.
Environment Enacted environ- Output/input of Output/input of Endpoint item

= input of
ment = equivocality equivocality of equivocality
equivocality.
~~~!ro:a~ity l;evel ,-a l;evel x\-a-b Ie/vel = X-a-b-c
~~~t~=aiitY '~' \;a ';;r~' ,\;-b le;;r~ = X-a-b-c
~ - - 1- - - -
Assembly Equivocality Assembly Equivocality Assembly Equivocality removed
rules removed= a rules removed b rules !I = a
activated
activat\~ L
L+ activated
activa~~ ~J activated
activ~~d +
\- + \- 1+ \- +
Interl cked Interlocked Interlocked
behaviors behaviors behaviors
selected selected selected
ENACTMENT: Individual SELECTION: Application RETENTION: Storage

and group processes of collective criteria comparison, integration
of attention and activity to the informational synthesis and protection
directed at the environment, input from the previous of new input and previously
producing variation. process. retained information
1. The Weick Organizing Model.
Figure 1.
of middle-level theory. It dealt with a limited range of phenomena, informa-

tional ones, and had other characteristics of a middle range theory, which will
be described in more detail below. In addition, it was unusual in attempting
to make explicit the linkages between processes at the system level and at the
level of interaction of the individuals involved in the organization. It was also
unusual in emphasizing, not merely adaptation to environmental change, but
evolution of the organization.
The framework of the model was derived from social-evolutionary theory
(Campbell, 1965). The core of the theory consists of three cyclic organiza-
tional processes: (1) encountering informational variations in the environment,
(2) selection of some of the variations so encountered, and (3) their retention.
The environment in question is an enacted environment, generated by atten-
tional processes of organization members, and may thus consist in part of ma-
terial that is much more structured, and much less diverse, than that which
makes up the totality of the environment proper. The selection process provides
both for organizational stability and for organizational change, by either reject-
ing or accepting a variation. The retention process consists not merely of storage
of input from the selection stage, but also of processes that organize and inte-
grate the input, and which may also feed back to earlier stages of the process.
Figure I depicts the Weick model.
The informational inputs to the organization are equivocal (or uncertain), and
organizational processes serve to remove some of this equivocality. It is crucial
that the organization itself be complex, or equivocal, enough to deal with the
level of equivocality in its input. Within each of the three stages of the cycle, a
similar set of processes accomplishes this work. Equivocality is registered and
adjusted to by the differential activation of standardized processing rules; the
more equivocal the input, the fewer of these rules can deal with the input. The
rules specify the assembly and application of particular types of interactive be-
haviors that will reduce the eqUivocality of the input; the more rules are applied,
the fewer the behaviors that can satisfy all of the rules. And finally, the fewer
the behaviors applied, the less equivocality is removed from the input. Thus, low
equivocality input activates many rules, is subject to few interlocked behaviors,
and has relatively little equivocality removed.
Having asserted, above, that the Weick theory is an MRT, it is now appro-
priate to demonstrate that this is the case, or perhaps more accurately, the
extent to which this is the case.
INFORMATION THEORY AS A MIDDLE RANGE THEORY
Lacking any single indicator of middle rangedness, it seems appropriate to

evaluate a theory separately on each applicable characteristic. To illustrate this
process, several of the least redundant characteristics have been chosen. In each
case, information theory in general, and the Weick theory in particular, will be
compared with a theory that is clearly a general theory, systems theory as
represented by Miller (1971) and Katz and Kahn (1966), and also with a clearly
middle range theory, so labelled by Merton (1968, p. 64), cognitive dissonance
theory (Festinger, 1957). A listing of the applicable characteristics and evalua-
tions of relevant theories follows.
LA.l. Number of empirical laws. Miller listed 36 hypotheses, which had the
form, and, he suggested briefly, the support, to be considered empiri-
cal laws. Of these, 24 related more or less directly to information.
Festinger's data chapters listed something over a dozen empirical laws
that were explicitly explained by dissonance theory. Weick listed
about a half dozen such empirical generalizations in a somewhat less
explicit way, but also listed numerous theoretical propositions ex-
plained by his version of information theory, some of which may be
considered stand-ins for empirical laws. Thus, with respect to this
characteristic, information theory appears middle range by compari-
son with general systems theory, and approximately equivalent in
range to dissonance theory. A comparison on variety (characteristic
2) produces similar results.
LB.3. Number of units of analysis. Information theory is clearly more
middling than systems theory on this characteristic; Miller listed 19
subsystems of which nine are informational, whereas Katz and Kahn
listed 5 subsystems of which one (the managerial) is primarily con-
cerned with information. Weick and Festinger, on the other hand,
seem quite comparable on this characteristic. Weick's theory involves
about ten units, such as individuals, rules, behaviors, and so on. Fes-
tinger's theory involves about eight units, including cognitive ele-
ments, individuals, behaviors, and so forth. Thus Weick's theory is
clearly middle range on this basis. A comparison on number of prop-
erties (characteristic 4) produces similar results.
ILA.6. Generality 1: basic problems. The difficulty with applying this crite-
rion of middle rangedness is to define the basic problems of the dis-
cipline, since presumably different general orientations to a field may
identify different sets. However, it is possible to compare systems
and information approaches in Katz and Kahn; their latter chapters
were given over to seven "genotypic categories of phenomena" (1966,
p. 453), of which only one is informational. It is possible to compare
the seven components of organization theory discussed in Weick's
third chapter with the five major areas of theory and data discussed
by Festinger's later chapters, although the latter seem somehow a bit
less basic. Thus, information theory is less general! than systems

theory, and may be equivalent to or slightly more general! than
dissonance.
II.B.8. Comparisons regarding clarity, precision, and specificity must be
made subjectively. There appears to be no difference in clarity, and
so forth, between the general and the informational components of
Miller and of Katz and Kahn. However, Weick appears somewhat less
precise and less specific than Festinger, suggesting that Festinger is
more middle range on this characteristic.
9. GeneralitY3: abstractness. This characteristic is difficult to evaluate
except in cases of theories presented in explicit, formal, proposi-
tional style. However, there seems to be no important difference in
abstractness between general systems and information theories. On
the other hand, Weick's information theory is somewhat more ab-
stract, further from observables, than is dissonance theory, and is
therefore less middle range.
10. Development from data up. Although Miller, Weick, and Katz and
Kahn all cited data to illustrate their theories, all were clearly de-
veloped deductively, downward. However, Festinger (1957, pp.
v-vii) was explicitly developed by working up from empirical gen-
eralizations. Thus, only dissonance theory qualifies as middle range
in relation to this characteristic.
In summary, both by differentiation from general systems theory, and by

similarity to a recognized MRT, information theory may be considered middle
range in terms of six of the ten characteristics. No comparison is possible on one
of the characteristics, and information theory appears not to be middle range
with respect to three of the characteristics. While the balance may appear to be
in favor of considering information as an MRT, it will be argued below that the
three characteristics on which it fails to qualify are crucial.
OPERATIONALIZING AND TESTING WEICK'S THEORY
Gilfillan (1978) represented an attempt to test a number of hypotheses derived

from Weick's information-theory approach to organizations. As such, the study
illustrates the problems in operationalizing aspects of a certain type of middle
range theory, and these problems, in turn, suggest the importance of some of
the characteristics of middle range theory listed above.
Two similar big-band jazz orchestras were studied, for a period of three aca-
demic quarters, using structured observational methods to record a number of
items of data on each instance of playing, as well as on other events. Inputs to
these organizations were new written arrangements, which were equivocal both
in the sense of being made of unknown notes, and of being related in unknown
ways to the jazz idiom, standards of musical quality, and the bands' favored
styles of playing. Since each musician received only one-twentieth of the music
written down, this was very clearly a case of generating an enacted environment;
the tune had to be played at least once before it could be known as a whole. Fol-
lowing this initial playing, the first selection process was invoked; the arrange-
ment might be rejected if it was either very high in equivocality (chaotic or
senseless) or very low (banal) (Meyer, 1957). If the arrangement was retained,
it was repeatedly cycled through the organizational processes, removing the
unfamiliarity of the notes, and later the more subtle equivocality involved in
interpretive questions, until the number was considered ready for public per-
formance, which was the principal dependent variable.
Four hypotheses were derived from the foregoing analysis, which could be
tested given the assumption that the new arrangements comprising the input to
the orchestras were of approximately equal equivocality, so that outcome dif-
ferences would reflect primarily variations in organizational processing.
Hypothesis 1. The more an arrangement is fed back through the organizational

processes, the more likely it is to be made ready for public performance, by the
removal of sufficient eqUivocality. This cycling is of two different types. One is
the repetition of the arrangement within a given rehearsal (Hypothesis 1b). The
other is returning to the arrangement at a later rehearsal, which may allow for
reconsideration, unconscious creative cognitive work, and the like, to occur
in the interim (Hypothesis la).
Hypothesis 2. Playing a tune in full will be more effective in removing equivo-

cality than playing in bits and pieces, because it brings more eqUivocality into
the processing. Especially after the basic, technical issues (what are the notes?)
have been resolved, the eqUivocality centers around interpretive issues. Because
these issues tend to involve the relationships of each part of the arrangement
to various others, and of each instrument to several others, processing will be
more effective when the whole arrangement is played (Hypothesis 2a), and when
the playing is by the full ensemble (Hypothesis 2b).
Hypothesis 3. While individual, ad-hoc playing may be useful to the individual

player, it will interfere with processing by the interlocking behaviors of the orga-
nization as a whole, interrupting and drowning out other work, preventing
restarting, and the like. There are two types of such a spill: the orchestra (or a
member) may continue playing after a halt has been called (Hypothesis 3a), or
an individual may begin playing something of interest to him or her, during a
pause (Hypothesis 3b).
Hypothesis 4. The more the discussion time devoted to the arrangement, the
more likely it is to be perfected for performance. If equivocality is relatively
high, and few rules are invoked, many behaviors will be activated, including both
playing and discussion. This should be especially true when interpretive, aes-
thetic issues come to the fore.
Further details of the research methodology, the data analysis, and the results
of this study have been reported elsewhere (Gilfillan, 1978). These will be men-
tioned below only as they bear on the question of the status of the Weick theory
as a middle range theory.
Three Difficulties in Operationalizing and Testing Weick's Theory
l. Clarity and Precision in Measurement of Information. New arrangements

that were input to these organizations were high in equivocality or uncertainty,
and correspondingly low in information. Since a new arrangement might contain
dozens or even hundreds of bars, and each bar might allow for 16 notes, spread
over a range of several octaves, it is clear that the number of potential combina-
tions of notes in an unknown score would be essentially infinite, as would its
uncertainty. Yet, this was the least important type of equivocality, and was
completely resolved into information within a few rehearsals, as the orchestra
members learned to play their parts letter-perfect.
It seems reasonable, at least, that this simpler type of equivocality could be
measured exactly. Information theorists have devised methods for the informa-
tional analysis of music using computer technology. A body of similar melodies
is analyzed as a Markov chain, probabilities being worked out for each of the
several notes that might follow a given note or combination of notes. Having
done this, the computer can generate melodies that are usually, not always,
judged acceptable. However, there are several flaws in this approach. It only
works for very simple types of melodies (lullabies and hymns), or for very struc-
tured arrangements of a given melody (16th century counterpoint). It takes no
account of the information generated between the parts of long and/or complex
musical pieces, of the effect upon the listener, or of the possibility of changes in
the source that generates the chain. Cohen (1962) has criticized a variety of as-
pects of this approach. Other approaches (Kraehenbuehl and Coons, 1959;
Moles, 1966; Meyer, 1957) have dealt with the information content of music in
other ways, and the relation of this information to musical meaning, the latter
involving the current state of the perceiver in the calculation of probabilities and
expectations. None of these has provided an acceptable measure, or set of mea-
sures, which could be used to measure the equivocality of musical input, or the
equivocality reduced by organizational processes (Galbraith, 1977, p. 38).
Lacking such a measure, two compromises were forced upon the research
design. First, it was necessary to assume the approximately equal equivocality of

input arrangements, if variations in outcome were to be attributed to organiza-
tional processes rather than preexisting differences in arrangements. The data
demonstrated this assumption to be false, as a result of which the sample of
input arrangements had to be restricted by post-hoc methods to the tunes that
were taken seriously by the orchestra. Second, it was necessary to use a judg-
ment of readiness for public performance as a surrogate for an information-
generated (or eqUivocality-removed) measure of outcomes. It was not an
unreasonable substitution, but a more direct, less subjective measure would
clearly have been preferable.
It should also be noted that the orchestras under study had only one principal
input, arrangements. Turnover was low, so that input of personnel could be
ignored. Other support was provided by the university (e.g., space) in such a way
that it constituted merely a non demanding environment. Most other organiza-
tions have a variety of inputs in a variety of categories, obviously making the
specification of their input equivocality much more difficult.
This problem is essentially a problem of the weakness of information theories
of organization with respect to characteristic 8 of midrange theories: clarity and
precision. The concept of information has an absolutely clear and precise mean-
ing in the context in which it originated. The Shannon (1964) formula for H
provides an exact mathematical measure of the information contained in a mes-
sage consisting of, for example, electrical impulses. Furthermore, this method
may be profitably extended to certain other areas where the issues are essentially
dichotomous; Phillips and Conviser (1972), for instance, used a measure related
to H to define group membership/nonmembership on the basis of individuals'
participation or non participation in various activities. Obviously, the method
would be in difficulty if degrees or amounts of participation were the issue.
More generally, the extension of the concept of information to cover situations
where the informational issues are highly complex and multivalent reduces the
preciseness of the concept, and therefore, its operationalizability. Thus, the
fact that the theory did not satisfy the formal characteristic was reflected in
concrete difficulties in operationalization.
2. Problems Related to Grand-Theory Origin. Weick's theoretical model dealt

with the assembly of organizational processes by interlocking the behaviors of
(implicitly) homogeneous members. It described three organizational processes,
but did not even identify them with subsystems, the usual approach to structural
differentiation in systems theory. Similarly, although the three processes each
were presented as separate equivocality-removing operations, they were de-
scribed as identical in form, and any operating differences that might result from
their different functions were left unstated. Thus, it is reasonable to say that the
theory contained several suggestions that the organization as whole is the pri-
mary equivocality-reducer. This organismic assumption is shared with many
other systems approaches, and has been frequently criticized (e.g., Dahrendorf,
1958, p. 125).
One result of this orientation in the theory was the generation of Hypotheses
3 and 2b, that individual or sectional playing would interfere with organizational
processing. These hypotheses were by no means counter-intuitive; both the
leader and other orchestra members often expressed annoyance at ad-hoc play-
ing and at rehearsing single sections. However, the data were disconfirming.
Some of the work that needs to be done to get any organizational product out
is individual or group work. Individual players must master difficult passages,
sections must adjust their loudness, and these processes, though they contribute
to the fmal goal, go on essentially independent of other parts of the organiza-
tion. A revised theoretical model, including individual and group equivocality-
reducing processes, was the result of these considerations.
These problems appear to derive from the fact that the theory was largely
deduced from a general framework, rather than built up, as suggested by charac-
teristic 10, from empirical generalizations. A theory deduced from another
general framework (e.g., Dahrendorfs) might well correct for these problems,
but at the risk of missing other aspects. On the other hand, it is highly unlikely
that a midrange theory built up from empirical laws could manage to slight the
contributions of individuals and groups to organizations.
The orientation of the theory was reflected in other ways, as well. First, the
research simply assumed the existence of a common organizational goal of
making new arrangements ready, rather than researching the goals, implicit and
explicit, held by the organization's members, shared or not. Second, an arrange-
ment consists of 20 individual parts, each with its own level of equivocality. This
fact was not operationalized in the research design, nor explicitly assumed (as
equivalence of arrangements was). Clearly, however, any thorough informational
approach would have to include an attack on the problem on a part-by-part basis.
3. Abstractness Problems. A third set of problems in the research may be re-

lated to the third characteristic of MRTs which did not fit the Weick theory,
number 9, generalitY3, abstractness, or the distance from theoretical concepts to
observables. ill operationalizing Hypothesis 1, the measure originally selected
as an indicator of cycling was repetition of the arrangement within the rehearsal
(now labelled Hypothesis I b). Preliminary data analyses indicated that this
variable behaved as predicted, but that another, repetition of the arrangement
at many rehearsals, explained more variance, leading to its inclusion in the form
of Hypothesis la. This was a problem of abstractness, in the sense that the
theory gave no guidance whatever in determining the length of a feedback cycle;
the original choice was made on the basis of other considerations. This theo-
retical problem might also have been a serious empirical problem, had the
initial assumption led to the collection of data relevant only to the original
operationalization.
Other difficulties with operationalization also related to the problem of
abstractness. For example, the operationalization of the idea of multiple inter-
locked processing behaviors in terms of playing and talk (Hypothesis 4) was
clearly a rather coarse classification; however, it was not clear how to distinguish
finer categories of behaviors for this purpose. Similarly, some central aspects of
the theory remain unoperationalized, including the separation into three process
stages, and the notion of assembly rules. Making assembly rules observable
would involve gathering data on how the leader chooses, during each pause,
the sections of the band and the parts of the arrangement to be played next,
how the various members decide upon various twists of interpretation, and so
on. These may be very difficult to observe, and in fact may be largely unver-
balizable, expressible only by playing or singing. In general, operationalization
in this study depended to a very great extent upon detailed knowledge of music
and orchestras, and even with that knowledge often went astray. While all
operationalization requires knowledge of the concrete research situation, the
knowledge required may have been greater, and the outcome less reliable,
in this case than in the case of the usual MRT. Thus, the high level of abstract-
ness, or of generalitY3, of the Weick theory, caused difficulties both formal and
operational.
CONCLUSION
While the Weick theory has the majority of the ten characteristics of an MRT,
it does not have three. While one might be inclined to adopt a majority decision
rule, and to classify it as a theory of the middle range, the foregoing analysis
indicates that this would be a mistake. With the appropriate reservation, that
this analysis has been based on a single theory and a single research study, it is
clear that the characteristics listed are not merely formal, that they do have
substantial impact on the testability of a theory. Further, since problems of
abstractness, clarity, and top-down derivation seem to produce somewhat dif-
ferent difficulties, my previous impression that they are each important, and are
not synonymous or reducible to a single characteristic, is also supported. To
qualify as a middle range theory, then, a theory must meet a substantial number
of criteria, not just one or two. Not every limited theory is a theory of the
middle range; it seems possible, though this analysis has not demonstrated it
conclusively, that the conceptual, and especially the qualitative characteristics
(type II.B.) are somewhat more important than the others.
It is especially tempting to believe, with Dubin, that characteristic 10, build-

ing up from empirical generalizations, is unimportant. To work in the other
direction seems to offer tremendous advantages in the form of cross-field gener-
alization, but the disadvantages outweigh them. Merton has contrasted two types
of grand theories, those built up after decades or centuries of empirical laws and
MRTs, and those generated quickly, de novo. The latter, he suggests, might bet-
ter be called "general orientations toward data" (1968, p. 52), or approaches;
others have referred to these as "sensitizing concepts" (Blumer, 1954, p. 47)
or "taxonomies" or "frames of reference" (Zetterberg, 1965, p. 29). These are
useful in that they suggest new issues and insights. In the present case, informa-
tion theory certainly stimulated a richer and more meaningful approach to input
arrangements than regarding them as material to be learned, in the traditional
learning theory sense, could have. However, these frames of reference should be
regarded, and used, suggestively, as analogies; time devoted to trying to devise
empirical tests for them may be more usefully spent on true MRTs.
This suggests that progress in organization theory will be something like
working out on a trampoline. A big push at the beginning is futile; it merely
pushes down the canvas. Height is gained by repeated small bounces, up from
the data a little way, and then back down, over and over again. Eventually,
enough height is gained to enable one to do acrobatics. To be sure, the same
acrobatics might be done while skydiving, but the trampoline method avoids
the waste of time and gas going up in the plane, and the risk of floating grace-
fully down into a manure pile, or of streaming to a crash.
REFERENCES
Abel, T. The Foundation of Sociological Theories. New York: Random House,

1970.
Blumer, H. What is wrong with social theory? American Sociological Review,
1954,19,3-10.
Campbell, D. T. Variation and selective retention in socio-cultural evolution. In
H.R. Barrington, G.I. Blanksten, and R. Mack (Eds.), Social Change in Devel-
oping Areas. Cambridge, Mass.: Schenkman, 1965.
Cohen, J. E. Information theory and music. Behavioral Science, 1962, 7, 137-
163.
Dahrendorf, R. Out of utopia: Toward a reorientation of sociological analysis.
American Journal of Sociology. 1958,64. 115-127.
Dubin, R. Theory Building. New York: Free Press, 1969.
Festinger, L. A Theory of Cognitive Dissonance. Stanford: Stanford University
Press, 1957.
Galbraith, J. Organization Design. Reading, Mass.: Addison-Wesley, 1977.
Gibbs, J. Sociological Theory Construction. Hinsdale, Ill.: Dryden, 1972.
Gilfillan, D.P. Organizational Processes in Two Jazz Orchestras: Some Evidence

Regarding the Social Psychology of Organizing. Doctoral dissertation, Univer-
sity of Minnesota, 1978.
Hage, J. Techniques and Problems of Theory Construction in Sociology. New
York: Wiley, 1972.
Kaplan, A. The Conduct of Inquiry. San Francisco: Chandler, 1964.
Wiley, 1966.
Ko1b, W.L. Review of social theory and social structure by R.K. Merton. Ameri-
can Journal of Sociology, 1958,63, 544-545.
Kraehenbuehl, D., and Coons, E. Information as a measure of the experience of
music. Journal of Aesthetics and Art Criticism, 1959,17, 510.
Merton, R. K. Social Theory and Social Structure (1968 enlarged ed.). New
York: Free Press, 1968.
Meyer, L.B. Meaning in music and information theory. Journal of Aesthetics
and Art Criticism, 1957,15,412-424.
Miller, J.G. The nature of living systems. Behavioral Science, 1971,16,278-301.
Moles, A. Information Theory and Esthetic Perception. Urbana: University of
Illinois Press, 1966.
Phillips, D.P., and Conviser, R.H. Measuring the structure and boundary proper-
ties of groups: Some uses of information theory. Sociometry, 1972, 35,
235-254.
Shannon, C. E. The mathematical theory of communication. In C. E. Shannon
and W. Weaver. The Mathematical Theory of Communication. Urbana: Uni-
versity of Illinois Press, 1964.
Weick, K. E. The Social Psychology of Organizing. Reading, Mass.: Addison-
Wesley, 1969.
Zetterberg, H.L. On Theory and Verification in Sociology (3rd ed.). Totowa,
N.J.: Bedminster, 1965.
5
COMMENTARY
ON GILFILLAN,
MORROW- MUCHINSKY,
AND BLUEDORN-EVERED
The Circle of Inquiry
Louis R. Pondy,* University of Illinois
INTRODUCTION
Each of the three papers in this section of the book makes important contribu·
tions to our understanding of middle range theory. Gilfillan has made a valiant
attempt to make the concept of middle range theory operational by laying down
ten criteria that distinguish middle range theories from empirical generaliza·
tions on the one hand and from grand theories on the other hand. Morrow and
Muchinsky have argued that the ultimate goal of scientific inquiry is to develop
general theories, and that the use of middle range theories (MRTs) can play an
important role in this process through the gradual consolidation of MRTs into
theories of increasing generality. Bluedorn and Evered, in contrast to Gilfillan,
have rejected the notion of middle range theory as a well·defined absolute con·
cept and have suggested instead that middle rangedness is a relative concept and
that what is at issue is not the use of MRTs but the employment of a middle
range strategy for theory construction. Furthermore, they argue that the middle
range strategy is only one of several possible strategies, and that we in organiza-
*1 would like to express thanks to Patricia Templin and Roger Evered for helpful con-
versations during the preparation of this paper, and to Patricia Hollahan for assistance in
copy editing the final version of the manuscript.
61
tional behavior should avoid anointing the middle range strategy as the best and
only acceptable way of conducting scientific inquiry. Taken together these
contributions resolve certain problems, but they also expose certain conflicts
and gaps in our understanding of how to conduct inquiry into organizational
phenomena.
I should like to organize my own comments around: (a) one of the conflicts
among the three papers, (b) one gap not closed by any of the papers, and (c)
my own proposed formulation of the role that MRTs can play in organiza-
tional research. The conflict concerns whether middle range theory exists as a
well-defmed concept, or whether what exists instead is a middle range strategy.
The gap arises from an inadequate discussion of what "progress" or "advance-
ment" means in scientific inquiry. My proposed formulation is called the circle
of inquiry and attempts both to resolve the conflict and to fill the gap men-
tioned above.
Throughout my preparation of these comments, my posture has been future-
oriented: What can we say and do here that will help to make organizational
research more fruitful in the future? I have resisted the temptation to spend time
trying to determine what Robert Merton really intended by the concept of mid-
dle range theory. Merton represents only one of organizational behavior's parent
disciplines, with his own set of biases, problems, and concerns. It is time for us
in organizational behavior (O.B.) to solve our own problems, rather than import-
ing prefabricated solutions. This commitment to authenticity-to be authors of
our own scripts-obligates us to examine in depth the nature of our subject mat-
ter and our relationship to it. Let us be "phenomeno-Iogical" in the literal sense
of the word, rooting our logic in the nature of our phenomena, not borrowing
a logic from some other domain.
Before getting on with the heart of my comments, let me share with you my
two greatest fears about the outcome of this examination of middle range theory.
One fear is that we will take middle range theory too seriously; the other is that
we will fail to take it seriously enough. We will have taken middle range theory
too seriously if we outlaw any approach to inquiry that is not a middle range
theory approach. We will not have taken it seriously enough if we change only
our language but not our habits of inquiry, merely relabelling all of what we
have been doing as middle range theory. Our purpose in this endeavor ought to
be to develop an optional strategy, middle range theory, for organizational re-
search, not to argue that it is the only acceptable strategy.
THE CONFLICT: MIDDLE RANGE THEORY OR

MIDDLE RANGE STRATEGY?
Our field has a pernicious habit of reifying what we study. We tend to treat
organizations as if they were objective things or entities with persistent and
COMMENTARY ON GILFILLAN, MORROW-MUCHINSKY 63
measurable properties. As a consequence, we miss the fact that organizations

are socially constructed realities, collections of processes that do the work of
organizing. Because we have succeeded in devising measurements, we have de-
luded ourselves into thinking that properties exist which correspond to those
measures and that entities exist which exhibit those properties. Thorndike has
often been quoted as saying, "If something exists, it exists in some quantity and
if it exists in some quantity, it can be measured." What is seldom pointed out is
that the converse is not true: The fact that a measure exists and is alleged to
be a quantitative description of some thing does not necessarily imply that that
thing exists.
Reification could haunt our examination of middle range theory because it
appears to be a general difficulty with our patterns of thought. This is revealed
most graphically in the question addressed by these three papers: "What is
middle range theory?" The fact that a syntactically meaningful question can be
formed using the phrase 'middle range theory' does not mean that middle range
theory exists, either as an objective entity or even as a well-defIDed concept. A
more appropriate question would be: "Does the concept of 'middle range the-
ory' make sense at all?" I should like to suggest that the answer to that question
is "no." This stand places me on the side of Bluedorn and Evered and in op-
position to GilfIllan (and, to a lesser extent, to Morrow and Muchinsky). It
seems apparent from GilfIllan's analysis that he believes theories are entities,
objective things, with measurable properties. Indeed, he proposes a set of ten
defIDing characteristics.
There are special problems with the characteristics themselves. For example,
Gilfillan assumes that the number of empirical propositions derivable from any
given theory is well defIDed. But this is clearly absurd. Propositions can be framed
for each specific empirical setting, or each set of contingent conditions. Consider
the research on administrative intensity. It has been hypothesized that size and
administrative ratio are negatively correlated, but this hypothesis can be made
contingent on technology, ownership, and so forth, or can be tested for business
firms, hospitals, government bureaucracies, and so on. Is that hypothesized cor-
relation one hypothesis, or several? The answer is that it is not clear ... because
the question is founded on the false presumption that empirical propositions are
sufficiently well defIDed that they can be counted, as one might count apples or
rocks. Counting the propositions in a theory is more like trying to count the
number of ideas that could be evoked by someone's reading of a T.S. Eliot
poem. Instead, the propositions derivable from a given theory depend on the
person using the theory, on his or her creativity and insight, on the settings in
which the theory is interpreted, and so forth. So any of Gilfillan's defIDing char-
acteristics that depends on counting propositions is faulty.
But the more serious and fundamental flaw in Gilfillan's analysis is his initial
presumption that any given theory can be described absolutely and unambigu-
ously as middle range or not. He modifies this presumption slightly by arguing

that there is a continuum of "middle rangedness," to use his terminology, along
which theories can be arrayed. But this does not avoid the fundamental diffi-
culty arising from his implicit assumption, which I believe to be faulty, that
theories are stable, well-defined entities with measurable properties. This as-
sumption is tantamount to assuming that a given theory can be lifted out of the
context of inquiry, out of its place in the historical development of theory, and
out of its place in the stream of empirical plus theoretical thought and given a
meaning as an isolated objective thing rather than as an integral part of a richly
interconnected set of concepts and processes.
But if theories are not entities, what are they? I would like to propose that
we think of theories as stages in a process of inquiry, themselves processes, so
that we should be speaking of middle range theorizing. In this sense, the middle
range strategy is a strategy of inquiry that makes use of an intermediate process
of theorizing that connects low level empirical facts or generalizations with more
abstract conceptualizations. We can refer to these abstract conceptualizations
as root metaphors or fundamental images of the subject matter. Any middle
range theorizing must be interpreted in the context of the root metaphor from
which it derives in order to be understood properly. For example, Weick's
(l969) theorizing about organizing clearly derives from the phenomenology of
Alfred Schutz (1967), the symbolic interactionist orientation of George Herbert
Mead (1956), the general systems and ecological orientations of Gregory Bateson
(l972) and others, and Ross Ashby's (1956) cybernetics. By lifting Weick's
theory out of that context, and trying to examine it as an isolated entity, Gilfil-
lan has failed to treat MRT as one stage in a process of inquiry linking empirical
facts with the abstractedness of root metaphor. He concludes that Weick's
theory is not middle range enough to be empirically tested. Does this mean that
Weick's theory is useless? Of course not. It merely means that work remains to
be done to link up Weick's insights with empirical procedures and with other the-
oretical developments. The terms in Weick's theory can not be operationalized
directly; they must be further specified in some concrete empirical context.
Bluedorn and Evered's treatment is more nearly consistent with my view of
middle range theorizing as a part of the entire process of inquiry. They specifi-
cally reject the notion that MRT is an absolute concept, and see it as only one
possible route from facts to grand theories. Furthermore, a theory in their view
can change status from grand theory to MRT as wider abstractions become
available. I shall have other disagreements with them to air later in my com-
ments, but on this we seem to be in agreement.
Morrow and Muchinsky occupy an intermediate position between Gilfillan
and Bluedorn-Evered. On the one hand, they seem to have a tendency to think
of any given theory as being unambiguously middle range or not. On the other
hand, they recognize explicitly the process of consolidating MRTs into grand
theories and so see MRTs in the context of inquiry as a whole, and they do seem
to recognize that MRT is more a process of theorizing than a well-defined entity.
Indeed, they use the phrase "middle range theorizing" in their paper, and place
considerable emphasis on the consolidation process. And in private correspon-
dence, Professor Morrow has indicated that they see MRT as one of only several
strategies of inquiry. So perhaps I am just overreacting to the fact that Bluedom
and Evered laid out a more systematic array of theory construction strategies
than did Morrow and Muchinsky.
To conclude this section, my judgment is that we should reject the notion
that middle range theories exist in some independent objective sense. Gilftllan
has provided a valuable service in trying to create measurable properties of MRTs
and in demonstrating that particular strategy is not a fruitful one. In my judg-
ment, no more of our resources should be invested in such efforts. Instead, let
us choose as one of the slogans of this investigation: "Middle range strategy, not
middle range theory."
THE GAP: WHAT IS "PROGRESS?" IN SCIENCE?
All three papers address the question of whether using MRTs will result in more
progress for scientific advancement. But none of them explicitly discusses alter-
native meanings of "progress."
I have inferred from Morrow and Muchinsky's paper that the ultimate goal of
scientific inquiry is the development of general theories, a goal shared by Merton
himself. Furthermore, Morrow and Muchinsky consider whether progress takes
place through the gradual consolidation of MRTs into more general theories.
Bluedom and Evered seem to be saying that theory construction is the aim of
scientific activity. Gilfillan is less explicit as to his defmition of progress. From
his example of Weick's theory, we might conclude that progress is the testing of
scientific theories. Taken together these defmitions of progress constitute a
limited definition of what we might take as the indices of scientific progress. If
we are to assert that the middle range strategy of inquiry, or some other strat-
egy, will permit us to make more progress, then we had better have a clearer idea
of what is meant by progress, and should address the issue frontally.
Let me begin the task by listing some possible indices of progress, and permit
me to make clear at the outset that my defmition of progress includes all of
these; the problem is not to choose which one really is the best single indicator
of progress. (In the next section, I propose a model of inquiry in which these
phases of progress are integrated.) I propose the following eight indices of
scientific progress.
I. The accumulation of factual knowledge about phenomena through

empirical research. Which facts to select for attention is determined by
the attention-directing property of theories. Theories tell us that some
facts are significant and that others should be ignored.
II. The accumulation of new interpretations of those empirical facts by
placing them within new theoretical frames, or by placing them in new
juxtaposition to one another. This reinterpretation proceeds through
secondary analysis, but it is distinct from using old data for mere em-
pirical testing of a new theory. The new theory gives the old data new
meaning.
III. The verification or testing of theories by confronting them with em-
pirical facts. Strictly speaking, what are tested are hypotheses derived
from theories. But in any case, what is at issue here is not the interpre-
tation of the facts, but the mere testing of hypotheses.
N. The forming or shaping of theories by observing empirical regularities
and subsequently by generalizing those regularities to a higher level of
abstraction.
V. The development or creation of more general theories. This may pro-
ceed through the consolidation process described by Morrow and
Muchinsky, or through other more direct processes identified by Blue-
dorn and Evered-either through a direct inductive leap from data or
through some poorly understood act of creative intuition. A good
example of the consolidation process is Homans' (1950) The Human
Group, which resulted from his interpretive integration of several
studies and theories of interpersonal and group processes.
VI. The specification of more nearly testable, less abstract theories from
very general theoretical orientations. What I have in mind here is tak-
ing a theoretical system stated as a root metaphor and concretizing it,
or de theorizing it. Weick (1974) has done something like this in his
Behavioral Science article by taking general systems theory and show-
ing that it could be made operational and less metaphorical. This is
middle range theorizing from the top down; it is the opposite of the
consolidation process.
VII. The development of the capacity for solving practical problems of
organization. This capacity may be tacit; that is, unlike the previous
six indices, it need not rely on making our knowledge or understand-
ing explicit, but may reside instead in the intuitive experience of
practitioners.
VIII. The development of a capacity to describe and to understand unique
concrete events and situations. This is more than the accumulation of
mere empirical facts. It is the assembly of integrated, interpretive de-
COMMENTARY ON GILFILLAN. MORROW-MUCHINSKY 67
scriptions of specific settings, what Clifford Geertz (1973) in The

Interpretation of Cultures has called "thick description" -ethnographic
techniques at their best and fullest.
This list of indices of scientific progress is more nearly complete than those cri-
teria considered in the three papers. Some might argue that we have omitted
certain purely instrumental indices such as the development of more reliable
measurement procedures, and I would agree that these indices are indeed purely
instrumental. The eight criteria I have listed have the status of goals of inquiry,
whereas the development of better measurement procedures is merely a means
to those goals. I do not yet wish to refer to them as individual goals, because
they take on meaning as goals only in the context of each other.
Before going further, I should like to discuss one very special issue about
progress that surfaced in the Morrow-Muchinsky paper. Their treatment of con-
solidation as a gradual process smuggles into the analysis an assumption of
scientific progress that is directly counter to Kuhn's (1970) theory of scientific
revolutions. Nowhere in the Morrow-Muchinsky analysis is there any room for a
radical shift in the underlying root metaphor from which the theories derive. In
a private communication, Professor Morrow said that she and Muchinsky con-
sidered the Kuhnian model but chose not to discuss it. She observed that the
middle range orientation is probably consistent with what Kuhn called the
"normal science" phase in which the reigning paradigm is not questioned. If
correct, this observation would mean that adopting a middle range strategy
exclusively would preclude the most fundamental and important forms of scien-
tific progress - the creation of new paradigms of inquiry. Morrow argued privately
that consolidating MRTs can highlight anomalies and thus help to precipitate
paradigmatic revolutions. But I am more convinced by her prior argument that
adopting the middle range strategy would facilitate progress within a normal
science mode, but would foreclose more revolutionary forms of progress. One
wonders whether Merton's methodological orientation toward middle range
strategy was not a reflection of his own substantive orientation toward func-
tionalism and gradualism.
THE CIRCLE OF INQUIRY
In this section, I attempt to place the middle range strategy of theorizing within
a broader context of scientific inquiry. In their paper, Bluedorn and Evered
present MRT as a strategy of theory construction, and thus, in their Figure 1, all
of the arrows of development run from left to right, from the empirical end
toward the theoretical end of the scope continuum. (What is seen as strategy
numbers 7 and 9 in their model appears to be what Morrow and Muchinsky term
consolidation.) Is there some way in which we can build upon and extend the
Bluedorn-Evered model to defme the ends of inquiry so that it includes theory
construction and other forms of scientific inquiry as well? (I wish to acknowl-
edge that Bluedorn-Evered did not assert that the aim of inquiry was limited to
theory construction; they were simply silent on the issue.) Theory is not only a
summary of empirical facts. It is also a guide to further inquiry; it tells us what
to look for and what to ignore. And it is a framework for interpreting the mean-
ing of empirical facts. These are my indices I and II in the list of indices of
scientific progress from the preceding section. Within the context of the
Bluedorn-Evered model, this suggests that arrows should also be running from
right to left, connecting theory back to the factual level of empirical generaliza-
tions. While the present version of the Bluedorn-Evered model suggests a uni-
directional movement toward more precise and broader scope theory, my
extension of their model suggests a circle of inquiry, flowing back and forth
between the presuppositions of theory and the observations of fact. (My circle
of inquiry derives from the concept of a "hermeneutical circle" as described by
Heidegger (1962) in Being and Time or Habermas (1971) in Knowledge and
Human Interests.) The circle is shown graphically in Figure 1, which follows.
The Roman numerals I, II, Ill, and IV correspond to the first four indices of
progress described in the preceding section. At this stage, no distinction is made
between MRTs and theories of higher levels of abstraction. Inquiry within this
model is the joint action of theorizing and empirical investigation. Emphasis
is placed on both processes working together, round and round, not in a linear
movement toward ever grander theory, but in a mutual process of using maps
(theories) to explore new territory (collect new facts), and improving maps in
accord with the results of exploration. The circle of inquiry suggests that we
are both explorers and map-makers working together.
Where does the middle range strategy fit into this scheme? It suggests that
the leap from facts to grand theories is usually too large to take in a single
bound and that interposing middle range theories as stepping stones between
facts and grand theories makes that transition more nearly feasible. But it also
suggests that grand theories are frequently too vague to serve as guides for select-
II: Interpretation
I
FACTS
I: Selection
I
THEORIES
I IV: Shaping
III: Verification
t
Figure 1. The Circle of Inquiry.
ing facts to observe and interpret because grand theories are too far removed
from concrete situations. Using MRTs provides a way of continuing to root
empirical inquiry in general theory while still providing a close link to concrete
events. Stepping stones are helpful for travelling in both directions. Not only
should we, according to this version of the middle range strategy, attempt to
consolidate those stepping stones into grand theories (my index V), but we
should also concretize grand theories down into generalizations more directly
useful for selection and interpretation of empirical facts (my index VI).
This expanded version of the circle of inquiry is depicted in Figure 2, which
follows.
Indices I through VI are represented in Figure 2 as transitions between fact,
middle range theories, and grand theories. It is more difficult to represent indices
VII and VIII on the circle of inquiry. Both have to do, in a sense, with the
empirical analog of grand theorizing; both represent intense probing, one practi-
cal, one descriptive, into some specific empirical setting. They might be de-
scribed as "grand empiricizing."
This line of thinking suggests that we should carry the parallel between facts
and theories a step further. Suppose we make an analytical distinction between
middle range facts and grand facts. By grand facts I mean intensive, contextually-
sensitive case descriptions of concrete situations - what Geertz (1973) calls
"thick descriptions." And let us define middle range facts as factual descriptions
in which the empirical context of those facts is not fully specified. Then placing
middle range facts into context and thereby creating grand facts is what I have
labelled index VIII in my list of indices of scientific progress. Index VII, devel-
oping a capacity for problem solving, can then be seen as using an understanding
of the factual context to shape and give meaning to specific factual descriptions
of the middle range. These ideas are summarized graphically in the third ver-
sion of the circle of inquiry in Figure 3.
MIDDLE
TH~;IEES~
~I: Concretization
Interpretation
GRAND
FACTS THEORIES
~:
Verification
Shaping Consolidation
MIDDLE
RANGE
THEORIES
Figure 2. The Circle of Inquiry, Expanded.
MIDDLE I: Selection MIDDLE

/ RANGE
FACTS • .
II: Interpretat~on RANGE -..........
THEORIES "
J' VIII: Thick Description VI: Concretization
\
GRAND GRAND
FACTS THEORIES
~ VII: Problem Solving V: consoli!ation

~ MIDDLE III: Verification
RANGE - - - - - - - - - - - - .
MIDDLE
RANGE
~
FACTS IV: Shaping THEORIES
Figure 3. The Circle of Inquiry, Third Version.
Given this general scheme, scientific progress can be defmed as completing

the circling of inquiry, traversing it repeatedly as our accumulation of factual
descriptions and our understanding of them becomes deeper and more refined.
The middle range hypothesis can then be restated as follows: Progress through
the circle of inquiry will be enhanced by the use of intermediate levels of both
fact and theory.
This proposal of a circle of inquiry is not meant to criticize the three papers
for incompleteness. Indeed many of the elements of the circle were found in
the papers, and the papers collectively evoked the idea of the circle of inquiry.
What the circle of inquiry has done so far is to place the concept of middle range
theorizing in a broader context than that suggested by the papers, and in doing
so, to represent more fully what scientific inquiry is and can be.
ANOTHER MEANING OF THE WORD "CIRCLE"
In the above discussion, the word "circle" was used to suggest repetitive, inter-
locked processes. But there is another use of the word "circle" that suggests
"active intellectual group," as in Vienna Circle. This reminds us that science is
a social process, a point touched on by Morrow and Muchinsky in their discus-
sion of cooperative and competitive elements in science. More broadly, it sug-
gests that we might profitably be more aware of the organization of scientific
inquiry, the division of labor into empiricists, grand theorizers, and middle range
theorizers, the ways of coordinating their efforts, and the pathologies of inquiry
that can result when we do not. Hopefully, this book will help to reinforce the
social interpretation of "circle" as well as its procedural one, and serve to inte-
grate empiricism, grand theorizing and other strategies of inquiry, not to define
them as subordinate to a new orthodoxy of middle range theory.
REFERENCES
Ashby, Ross. An Introduction to Cybernetics. London: Chapman & Hill, 1956.

Bateson, Gregory, Steps to an Ecology of Mind. New York: Ballantine, 1972.
Geertz, Clifford. The Interpretation of Cultures. New York: Basic Books, 1973.
Habermas, Jurgen. Knowledge and Human Interests. Boston: Beacon, 1971.
Heidegger, Martin. Being and Time. New York: Harper & Row, 1962.
Homans, G.C. The Human Group. New York: Harcourt, Brace, 1950.
Kuhn, Thomas S. The Structure of Scientific Revolutions (2nd ed.). Chicago:
University of Chicago Press, 1970.
Mead, George Herbert. In A. Strauss (Ed.), Social Psychology. Chicago: Univer-
Middle range theories of social systems. Behavioral Science, 1974,19, 357-367.
Schutz, Alfred. The Phenomenology of the Social World. Evanston, IL.: North-
western University Press, 1967.
Weick, Karl E. The Social Psychology of Organizing. Reading, MA.: Addison-
Wesley Publishing Company, 1969.
6 METAPHORS, THEORIES,
AND THE PROCESSES
OF SCIENTIFIC INQUIRY
A Reply to Pondy
Pondy has found much to criticize in my paper. I find that in some cases, I agree
with him because he has mistaken or misinterpreted my points, in other cases,
that I must disagree on quite substantial grounds, and in still others, that his
criticisms are merely careless errors. I have attempted to group my responses to
nearly a score of issues in his commentary according to general themes: reifica-
tion, goals and progress in science, and so forth.
THE DEFINITION OF MIDDLE RANGE THEORY
Are the Ten Defining Characteristics Mine? In several places, Pondy stated that
I laid down, proposed, created, and in some sense was responsible for the ten
characteristics that may be components of a defmition of middle range theory.
In fact, that section of my paper was thick with references to Rage, Dubin,
Gibbs, Kaplan, and other major figures of the theory-building literature. I re-
viewed and synthesized that literature, and then attempted to evaluate those
characteristics. I cannot honestly claim any of the characteristics as mine by
invention. My conclusion might possibly suggest that the qualitative characteris-
tics are mine by adoption, but it is phrased very tentatively, and I would prefer
72
METAPHORS, THEORIES, AND THE PROCESSES OF SCIENTIFIC INQUIRY 73
to defer that adoption until I have had much more opportunity to evaluate
them further.
Is MRT Well Defined? Pondy has charged that, in contrast to Bluedorn and
Evered, I considered the concept of middle range theory (MRT) to be well de-
fined. In fact, I pointed out in my paper that the defining characteristics overlap
each other, that the terminology used is inconsistent and unclear, that there are
differences among the writers on theory, that some of the characteristics pro-
posed by Merton do not distinguish MRT from other theory, and that Merton's
basic definition of MRT "has the advantage of simplicity, but the disadvantage
of vagueness." Pondy can not have read this section very carefully.
WHAT IS MERTON'S HYPOTHESIS?
In discussing Bluedorn and Evered's very useful typology of strategies of theory

construction, Pondy drew a sharp and clear distinction between their views and
another view which, in the context, must clearly be ascribed to either Merton
or Gilfillan or to both. This view favors "anointing the middle range strategy as
the best and only acceptable way of conducting scientific inquiry." Regardless
of how one interprets the implied ascription, I think it is fair to say that this is
an argument against a straw man. Pondy wanted to "develop an optional strat-
egy"; Merton and I would agree that there are many options in scientific inquiry
but we would differ in suggesting that all of these options are not equivalent.
Merton (1968) had both a descriptive and a prescriptive aspect. Descriptively,
he discussed grand theory, empirical generalizations, and data gathering as well
as MRT. Prescriptively, he asserted that, at the present time, the pursuit of
MRTs (or the MRT strategy) as a "major (but not exclusive) concern" (Merton,
1968, p. 50) would be more productive in advancing science, than either pur-
suing grand theory or focusing on empirical generalizations alone. I restated this
assertion as an hypothesis. Merton believed the hypothesis to be true, based on
his reading of the history of science. Bluedorn and Evered, based on their read-
ing of history, believed it to be false; this difference may suggest some difficulties
in attempting to test hypotheses using archival data. As I indicated in my paper,
I believe the hypothesis to be unproven, but at least potentially testable, using
appropriate data gathering methods. Neither Merton's chapter nor my article
contained any suggestion that MRT be anointed the "only acceptable way of
conducting scientific inquiry." The point is both narrower, and more reason-
able, than that: Some strategies with respect to theory may be more productive
than others. Nor can there be any doubt that, whatever the theoretical strategy
adopted, a variety of other activities is also part of "conducting scientific
inquiry ," including gathering facts, solving practical problems, and the like.
STRUCTURE, PROCESS, REIFICATlON, AND ENTITIVITY
A number of Pondy's specific criticisms and comments fall under this general
heading. His basic position was that organizations are "socially constructed
realities, collections of processes that do the work of organizing," and I fmd
it impossible to disagree with that. Unfortunately, however, he went further
by attempting to "stamp out nouns" (Weick, 1974, p. 358; Bateson, 1972,
p. 334), to deny that organizations (and MRTs) are "objective things or en-
tities with persistent and measurable properties." In passing, I think it appro-
priate to note that processes have persistent and measurable properties too, such
as rate, but that is not the issue. The question is, does a process viewpoint ex-
clude an entitive, or structural, viewpoint?
It is reasonable to regard an automobile as largely a process from one form of
iron oxide, hematite, or iron ore, to another, rust, and indeed it is an unfortu-
nately fast process. Nevertheless, for practical and legal purposes, it is useful
and necessary to consider it as a structure. The process description is an accurate
but only a partial description, and a description useful only for some purposes. I
don't buy the car for its oxidizing potential. Similarly, a chair is a socially con-
structed reality, and may also be differently constructed as a substitute ladder, a
collection of pieces of wood, a defense against lions, a collection of molecules,
or an area of space thinly populated with elementary particles. The point is that
it is wrong to assert that the object in questions is only one of its many possible
aspects.
With regard to social entities such as organizations, the best and last word on
this subject has probably been that of Donald T. Campbell (1958), whose main
point was that entitivity is a matter of degree, of the extent to which social
objects have such characteristics as similarity, common fate, and so on, and
therefore, of the extent to which they are perceived as a unit. The continuing
existence and success of macro organizational behavior (0. B.) is a testament to
the usefulness of thinking about organizations as objects, with structures.
Are Theories Entities? Pondy clearly disagrees with me, and as I recall his
spoken comments in Vancouver, he went even further in asserting that what is
important is the theorizing, not the theory. Theorizing, a process aspect of
science, is very interesting and very important; it is studied by philosophers of
science (under such headings as logic and as the discovery problem) and by psy-
chologists (creativity, problem solving) among others. Nevertheless, it is a means
to an end, and the end is knowledge, intersubjective, reliable, precise, systematic
knowledge, for which read theory. The fact that theories evolve over time, or
that they appear somewhat different to different observers, no more affects the
entitivity of theories than it does that of cars, chairs, or organizations.
Has MRT Been Reified? Pondy asked the question "Does the concept of mid-
dle range theory make any sense at all?" and suggested that the answer is "no."
As I recall his spoken comments, that "no" was "unambiguous." He suggested
that, in the place of senseless, reified MRT, we think about middle range the-
orizing. Yet even Pondy has been unable to act upon these assertions. Late in
his comments may be found a discussion in which the words "middle range
theories" and "grand theories" are used quite concretely, and indeed, are com-
pared to other quite concrete entities, such as maps and stepping stones. He
can't have it both ways.
Is MRT an ''Absolute'' Concept? Pondy asserted that, while Bluedorn and

Evered considered middle rangedness to be a relative concept, I considered it
to be absolute, which I take to mean nominal, or categorical. I believe there is
no evidence in my paper to support such an interpretation, which is of course
at variance with my opinion. While space limitations prevented my discussing
this issue in detail, I did mention it explicitly, and several aspects of the paper
clearly illustrated a relative use of the term. Of the ten characteristics discussed,
only one is categorical, and the conclusion of that section discussed the desirabil-
ity of combining them into an index. The comparisons of systems theory, cog-
nitive dissonance theory, and Weick's organizing theory, and the discussion of
sources of difficulty in operationalizing Weick's theory, should have made this
even clearer.
Measuring the Degree of Middle Rangedness. If we accept the notion that an

MRT is not a member of a nominal, categorical, or absolute set of things, but
rather is an entity located in the middle of a continuum, the question then be-
comes: How can we place any given theory on that continuum? This is the
question the ten characteristics may help us to answer. I agree with Pondy
at least to the extent that I see methodological problems in using the quantita-
tive characteristics to compare theories, but these are not fatal.
It is certainly true that systems theory has produced more than 36 hy-
potheses, and that Weick's theory supports more than six. It is also true that it
is virtually impossible to make a definitive comparison on this basis, since to
do so we would have to make the comparison on the basis of all possible hy-
potheses derivable from the theories. This we cannot do, since all have probably
not been made, and may never be made. Furthermore, there is a time problem:
Theories evolve, so that dissonance 1962 is not the same as dissonance 1978.
Does this mean, as Pondy indicated, that the effort is "not fruitful," or even
"absurd"? Of course not. For the example in the paper, I chose one of many
possible measurement methods, none perfect. My choice was rather operationist:
The range of the theory was operationalized in terms of what the author of the
original publication had included in that publication. Another, more laborious

method, would be to do a literature search, and tabulate and compare either all
published derivations, or all published and empirically supported derivations.
Clearly, the nature of the problem does not permit ratio measurement, and
probably does not permit interval measurement, but I believe it does permit at
least ordinal measurement.
GOALS AND PROGRESS IN SCIENCE
What Is Progress? Pondy listed eight indices of scientific progress, with which I
suspect not only I, but virtually everyone else could agree. I would add at least
one more, the development of new research methods and instruments. Surely
the invention of the telescope or the microscope, the cyclotron or the electron
microscope, the l.Q. test or the Skinner box, were not merely "purely instru-
mental" as Pondy states, but were in addition, great advances in applied science,
and generated whole new fields of study in many cases.
These nine indices of scientific progress can be ordered in at least three ways.
First, the preferences of the individual scientist will determine a personal ranking
in importance, the methodologist preferring to search for new methods, the
practitioner to search for new applications, and so forth. Second, they may be
flow-charted, in the manner of Pondy's circle, indicating how each relates to the
others. Third, and most important, however, they can be organized as means
and ends.
Pondy argued that all eight ( or nine) indices of progress are equivalent in
being ends, or goals. Further, his analysis suggests that the development of gen-
eral theory is not a meaningful definition of progress. I disagree, as I suspect
most scientists would. I think the development of general theory is a goal at a
higher level than the others; the others are contributory to and/or derivative
from it. Virtually any source that discusses the goals of science states these in
one of two ways: as description, explanation, and prediction, or as the produc-
tion of knowledge of a certain type. Obviously, the most adequate description,
explanation, and prediction are provided by the most general theory. The knowl-
edge which is the objective is usually qualified as follows (an example chosen
from the book nearest to hand): "The quest for scientific knowledge is therefore
regulated by certain standards or criteria (or) ... ideals to be approximated, but
perhaps never fully attained. The most important of these ... are: (1) Intersub-
jective testability ... (2) Reliability, or a sufficient degree of confirmation ...
(3) Definiteness and precision ... (4) Coherence or systematic structure ... (5)
Comprehensiveness or scope of knowledge" (Feigl, 1949). In combination, these
also constitute a reasonable definition of general theory.
In summary, then, while all of the indices are desirable goals in themselves,
and contribute to scientific progress, all the others are also means to the ultimate
goal, general theory, as defined above.
Normal and Revolutionary Science. Although it is not immediately relevant

to my own paper, I cannot help commenting upon another issue raised by
Pondy in this context. Pondy claimed that "adopting a middle range strategy
would preclude the most fundamental and important forms of scientific progress
-the creation of new paradigms of inquiry." This comment is based upon the
most fundamental misunderstanding of Kuhn, who was quite clear that "re-
search under a paradigm (normal science) must be a particularly effective way of
inducing paradigm change" (Kuhn, 1970, p. 52). The latter comment I take to
be wholly consistent with Merton's hypothesis.
Solving Our Own Problems. Near the beginning of his comments, Pondy called
for those of " ... us in O.B. to solve our own problems, rather than importing
prefabricated solutions." I have difficulty knowing whether he really meant
it, given his later utilization of Heidegger (a philosopher) and Habermas (a
Frankfurt-school sociologist). But let us suppose he was serious about this
"authenticity": will such a strategy contribute to the advancement of O.B.?
I suggest that, while O.B. has both substantive and methodological problems,
which are to some extent unique, the field is not and cannot be separated from
the other behavioral and social disciplines. As Weick has argued, organization
behavior is behavior that happens to take place in organizations (1969, pp. 25-
27). Organizational behavior is, has been, and must remain at the intersection
of the related diSciplines. There are political and organizational forces that tend
to artificially separate related fields, but for the sake of scientific adequacy, it is
important to avoid "the ethnocentrism of disciplines" (Campbell, 1969).
ROOT METAPHORS AND GENERAL THEORY
Are They the Same? There is a good deal of ambiguity among writers on theory
as to what constitutes a general theory. Merton provided a good example: his
chapter began with the observation that "the word theory threatens to become
meaningless because of the diversity of its referents, "from minor working hypo-
theses, through comprehensive but vague and unordered speculations, to axiom-
atic systems of thought" (1968, p. 39). In the following paragraph he proposed
to restrict the use of the word to "logically interconnected sets of propositions
from which empirical uniformities can be derived." Yet, a page later, he referred
to the "seminal ideas" or metaphors from which several natural-science theories
derive as themselves theories: "Each of these theories provides an image that

gives rise to inferences" (l968, p. 40). A comparable, more or less mirror-image
problem has been seen in the many different ways Kuhn used the work "para-
digm" (Masterman, 1970, pp. 61-65).
Pondy has referred to "root metaphors" in a way that suggests that he be-
lieves that they are the same as grand theory, and the fact that root-metaphor-
making does not have a separate place in his circle of inquiry reinforces this
impression. The point I was trying to make in my conclusion, albeit too briefly,
was that there is an alternative way of looking at these things, distinguishing
them, and that this way may well be productive. I take Weick's clockface image
at the fmal session of this conference to be supportive of this notion: Theory at
two or three o'clock is substantially different from theory at eight or nine, and
therefore must be treated differently (Weick, 1979). This point is also supported
by Merton (1968, p. 52) and by others cited in my paper. As in the case of the
distinction between MRT and grand theory, I suspect the difference here to be
one of degree, rather than of kind. Further, I think we understand how to deal
with grand theory far better than how to deal with metaphors, paradigms,
and so on.
How are they Related? If we distinguish between root metaphors, general

orientations, sensitizing concepts, frames of reference, paradigms, and so forth
on the one hand, and general theories on the other, some interesting conse-
quences follow. Bluedorn and Evered's Figure 1 must be modified, extending
the explicitness dimension downward, to make room for root metaphors lower
than all of the existing diagrammed relationships in explicitness, and adding a
number of strategies connecting root metaphors with other points on the figure.
Similarly, the shape of Pondy's circle of inquiry must be drastically revised, as
indicated in Figure 1. The figures are alike primarily in showing the several con-
crete aspects of science connected by processes, and in the Roman numeral
scheme for the processes. Considering the theme of this conference, it is appro-
priate that my figure restores the centrality of middle range theory, both spa-
tially, and in terms of the many processes in which it is implicated.
The most important aspect of Figure 1 is that it clearly differentiates between
processes connecting grand theory and middle range theory, and processes con-
necting paradigms and MRT. I have labelled two processes concretization, using
a term from Pondy, but have also distinguished them. Concretization of grand
theory (VIa) is primarily a deductive process, while concretization of paradigms
probably involves a combination of deduction, artistic insight, and other cogni-
tive processes. A similar distinction can be made between the processes that sup-
port paradigms and grand theories. Grand theory is verified, usually indirectly,
through confirmation of middle range theories (processes III and V). However, I
VII Problem Solutions
Root Metaphors. Paradigms (etc.)
/I Idiographic.
Ethnographic.
or "Thick"
Descriptions
X VIb
X XII (not "grand
Empirical
/1""\
la, II
Generalizations ..
Middle VIa
Range ...- - - - - - - - - - -
(not "mi ddl e ~ Theory
VIr::""') III. IV. - ~

Probl em Probl em
Solutions Solutions
Research
Methods
Figure 1. An Outline of the Content and the Processes of Scientific Inquiry: The Circle
of Inquiry, Revised.
Note. Key to the numbering of the processes:
la. Selection of data

lb. Selection of methods
II. (Re)interpretation of existing data
III. Verification of theory
IVa. Shaping of theory (induction)
IVb. Shaping of Theory (limitation)
V. Consolidation of theory
Via. Derivation (deductive)
Vlb. Concretization (deductive and intuitive and ?)
Vila. Problem solving (tinkering)
Vllb. Problem solving (engineering)
VIIc. Problem Solving (insight)
X. Assuring or belief-supporting.
XII. Suggesting, or sensitizing to, particular empirical areas
XIII. Abstracting (induction from one or few cases)
XIV. Grand inductive leaping
XV. Inclusion (artefact paradigm problem-solution)
79
have labelled the connections between theory and paradigms as "assuring" pro-
cesses, to indicate that, while root metaphors may not be verified in the ordinary
sense, the existence of congruent MRTs and general theories, as well as facts,
at least increases our belief or faith in the metaphor. The former is a formal
scientific process, reasonably well understood; the latter is akin to argument
from example; we know less about how to do it usefully in a scientific context.
What Is the Role of "Thick Description"? I am quite uncomfortable with the

term, grand facts, for I think it implies an invidious distinction among facts. If
there is to be such a distinction, there is good reason to believe that middle
range facts (I prefer Merton's term, "empirical generalizations") should have
the place of honor (platt, 1964; Chamberlin, 1965). Be that as it may, discussion
of "thick" idiographic, or ethnographic descriptions is in order. They are impor-
tant, not only in and of themselves, but in relation to other data (Mack, 1969).
Furthermore, they are not done nearly frequently enough (Snizek, 1976).
I doubt that many thick descriptions are generated by aggregating empirical
generalizations, as depicted in Pondy's Figure 3. Rather, they are usually col-
lected separately, and using separate research methods, of which ethnography
(or participant observation), deep psychoanalytic interviewing, and some orga-
nizational case studies are examples. Furthermore, their immediate consequences
are not restricted to problem solving: they may also sensitize investigators to
factual issues, which can be further studied using other more rigorous and/or
more representative methods, and they may also be used to generate MRTs
and/or grand theory directly.
That the latter uses are problematic may best be indicated by two well-known
examples. Two anthropologists, Redfield (1930, 1941) and Lewis (1951,1953)
studied the same Mexican village, arriving at radically different thick descriptions
and radically different theories of peasant society. Even better known are the
many schisms among Freud and his various students, caused, in large part, by
deriving different theories from thick descriptions, and resulting in controversies
over the scientific validity of psychoanalytic theory, and the usefulness of
psychoanalytic treatment, which persist to this day. Thus, I think it makes sense
to deal with thick descriptions, not as inevitably a part of a single great circle of
inquiry, but rather as separable, with unique relationships to a variety of aspects
of inquiry. There is a kind of parallelism between the relationships of MRTs and
paradigms and of empirical generalizations and thick descriptions; however, this
issue is not one of grandness, but rather of our ability to deal adequately with
the latter member of each pair.
Other aspects of Figure 1 may be briefly explained. The role of methods is
explictly recognized, as suggested under the heading, "what is progress," above.
Practical problem solving emerges from all aspects of the scientific enterprise:
from facts (as a result of tinkering), from MRTs (engineering), even from root
metaphors (insight or inspiration; vide Archimedes' bathtub, or Kohler's apes,
1925). Finally, neither Pondy's circle nor my own figure possesses a necessary
third dimension, to indicate in graphic form that what's involved is not horizon-
tal tail-chasing, but rather an upward spiral.
ANALYSIS OF A THEORY
There are a number of ways of analyzing, examining, interpreting, or giving

meaning to any given theory. One of these is empirical, that is, testing the the-
ory. Another is conceptual analysis, clarifying concepts, assumptions, relation-
ships and conclusions of the theory. Theoretical elaboration might be yet another.
A fourth is establishing connections between the given theory and others of
similar range, more general theories, and/or paradigms, showing how the theory
is consistent with existing knowledge in the field. Finally, there is satori, ex-
perienced wholeness, in which the theory is subsumed into the Zen "Void of the
Unmanifest" (Humphreys, 1968, p. 113).
How Should Weick's Theory Be Analyzed? My analysis of Weick's (1969)

organizing theory involved the first two of the ways listed above, empirical
testing and conceptual analysis. Pondy did not seem to object to the way I did
this; rather, he criticized me for something else which I did not do. Pondy as-
serted that I should have examined the theory in its context, which, as he cor-
rectly observed, includes phenomenology, symbolic interactionism, general
systems theory, and cybernetics, among others. Further, he asserted that I had
lifted the theory "out of the context of inquiry, out of its place in the historical
development of theory, and out of its place in the stream of empirical and the-
oretical thought." What could this assertion, or criticism, mean?
It seems unlikely that this could mean doing a different kind of empirical
analysis. It is true that my research dealt only with a portion of the theory, pri-
marily its informational aspects. This, it seems to me, was virtually inevitable: I
can think of no example of all aspects of a complex theory being successfully
operationalized in a single test. It follows, a fortiori, that operationalization of
the entire theory, and its context too, is unlikely.
One potential meaning of Pondy's criticism is that I should have done a
discursive review, relating the organizing theory to its antecedents in phenom-
enology, symbolic interactionism, and so on. This, however, seems to me to be
quite redundant, since Weick's book did the job very well.
The remaining possible meaning of Pondy's assertion is that I should have
attempted to express the wholeness of Weick's theory, in a context which
presumably includes not only behavioral science, but everything else as well. I
freely confess that I feel inadequate to such a task. However, if I understand
the intention of Zen training correctly, it is to train the novice without actually
telling him anything, for he can only experience what is important, not be
taught it. This suggests that an attempt by me or anyone else to express the
wholeness will be fruitless, that "all is incomprehensible" (Humphreys, 1968,
p. 14; Bobko, 1979), and that each person will necessarily have a different sense
of that wholeness, independently arrived at. Thus, in Zen terms, what Pondy
suggested would be futile. Furthermore, while the kind of Zen thinking which
Bobko recommended may be a way to knowledge, it is certainly not a way to
scientific knowledge, since intersubjectivity is one of the key characteristics of
scientific knowledge.
Thus, Pondy seems to have recommended that I do one of two things, one
of which is superfluous, the other of which is both philosophically futile and
scientifically "non-sense" (Bobko, 1979). Under the circumstances, I think my
preference to other activities is reasonable.
Is Weick's Theory "Useless"? Although I suspect that Pondy did not believe
it, I do in fact agree with him that Weick's theory is not useless. My paper dis-
cussed some aspects of its usefulness. I also agree with him that "work remains
to be done to link up Weick's insights with empirical procedures"; a major
section of my paper was devoted to precisely this issue. While I do not fully
agree that "the terms in Weick's theory can not be operationalized directly",
I did describe a number of difficulties that I encountered when I tried to "spec-
if[y them] in some concrete empirical context." This criticism seems to be
based on a misreading of my paper.
FUTURE DIRECTIONS
The real issue here is, how best to deal with theories that are on the upper part
of the range continuum, as I have argued Weick's is, and with paradigms or root
metaphors generally. I suspect that other such theories have difficulties similar
to the ones described in my paper. For example, a recent review of a book by
four British Marxists indicated that they found one of the mainstays of Marxist
theory, the labor theory of value, to be unrelated to the market price of goods,
useless for economic analysis, and "an obstacle to understanding how a capitalist
economy works" (Hindess, 1978, p. 80). Since Marxism is so widely known and
intensely studied and debated, it is remarkable that such a weakness took more
than a century to surface. However, one explanation of the delay may be the
kind of difficulties I had with Weick's theory, preventing the sort of testing that
would have rapidly revealed the flaw.
Thus, to return to my Figure 1, I have argued that there are important dif-
ferences between grand theory and paradigms, and between empirical generaliza-
tions and thick descriptions. We understand much better how to deal with the
first and third of these than the other two, for instance, how to relate each to
middle range theory in a productive way. I have not argued that paradigms and
thick descriptions are therefore useless, for I do not believe that. Rather, I have
argued that we should attempt to understand them better, and to understand
the processes by which they are linked to each other, to theory, and to empirical
generalizations. There is some indication that these processes are not the same as
the processes which link grand and midrange theory to empirical generalizations.
There are three implications of this analysis for future work in O.B. If we can
learn to tell the one from the other, we can, first, deal with theories and em·
pirical generalizations appropriately, using well-developed modes of thought.
Second, we can avoid applying those modes of thought inappropriately to
paradigms and thick descriptions. Finally, we can also proceed to develop appro-
priate modes of thought for dealing with the latter.
REFERENCES
Bateson, G. Steps to an Ecology of Mind. New York: Ballantine, 1972.

Bobko, P. Middle range theories of organizational behavior: Some implications
for the development of theory. In Pinder, C. and Moore, L., Middle Range
Theory and the Study of Organizations. Martinus Nijhoff, 1979.
Bluedorn, A. C. and Evered, R. Middle range theory and the strategies of theory
construction. In Pinder and Moore, op. cit.
Campbell, D.T. Common fate, similarity, and other indices of the status of ag-
gregates of persons as social entities. Behavioral Science, 1958, 3, 14-25.
--. The ethnocentrism of disciplines and the fish-scale model of omniscience.
In Sherif, M., and Sherif, C.W. (Eds.), Interdisciplinary Relationships in the
Social Sciences. Chicago: Aldine, 1969.
Chamberlin, T.C. The method of multiple working hypotheses. Science, 1965,
148, 754-759.
Feigl, H. The scientific outlook: Naturalism and humanism. American Quarterly,
1949, 1. Reprinted in H. Feigl and M. Brodbeck (Eds.), Readings in the
Philosophy of Science. New York: Appleton-Century-Crofts, 1953.
Gilfillan, D.P. Characteristics of middle range organizational theories and their
implications for operationalization and testing: A conceptual analysis with
empirical illustrations. In Pinder and Moore, op. cit.
Hindess, B. Interview, cited in A Marxist assault on Marx. Business Week,
July 17, 1978, p. 80.
Humphreys, C. Zen Buddhism. New York, MacMillan, 1968. Cited in Bobko,
op. cit.
Kohler, W. The Mentality of Apes. New York: Harcourt, Brace & World, 1925.
Kuhn, T.S. The Structure of Scientific Revolutions (2nd ed.). Chicago: Univer-
Lewis, o. Life in a Mexican Village: Tepoztlan Restudied. Urbana: University of
Illinois Press, 1951.
- - . Tepozthin restudied: A critique of the folk-urban conceptualization of
social change. Rural Sociology, 1953,18, 121-136.
Mack, R.W. Ego, energy, and the education of a sociologist. In I.L. Horowitz
(Ed.), Sociological Self-Images: A Collective Portrait. Beverly Hills, Sage,
1969.
Masterman, M. The nature or a paradigm. In I. Lakatos and A. Musgrave (Eds.),
Criticism and the Growth of Knowledge. New York: Cambridge University
Press, 1970.
Merton, R.K. On sociological theories of the middle range. In Merton, R.K.,
Social Theory and Social Structure. (1968 enlarged ed.) New York: Free
Press, 1968.
Platt, J.R. Strong inference. Science, 1964,146, 347-352.
Pondy, L.R. Commentary on Gilfillan, Morrow-Muchinsky, and Bluedorn-
Evered: The circle of inquiry. In Pinder and Moore, op. cit.
Redfield, R. Tepoztlan: A Mexican Village. Chicago: University of Chicago
Press, 1930.
- - . The Folk Culture of Yucatan. Chicago: University of Chicago Press, 1941.
Snizek, W.E. An empirical assessment of "Sociology: A multiple paradigm
science". The American Sociologist. 1976,11, 217-219.
Weick, K.E. The Social Psychology of Organizing. Reading: Addison-Wesley,
1969.
- - . Middle range theories of social systems. Behavioral Science, 1974, 19,
357-367.
- - . Remarks during the contemplative panel session, August 3. In Pinder and
Moore, op. cit.
IIWHY AND HOW
DOES MIDDLE RANGE
THEORY DEVELOP?
7
THE INEVITABILITY
OF MULTIPLE PARADIGMS
AND THE RESULTANT
NEED FOR MIDDLE
RANGE ANALYSIS
IN ORGANIZATION THEORY
Craig C. Pinder and Larry F. Moore, *
It has become fashionable in recent years for scholars to criticize "the state of
the art" in the study of organizations. A content analysis of many of these
criticisms and alleged problems reveals discontent about fundamental issues such
as the directions the field has taken or is taking, and about how research and
theorizing are being conducted. Controversy of such a basic nature and the
frequency of its appearance suggests to us that there is considerable disappoint-
ment and frustration on the part of many students of the field and a widespread
suspicion that the rate of progress in our learning about organizations is incom-
mensurate with the effort being expended.
The nature and incidence of this controversy seems similar to that which
Kuhn (1970) suggests occurs before the fall of a dominant paradigm and the
emergence of a new one. However, it will be argued in this paper that the high
incidence of controversy in our field in recent years reflects the simultaneous
existence of many images (or paradigms) about the subject matter studied, and
that this diversity of imagery and the controversy it generates are inevitable.
Two alternative philosophical positions will be adopted and it will be argued
*Tbe authors thank Peter J. Frost and an anonymous journal reviewer for comments
on earlier drafts of this manuscript.
87
within the bounds of each perspective that organizational behavior and organiza-
tion theory (O.B./O. T.) may never see the emergence of a single over arching
paradigm of the sort Kuhn (1970) describes in the context of other sciences,
certainly not in the foreseeable future.
Before our analysis proceeds, however, a few comments regarding the nature
of the paradigms are in order.
THE CONCEPT OF PARADIGM
Most scientists are familiar with the work of Kuhn (1970) on the revolutionary
nature of scientific development. Although he provided an important, if not
seminal (Masterman, 1970) understanding of the role of paradigms, there have
been a number of recent criticisms of the lack of clarity of Kuhn's paradigm
concept, since most of his discussions of the concept deal with what paradigms
do rather than what they are. Pursuant to this, Ritzer (1975) has elaborated on
Kuhn's definition. Since Ritzer's concept of paradigm is central to the analysis
that follows, we will examine it closely before proceeding.
For Ritzer, a paradigm is " ... a fundamental image of the subject matter
within a science ...." It is the "broadest unit of consensus within a science"
that serves to differentiate one community of scholars from all others.
There are two particular aspects of Ritzer's concept of paradigm that are of
special importance to our arguments. The first (and one that is also reflected
in the work of Masterman, 1970) is the notion that a scientific discipline can
simultaneously have several competing paradigms, each of which identifies a
scientific "subcommunity." Accordingly, for example, Ritzer has identified
three separate subcommunities, or paradigms, in contemporary sociology.
Pondy and Boje (1975) have subsequently explored the status of these three
in the context of organization theory. The three paradigms are referred to as the
social factist, the social behaviorist and the social definitionist points of view.
The second facet of Ritzer's concept that has relevance for our analysis is
his view that a paradigm consists of four basic components: (1) an "image"
of the phenomena and subject matter that are the focus of the discipline; (2)
an "exemplar", which he defmes as a seminal piece of work that articulates
and reports the essence of the image to scientists in that particular field; (3)
theories; and (4) research methods and instruments. But for Ritzer, the essence
of a paradigm consists primarily of the imagery created in the minds of scien-
tists by the substance of the phenomena being studied. Therefore, to say that a
discipline is a multiple paradigm science implies that different scientists within
the field hold different images about the essential nature of the phenomena
they study.
MIDDLE RANGE ANALYSIS IN ORGANIZATION THEORY 89
In the following sections it will be argued that O.B./O. T. is presently a multi-

ple paradigm science-although we do not necessarily accept the trichotomy of
images discussed by Ritzer (1975) and Pondy and Boje (I975)-and that this
condition of multiple imagery is responsible for much of the controversy in
our literature.
The first set of arguments to be made will adopt the epistemological perspec-
tive of "Realism" (Joad, 1957): the doctrine that phenomena and substances
have a real existence of their own, independent of the mind of the observer/
scientist. While there are many schools of Realism, the one tenet that binds them
together is a rejection of Berkeley's conclusion that to exist is to be either a
mind or an idea in some mind. In the context of O.B./O. T. then, a realist as-
sumes that organizations (or an organization) exist, and that it has certain more-
or-less stable and knowable properties that reveal themselves to study. This
existence and accompanying properties are independent from and external to
the mind of the observer. We believe that the vast majority of the extant litera-
ture in the field implicitly assumes this perspective. l
Now, within the confines of this perspective, we will argue that multiple
images and paradigms of organizations are inevitable for (at least) three reasons:
1. the vast real differences between organizations at any given point in time
2. the changing nature of organizations and their contexts
3. the reactive nature of the social scientific process of studying social
organizations
Following the presentation of these arguments, a different epistemological

stance will be adopted and the same (multiple paradigm) proposition considered
using a different argument.
ORGANIZATIONAL HETEROGENEITY AS A CAUSE

OF MULTIPLE IMAGES
Philosophically, it can be argued that nothing is static, that nothing in the

universe is constant over time, and that the state of being is a condition of be-
coming. Years ago, for example, sophists such as Heraclitus reasoned that an
individual can never step into the same river twice (Munitz, 1971, p. iv). In the
following remarks however, a less extreme assumption will be adopted. The as-
sumption is simply that some objects and phenomena are more stable over time
than others: that some phenomena change while the essential nature of other
substances and phenomena are static or slower to change.
Consider the prototypical nature of organizations now and in the past. The
progression in history from feudalism, through the industrial revolution and into
modem times, as charted by writers such as Bendix (1956) reveals that the
fundamental form and nature of human organization has undergone considerable
change. Further, it seems that organizations (and their environments) of the
present day are typically different from those found at the tum of the century
when the formal study of organizations began (Maccoby 1976). Moreover,
futurologists such as Bennis (1969), Toffler (1970), and Scott (1974) have pre-
dicted further drastic changes in organizations and their contexts in the future.
Now, as social, economic, and political environments change, organizations
change with them, or die. But these adaptive organizational responses do not
occur suddenly or all at once. At a given point in time it seems reasonable to
expect to find many organizations in a population that feature vestiges of
process and structure suitable for the past but which gradually change to be
adaptive in the future. Simultaneously, an observer will note many new (or even
old) organizations that bear the structural and process styles suited to their
present environmental circumstances. Still other organizations may feature
managerial and structural styles that are inappropriate for any set of circum-
stances. Not only do prototypical organizations of given environmental eras
differ as a function of their respective circumstances, but individual organiza-
tions at a given point in time reveal complex differences. In addition, these same
individual organizations themselves may be in the process of changing over time
to adapt to their changing environments. The faster the rate of change of proto-
typical or particular organizations in a population, the sooner well-established
imagery becomes outmoded and dysfunctional. Research guided by paradigms
based on outmoded imagery will naturally yield disappointing results and give
rise to the type of fundamental controversy mentioned earlier.
In short, it can be argued that O.B./O.T. presently features multiple para-
digms because different observers of organizations understandably see vast dif-
ferences between the (real) organizations they observe. These divergent images
give rise to divergent theoretical and applied ideas about organizations and
generate controversy and polemics among adherents of the different paradigms.
CONTROVERSY SUGGESTIVE OF DIVERGENT IMAGERY

AND MULTIPLE PARADIGMS
Examples of these differences and the controversies they have generated are
plentiful in the recent literature. Concerning the most appropriate imagery
for the field, consider Georgiou's (1973) attack on the goal paradigm and his
suggested substitution of an "organizations-as-marketplaces" image in its place.
Or consider the proposition by Cohen et al. (1972) and March and Olsen (1976)
that organizations are "garbage cans," as opposed to the rational systems most
previous thinking implicitly assumed. There is Barnard's (1938) view of orga-
nizations as social systems; Tushman's (1977) image of organizations as political
systems; and the notion of organizations as "loosely-coupled" systems (Glass-
man 1973; Weick, 1976). Consider, in fact, the recurrent attack on the systems
metaphor itself as a general framework for studying organizations (cf. Burack,
1975; Kast and Rosenzweig, 1972; Melcher, 1975; Milburn et al., 1975; Pondy
and Mitroff, 1978; Thayer, 1972; Weick, 1974a; Zaleznik, 1967). Each of these
authors either adopts or rejects an image of what constitutes organizations, and
again, most of them seem to adopt a realist perspective.
Notice that Nord's (1974) Marxist critique of the field attacks what we take
as given and what we view as manipulatable in understanding and changing
organizations. Recall Dunnette's (1966) observation pertaining to our susceptibil-
ity to fads and fashions and Naylor's (1971) perspective on how scholars should
be trained for the field. These sets of observations and the controversies they gen-
erated reflect assumptions about the nature of the field: what it is and how it
should be viewed. Reflect on the intermittent complaints about the necessity for
considering the sociological facets of organizations (cf. Perrow, 1970, 1972;
Pfeffer, 1976) and consider the ageless debate about the nature of the individual
human being and how it has been updated by contemporary organizational
psychologists such as Scott (1974), Deci (1974) and Locke (1977). Each of
these debates, suggestions, and comments reflects, in our view, the vast diversity
among their authors in the images they hold about the nature of organizations
and organizational behavior. Finally (although not exhausting the list of possible
examples), consider Bittner's (1965) views of organization as sources of rules for
conduct, stylistic unity, and corroborative reference.
In addition, consider some of the recent controversy regarding methodology.
Weick's (1974b) suggestion that we study "absurd organizatioTls," "micro-
organizations" and "everyday events" reflects an image of organization, as does
Piehl's (1974) admonition that we think in and look for nonlinear relationships.
Sayles' (1975) argument that we return to anthropological methods, and the
recurrent cry for the simultaneous consideration of variables from several levels
of analysis (cf. Fredericksen et aI., 1972; de Wolff and Shimmin, 1976) all re-
flect implicit imagery about the nature of organizations. The "organizational
realities" perspective of Frost et al. (1978) and the isolation of alleged "blind-
spots" by Frost (1977) and Weick (1977a) similarly reflect a belief about what
imagery is most appropriate. RimIer's (1976) allegation that the field is pre-
occupied with mathematics, psychology, and "trivia" also reflects an image of
what the field consists of and how it should be studied.
We could continue this list, but hopefully our point is made: Different
scholars are looking at organizations and are seeing different things; different
images follow and different methodologies proliferate until the literature is

jammed with pitifully low validity coefficients and myriad suggestions concern-
ing the cause of the problem and why we are not being better. The cause, we
believe, lies in the preceding section of this paper: Scholars in the realist tradi-
tion see different images because they are not looking at a homogeneous or
orderly population of phenomena. Organizations are incredibly diverse (both
between themselves, within themselves, and over their own life spans) such that
the chaos of the myriad images held to represent them is entirely understandable
and justified. In addition, the process of registering their essence for study is
complicated by the fact that organizations are social phenomena and researchers
are social stimuli: the old but still relevant problem of reactivity (Webb et al.,
1966).
The nature of the solutions offered varies almost as widely as the number of
criticisms raised. For example, Weick (1974b), has suggested abandoning the
search for general theory: at least, to give up our obsession with systems as the
general theory. Pondy and Mitroff (1977) have argued that we are attempting
to study phenomena at Boulding's (1968) most complex systems level ("multi-
cephalous" systems) with models and research methods of much lower systems
complexity. In fact, they argue that for the most part, we are not even function-
ing at the level of open-systems as post-Thompson (1967) academics claim we
are. In the end, however, these critics still leave us with systems as the recom-
mended image for our paradigm. Cummings (1978), in his recent attempt to
define the field of O.B., suggested that we construct "bridges" between micro
and macro parameters through the use of concepts such as organizational struc-
ture, which is normally seen as a dependent variable in "Organizational Theory"
yet is treated as an independent variable in "Organizational Behavior."
Masterman (1970) has suggested that progress in a science takes the form of
moving from the condition of multiple paradigms to the condition of a single
over arching paradigm, and that this development occurs when " ... someone
invents a deeper ... paradigm, which gives a more central (fundamental) insight
into the field ... (p. 74)." While Cummings would accept the present images of
the field and try to link them, and while Weick would suggest abandoning our
attempt to see the field exclusively through the systems model, Masterman pro-
poses that progress occurs when we transcend the present images (contrary to
Cummings) and seek a more central, fundamental image (contrary to Weick). As
stated earlier, it is our belief that because of the changing nature of the phe-
nomena we study in O.B./O.T., the search for uniparadigm science with a realis-
tic philosophy of science is both futile and misguided. As organizations and the
environments surrounding them change in the future, the images we form will
also need to change and keep pace. It is doubtful that our science-making cycle
can keep pace.
Whether this state of confusion reflects what Kuhn (1970) would call a pre-
paradigm situation or what Masterman (1970) and Ritzer (1975) would call a
multiple paradigm condition is hard to determine, but perhaps that issue, in
itself, is not important. What seems important to us is the proposition that while
there is considerable variety underlying the chaos and while this variety could be
beneficial if attempts were periodically made to reduce the incommensurability
among the images we hold, we are not aware of many attempts to do so. It may
be that the field is far too young for us to reasonably expect the emergence of a
single hegemonic image, or overarching paradigm.
Or fmally, it may be that the diversity of imagery and considerable contro-
versy we are witnessing attest to the futility of embracing a realist's view of
the world, and that a philosophical stance such as Subjective Idealism or Phe-
nomenalism is more justified. We turn now to that possibility.
PHENOMENALISM AND THE STUDY OF ORGANIZATIONS
There is at least one other major reason why our literature reflects the simul-
taneous existence of so many paradigms (images) of organizations. The reason
has been suggested by the writings of Weick (1969, 1977b) and requires, to be
understood, the rejection of the assumptions of Realism (as described earlier)
and the adoption of a different epistemology. The alternative viewpoint has its
roots in the thinking of Locke, Berkeley, Hume, Mach, and Schutz. Broadly
classified, these thinkers come from the tradition of Subjective Idealism or
Phenomenalism. Very briefly (and with a fear of unfairly typecasting Weick
or oversimplifying his perspective), Phenomenalism " ... is the doctrine that all
statements about material objects can be completely analyzed into statements
about sense-data (Whiteley, 1974, p. 712)." It " ... involves the thesis that
anything we know about material things may be expressed in statements re-
ferring solely to appearances (Chisholm, 1968, p. 83). " When Weick remarks
that organizations, organizational boundaries, and organizational environments
cannot be known but are simply what actors enact them to be, because they
exist only in the heads of the actors themselves, he sounds amazingly like
Berkeley and the physicist Mach who both believed that there is no physical
world behind the appearances and sensory stimulation we attribute to entities
that we think of as occupying space in the universe. Writes Mach:
Different ideas can express themselves with the same exactness in the domain
accessible to observation. The facts must hence be carefully distinguished
from the intellectual constructs the formation of which they suggested. The
latter-concepts-must be consistent with observation, and must in addition
be logically in accord with one another. Now these two requirements can be
fulfilled in more than one manner ... (Bradley, 1971, p. 83).
The important point for the present discussion is not the correct classification
of Weickian O.B. The important point is that from this general epistemology, as
from the realist view considered earlier, multiple paradigms of organizations are
inevitable. In fact, it may even follow that we can expect as many images and
paradigms of organizations and their contexts as we have observers of them.
Complicated by the practical problems of registering without interference the
mental maps and images actors hold (again, the issue of reactivity), organization
theorizing with a social defmitionist or phenomenalist viewpoint, in the limit,
would be chaotic from a scientific perspective. While ethnomethodologists claim
that organization theorizing by realists is fraudulant, adherents to the scientific
norms of replication, cross-validation and precision can claim that assumptions
from the social factist and social behaviorist schools (following Ritzer, 1975)
are necessary if chaos is not to prevail.
Pondy and Boje (1976) have reviewed the problems of managing a multiple
paradigm science. They conclude that a multiparadigm situation can be useful
for learning from the triangulation possible from competing perspectives. They
also discuss, somewhat optimistically, how efforts might be conducted to recon-
cile interparadigmatic differences to make the science more manageable. like
Pondy and Boje, we recognize the utility of multiple paradigms (although our
reasons for doing so are somewhat different), but unlike them, we do not
share their apparent optimism that reconciliation can take place. We predict
that the multiplicity of images that will abound in the heads of factists or
definitionists, (or realists or phenomenalists) will make the resolution to a single
paradigm hopeless.
RECAPITULATION AND IMPLICATIONS
Whether one assumes that organizations have independent existences of their

own, or that they and their environments are merely what we enact them to be,
the polyparadigmatic nature of the field can be explained and the high incidence
of fundamental controversy about imagery can be understood. Many of the most
frequently expressed complaints in the literature about methodology can also be
understood. Disagreement about imagery is conducive to allegations about poor
or inadequate methodology. For example, the often noted problem of our reli-
ance on quickly constructed research instruments with unknown psychometric
properties is not likely to improve. If organizations remain as heterogeneous as
argued earlier and/or if researchers continue to enact new images, new and dif-
ferent parameters will emerge and be identified accordingly. The large invest-
ment of time and other resources required to develop sound instrumentation

will likely be seen as increasingly too impractical in the context of such turmoil.
Consequently, we expect continued passionate exhortations on behalf of better
measurement for years to come. Further, if the prophesies of some organiza·
tional futurologists eventuate, we expect even more diversity of images and con-
troversy over methodology from realists, and an equal amount from the emergent
phenomenalists (or social defmitions). Our field may move further from, rather
than closer to, a condition of unity under a single Kuhnian paradigm.
If the foregoing remarks seem justified, what implications follow for students
of organization? We believe there are at least two major implications from
our analysis.
1. Understand and Accept Multiple Paradigms
The first obvious implication is that we simply accept the notion (for the time
being at least) of many different ways for viewing organizations and that pro-
ponents of the various images could be all somewhat correct. That is, it may be
that we should accept several simultaneous images as being more or less appro-
priate for any given set of organizations. Thus, for example, we might reason
that some (or all) organizations are more or less, and at varying points in time,
political systems, economic marketplaces, garbage cans, flypaper, cooperative
systems, and so forth. In fact, we may even consider system as a variable rather
than a fixed quantity and consider the possibility that some organizations are
more like systems than others.
We would not denigrate efforts to find bridges and to reduce the diversity of
paradigms in the field, but we would urge the vast majority of students in the
field who believe they are engaged in normal science not to set their expecta-
tions for general science too high, and would warn them to continue to expect
flak from adherents of (many) competing paradigms.
At the same time, however, we might suggest that journal editors require
polemics representing particular paradigms, as they attack the work of others
in different paradigms, to recognize and make explicit the differences of imagery
they hold, as a way of attempting the processes of "transpection" described by
Maruyama (1974). (Transpection involves seeing the world from the point of
view of alternative paradigms.) In other words, the price of admission into a
journal for the sake of polemizing would become the painful requirement that
the prospective critic articulate his or her underlying imagery as well as correctly
represent the assumed imagery of the author he or she seeks to assail. Then,
following the assault, the critic would be required to address the issue of whether
anything new has been learned about either of the two paradigms involved or
how they are similar or related. Otherwise, controversy such as we have wit-
nessed in recent years will continue to be only vacuous and destructive.
2. Abandon General Theory for Midrange Theory
A second implication rests on the assumption that organizations are incredibly

diverse (as argued above) and that different imagery may be appropriate for
different organizations in different contexts. The prescription following is that
we abandon, for some period of time at least, what can only be a futile search
for general theories that assume the universal acceptance of singular images.
Until something akin to strong inference (platt, 1964) can be developed and
deployed to systematically reduce the number of images we hold about organiza-
tions, acceptance of a single image seems unlikely and controversies of the sort
described above seem inevitable. Instead, we must seek to develop and refIne
midrange theories that seem to explain limited ranges of phenomena in limited
sample spaces.
Concentrating on the development of middle range theories of organiza-
tional behavior would appear to be particularly compatible with the polypara-
digmatic nature of our fIeld. A middle range theory, because of intentional
limits to content and scope, might more easily be based on a uniquely holistic
set of imagery and instrumentation. At the same time other limited theories
could be developed for other unique subgroups of organizations and/or content
areas. For instance, profitability might be an important problem for some types
of organizations but not for others. Further, the models constructed to explain
variability in profIts could differ from one class of organizations to others be-
cause of the inherent differences between varieties of profIt-oriented organiza-
tions. Middle range theorizing would thus involve both the systematic study of
organizational types (as suggested by taxonomists such as McKelvey, 1975,
1978; Pinder and Moore, 1979) and the nested analysis of various phenomena
within the contexts of these types. Emerging middle range theories could be
expected to compete with one another and/or be melded together as organiza-
tional phenomena change over time and as communalities are identifIed between
models and theories that were constructed for problems relevant to different
categories of organizations.
In short, we believe several benefits will accrue from focussing attention on
middle range theory development:
1. Limiting the scope of the phenomena under investigation should restrict

the number of contingency variables contributing to fIeld variance; thus,
complexity may be reduced to manageable proportions and predictive
power increased.
2. Careful sub grouping of organizations through taxonomic means should

help avoid the construction of overly general theories about unrelated
or weakly related phenomena.
3. Development of appropriately specialized measurement techniques and
instruments will be encouraged.
4. Recognizing the changing and multiparadigmatic nature of organizational
milieu should reduce adherence to outmoded or obsolete theories and
enhance the search for better ones.
Hopefully, ideas generated in this book will provide some impetus to the
acceptance of multiple paradigms and will offer further insight into the process
of middle range theory development and its possibilities for understanding
organizations.
NOTES
1. Two of the three current paradigms in sociology identified by Ritzer (1975}-the

social factist and social behaviorist paradigms- are consistent with a realist epistemology.
The institutions and social structures studied by factists and the behavioral phenomena of
interest to social behaviorists all are (implicitly, at least) assumed to have objective existence,
independent of the mind of the researcher. Ritzer's third paradigm - social definitionism - is
much closer to the epistemology of subjective idealism, to be discussed later.
REFERENCES
Barnard, C.1. The Functions of the Executive. Cambridge, Mass.: Harvard Uni-
versity Press, I 938.
Bendix, R. Work and Authority in Industry. New York: Wiley, 1956.
Bennis, W. G. Organizational developments and the fate of bureaucracy. In L. L.
Cummings and W. E. Scott (Eds.), Readings in Organizational Behavior and
Human Performance. Homewood, Ill.: Richard D. Irwin, Inc., 1969, pp. 434-
449.
Bittner, E. The concept of organization. Social Research, 1965,32, 239-255.
Boulding, K. General systems theory - The skeleton of science. In W. Buckley
(Ed.), Modern Systems Research for the Behavioral Scientist. Chicago:
Aldine, 1968.
Bradley, J. Mach's Philosophy of Science. London: The Athlone Press, 1971.
Burack, E. H. Aspects of the state of the art in systems theory: Accomplishments
and challenges. Organization and Administrative Sciences, 1975,6, 101-104.
Chisholm, R.M. Phenomenalism. In G. W. Engle and G. Taylor (Eds.), Berkeley's
Principles of Human Knowledge. Belmont, California: Wadsworth, 1968.
Cohen, M.D.; March, J.G.; and Olsen, J.P. A garbage can model of organiza-
tional choice. Administrative Science Quarterly. 1972, 17, 1-25.
Cummings, L. L. Toward organizational behavior. Academy of Management
Review, 1978,3, 90-98.
Deci, E.L. Notes on the theory and metatheory of intrinsic motivation. Orga-
nizational Behavior and Human Performance, 1974,29, 130-145.
de Wolff, C.J., and Shimmin, S. The psychology of work in Europe: A review
of a profession. Personnel Psychology, 1976,29 (2),175-196.
Dunnette, M.D. Fads, fashions, and folderol in psychology. American Psycholo-
gists, 1966,21,343-352.
Frederiksen, N., Jensen, 0., and Beaton, A.E. Prediction of Organizational
Behavior. New York: Pergamon, 1972.
Frost, P.J. Blindspots: Implications for teaching and application. Paper pre-
sented at the thirty-seventh annual meeting of the Academy of Management,
Kissimmee, Florida, 1977.
Frost, P.J.; Mitchell, V.F.; and Nord, W. Organizational Reality: The View From
Within. Pacific Palisades, Cal.: Goodyear, In Press, 1978.
Georgiou, P. The goal paradigm and notes toward a counter paradigm. Admin-
istrative Science Quarterly, 1973,18,291-310.
Glassman, R. B. Persistence and loose coupling in living systems. Behavioral
Science, 1973,21, 1-19.
Joad, C.E.M. Guide to Philosophy. N.Y.: Dover, 1957.
Kast, F., and Rosenzweig, J.B. General systems theory: Applications for organi-
zation and management. Academy of Management Journal, 1972,15, 447-
467.
Kuhn, T. S. The Structure of Scientific Revolutions. Chicago: The University of
Chicago Press, 1970.
Locke, E. The myths of behavior mod in organizations. Academy of Manage-
ment Review, 1977,2,543-553.
McKelvey, W. W. Guidelines for the empirical classification of organizations. Ad-
ministrative Science Quarterly, 1975,20 (4),509-525.
- - . Organizational systematics: Taxonomic lessons from biology. Management
Science, 1978,24, 1428-1440.
Maccoby, M. The Gamesman. N.Y.: Bantam, 1976.
March, J.G., and Olsen, J.P., Ambiguity and Choice in Organizations. Bergen:
Universitetsforlaget, 1976.
Maruyama, M. Paradigms and Communication. Technological Forecasting and
Social Change, 1974,6, 3-32.
Masterman, M. The nature of a paradigm. In I. Lakatos and A. Misgrave (Eds.),
Criticism and Growth of Knowledge. Cambridge: Cambridge University
Press, 1970.
Melcher, A.J. Theory and application of systems theory: Its promises, problems
and realizations. Organization and Administrative Sciences, 1975, 6, 3-10.
Milburn, T. W., Negandhi, A. R., and Robey, D. Evaluating systems theory con-
cepts: Comments. Organization and Administrative Sciences, 1975,6, 11-13.
Munitz, M.K. Identity and Individuation New York: New York University Press,
1971.
Naylor, J.C. Hickory, hickory, dock-let's turn back the clock. Professional Psy-
chologist, 1971, 217-234.
Nord, W.R. The failure of current applied behavioral sciences-A Marxian per-
spective. The Journal of Applied Behavioral Science, 1974, 10, 557-578.
Perrow, C. Organizational Analysis: A Sociological View. Belmont, Calif.:
Brooks/Cole, 1970.
- - . Complex Organizations. Glenview, Ill.: Scott, Foresman, 1972.
Pfeffer, J. Beyond management and worker: The institutional function of man-
agement. Academy of Management ReView, 1976, 1 (2), 36-46.
Piehl, D. Nonlinearity and the study of organizations. Academy of Management
Journal, 1974,17,760-767.
Pinder, C.C., and Moore, L.F. The resurrection of taxonomies to aid the devel-
opment of middle range theories of organizational behavior. Administrative
Science Quarterly, 1979,24,99-118.
Platt, J.R. Strong inference. Science, 1964,146, 347-353.
Pondy, L. R., and Boje, D. M. Bringing Mind Back In: Paradigm Development as
a Frontier Problem in Organization Theory. University of Illinois. Unpub-
lished manuscript, 1975.
Pondy, L.R., and Mitroff, I. Beyond open system models of organization. To
appear in B.M. Staw and L.L. Cummings (Eds.), Research in Organizational
Behavior. Greenwich, Conn.: JAI, 1978.
Pugh, D.S. Modern organization theory: A psychological and sociological study.
Psychological Bulletin, 1966, 6 6( 4), 235-25l.
RimIer, G.W. The death of management. Academy of Management Review,
1976,1(2),126-128.
Ritzer, G. Sociology: multiple paradigm science. The American Sociologist,
1975,10(3),156-167.
Sayles, L. Applied Anthropological Approaches. Paper presented at the 35th
Annual Meeting of the Academy of Management, New Orleans, 1975.
Schutz, A. The Phenomenology of the Social World. Evanston, Ill.: The North-
Scott, W. E., Jr. The effects of extrinsic rewards on intrinsic motivation: A
critique. Organizational Behavior and Human Performance, 1976, 15 117-
129.
Scott, W.G. Organization theory: A reassessment. Academy of Management
Journal, 1974,17, 242-254.
Thayer, F. General system(s) theory: The promise that could not be kept. Acad-
emy of Management Journal, 1972,15, 481-494.
Toffler, A. Future Shock. New York: Random House, 1970.
Tushman, M.L. A political approach to organizations: A review and rationale.
Academy of Management Review, 1977,2, 206-216.
Webb, E.J.; Campbell, D.T.; Schwartz, R.D.; and Sechrest, L. Unobtrusive Mea-
sures. Chicago: Rand McNally, 1966.
Weick, K.E. The Social Psychology of Organizing. Reading, Mass.: Addison-

Wesley, 1969.
- - . Middle range theories of social systems. Behavioral Science, 1974(b), 19,
357-367.
- - . Amendments to organizational theorizing. Academy of Management Jour-
nal, 1974, 17, 487-503.
- - . Educational organizations as loosely coupled systems. Administrative
Science Quarterly, 1976,21, 1-19.
- - . Blindspots in organizational theorizing. Paper presented at the thirty-
seventh annual meeting of the Academy of Management, Kissimmee, Florida,
1977(a).
- - . Enactment processes in organizations. In B.M. Staw and G. R. Salancik
(Eds.), New Directions in Organizational Behavior. Chicago: st. Clair Press,
1977(b).
Whiteley, C.H. Phenomenalism: Its grounds and difficulties. In W.P. Alston and
R.B. Brent (Eds.), The Problems of Philosophy. Boston: Allyn & Bacon,
1974,710-721.
Zaleznik, A. The Social Psychology of Organizations. Administrative Science
Quarterly, 1967,2,682-691.
8MIDDLE RANGE THEORIES
OF ORGANIZATIONAL BEHAVIOR
Some Implications for the
Development of Theory'"
DEFINITIONS
A disenchantment with global theories derived from open systems models has
led researchers to consider the efficacy of more specific, middle range theories
(see Pinder, 1977, for a review of this movement). According to a recent call
for papers, middle range theories are dermed as:
. . . [theories) that would attempt to predict and explain only a subset of
all organizational phenomena. As such, each midrange theory might make dif-
ferent sets of assumptions about organizations, consider different parameters
to be important and lead to distinctly different prescriptions for practice.
Each midrange theory may be based on a unique set of imagery of what con-
stitutes organizational behavior and require entirely different research strat-
egies and tactics (University of British Columbia announcement, 1977).
*1 am indebted to Peter Dachler for his help in the preparation of this manuscript. His
willingness to always not be distraught by my constant interruptions, listen to my ideas,
and provide helpful suggestions was imperative in keeping my infmities unbounded. 1 also
wish to thank Willard Larkin and Ken Smith for their time, helpful comments, and qualita-
tive knowledge.
101
According to this defmition, a mic.d1e range theory considers only a subset

of organizational phenomena. Logically, therefore, all theories are middle range
theories, since no theory can account for the infmity of variables impinging on
organizational behavior. Thus, an open systems theory could be considered a
middle range theory that focuses on a subset of philosophical, metaphysical,
anthropological, and social scientific notions. The definition of middle range
theory is therefore a relative one. What may be a middle range theory for one
investigator may be overabundant chaos for another investigator, or conversely.
Pinder's (1977) conception of nested levels of analyses and his use of the term
"lower-middle-range-theories" are necessary concessions to the relativity of what
constitutes a middle range theory.
If all theories are middle range theories, then what is the utility of this con-
struct? One answer has focused on the first sentence of the above defmition
of middle range theory. That is, subsets of organizations may help generate ap-
propriate taxonomic structures of organizations (McKelvey, 1975; Pinder and
Moore, 1978). Given a taxonomy, a particular (middle range) theory is invoked
depending on the type of organization encountered. This paper considers a
second set of reasons for the utility of the construct of middle range theory. By
the above defmition, such theories may arise as a function of "different sets of
assumptions about organizations" and "unique set(s) of imagery of what consti-
tutes organizational behavior." It will be argued that middle range theories,
which evolve from these latter reasons, provide social scientists with a needed
awareness of the multiplicity of explanation. The implications of this multiplic-
ity for the development of theory will also be discussed.
EXISTENCE OF MULTIPLE IMAGES
The existence of multiple paradigms in the social sciences can be readily docu-
mented. For example, Ritzer (1975) has indicated that sociology is a multiple
paradigm science. According to Ritzer, a paradigm is an image of a subject matter
that defmes what and how that subject should be studied. Thus, a paradigm/
image is a consensual unit of analysis that is clearly normative.
In psychology, Sampson (1977) contends that social psychologists have
emphasized research under the thesis of "self-contained individualism." This
thesis represents the belief that "syntheses of opposing or desirable characteris-
tics are located within the person rather than within an interdependent col-
lectivity" (p. 767). This image of a synthesis has dominated research in social
psychology as if self-contained individualism were a fundamental psychological
principle, which it is not. By example, (e.g., androgyny), Sampson indicates
that syntheses among persons provide alternative, and perhaps more desirable,
MIDDLE RANGE THEORIES OF ORGANIZATIONAL BEHAVIOR 103
solutions to contemporary social issues. A similar example of multiple images

is provided by Nord (1974) who indicates that the "Modern Human Resources
Management (MHRM)" movement focuses on individuals becoming, while a
Marxian perspective stresses mankind becoming. Nord contends that failure to
consider the Marxian image of man has led, in part, to the failure of applications
under the MHRM paradigm.
THE NATURE OF MULTIPLE IMAGES
As noted above, some researchers advocate a taxonomic approach to theory

development. In this case, different theories may be appropriate for differing
cells in the taxonomic structure. However, the multiple theories derived from
this taxonomic approach are not necessarily multiparadigmatic. As Ritzer
(1975) states, " ... theories are not paradigms. Rather, theories are but compo-
nents of far broader paradigms" (p. 158). Therefore, the existence of multiple
paradigms does not necessarily follow from a taxonomic approach. If the entire
taxonomy operates within one paradigm, then results of multiple theories may
be additive. That is, these results cumulate towards the solution of the problems
posed within the paradigm as in Kuhn's (1970) definition of normal science.
However, by considering the above examples of multiple images, it can be
seen that the root of much of the paradigm multiplicity in the social sciences is
more fundamental than a taxonomic "slicing of the pie." In the above examples,
multiplicity is a function of basic assumptions about the nature of man and
society. As Ritzer notes, differing paradigms are a direct function of this type of
fundamental difference in assumptions. Israel (1972) renames these assumptions
"stipulations," in order to indicate that they are normative, value-laden, and
prescientific. For Israel, these stipulations concern the nature of knowledge,
nature of man, and nature of society. These differing stipulations may lead to:
(1) lack of consensus about relevant variables, (2) differing descriptions of phe-
nomena, (3) differing evaluative criteria, and (4) differing models for the combi-
nation of parameters. That is, these multiple starting points (stipulations) lead to
different, and sometimes conflicting, research paradigms (images) and suggested
programs for organizational change, rather than equifinality of effort. Thus, re-
search results from differing paradigms are not necessarily additive (in Kuhn's
sense), since results are based on differing, value-bound, normatively chosen
dimensions and assumptions.
If the study of organizational behavior is multiply-imaged, what are some
implications for the development of theory? It will be suggested that unlike
problem solving within a paradigm, the union of our multiply-imaged results
does not progress towards a truer representation/explanation of the processes
which determine observable phenomena. However, these results do enhance

the quality of our knowledge. By accepting the multiplicities, we allow our-
selves the freedom to consider alternative ways of constructing social phe-
nomena. In fact, these multiple alternatives will provide a form of multilectic
for understanding our research endeavors.
To recapitulate, all theories can be considered middle range theories. Although
some middle range theories are intraparadigmatic, many middle range theories
are based on differing normative assumptions about the nature of man, society,
and knowledge. The rubric of middle range theory forces an awareness of these
partially overlapping, orthogonal, or even conflicting multiple images. Two
important questions immediately arise. First, "What is the nature and methodol-
ogy of multiply-imaged research?" Second, "What changes in ourselves are
necessary in order to conduct this research effectively?" An attempt is made to
answer these questions by drawing upon exemplars from fiction, philosophy,
and applied multivariate statistics.
EXEMPLARS
John Steinbeck
In The Log from the Sea o[Cortez Steinbeck (1941/1962, pp. 131-151) con-
siders two modes of thought-teleological and nonteleological. For Steinbeck,
teleological thought represents the search for causes and answers to the question,
"Why is X so?" He also notes that "conflicts may arise between any two or more
of the 'answers' brought forth by ... teleologies" (p. 143). Thus, Steinbeck may
be interpreted as struggling with the problem of conflicting, multiply-imaged
explanations. To resolve this problem, Steinbeck appeals to the need for a new
non teleological explanation in the form "X is so because it's so." Taken literally,
this formulation could be dismissed as naive description. However, for Stein-
beck, this nonteleological statement results from an awareness of a multiplicity
of explanations. The statement "X is so because it's so" subsumes causal (teleo-
logical) explanations, as well as descriptive statements, "since they [teleological
answers] are part of the picture even if only restrictedly true" (p. 144). The
non teleological statement allows for all possible modes of explanation, as well
as for multiple images within each mode. To know a thing includes all percep-
tions, causal and acausal, and their interrelationships:
The truest reason for anything's being so is that it is. This is actually and
truly a reason, more valid and clearer than all the other separate reasons,
or than any group of them short of the whole. Anything less than the whole
forms part of the picture only .... A thing may be so "because" of a thou-
sand and one reasons of greater or lesser importance .... The integration
of these many reasons which are in the nature of relations rather than reasons
is that he [man] is. (pp. 148-149).
Thus, Steinbeck accepts the existence of multiple images. His integrative,
non teleological statement subsumes all multiple images as a buzzing and alive
set of apparent conflicts, all nonconflicting by the fact that they each reflect
an aspect of what is. Knowledge then becomes the set of relations between our
multiple reasons (images) and the actual phenomenon of study. We now turn to
factor analysis in order to pictorially depict the nature of this nonteleological,
conflict-free thought.
Factor Analysis
The input for a factor analysis program is generally a correlation (or covariance)
matrix. The matrix is then transformed, mathematically and statistically, in
order to uncover underlying structures in the data and to test hypotheses about
such structures. The technical development of this technique led Cronbach
(1957) to state "Factor analysis is rapidly being perfected into a rigorous
method of clarifying multivariate relationships" (p. 675). Current researchers,
however, are becoming increaSingly aware that factor analytic results are a func-
tion of subjective decision rules regarding the nature of the data to be collected
for analysis, the number of factors, ope rationalizations of simple structure, types
of rotations, and assumed factor intercorrelations (cf. Francis, 1974). Thus, with
the same data, different researchers may produce differing structures, just as dif-
fering images of organizations may lead to differing middle range theories. As
Rozeboom (1966) says of factor analysis, "Aren't we just going in circles? Well,
yes-in a way we are, but sometimes the view from one point on a circle is more
interesting than from another" (p. 213). Thus, our multiple images are views
from differing points on the circumference of the circle. We are now ready for
the epistemological jump necessary for discovering the essence, indeed, the
center of the organizational phenomena that we study. That is, we are now
ready to take the Zen leap to ...
Points on the circumference of a circle, instead of merely looking to the self-
same centre, are at the centre all the time. This means, of course, that the
circle folds up, as it were, into the Void of the Unmanifest (Humphreys,
1968, p. 113).
Thus, just as with Steinbeck, the multiple images can not be conflicting at all,
for they are all aspects of the same center of the circle. The previous quote by
Humphreys concerns the Zen wholeness of multiple approaches. We now turn
to the underlying philosophy of that wholeness.
Zen
The Zen developmental process consists of three stages. The first stage, nonsense,
implies a world view that is unidimensional, naive, descriptive, and bounded by
Marxian (Groucho) humor. The second stage, logic, as typified by Western civili-
zation, is replete with dualistic notions (e.g., good/bad, logical/illogical, active/
passive) and emphasizes the rational search for an understanding of behavioral
phenomena. like Steinbeck, multiplicity of images provides the impetus for
a new mode of explanation. That is, we must comprehend that the part (middle
range images) is greater than the whole since the ''whole is complete, which is
finite; the part is unfinished, and that is infinite" (Humphreys, 1968, pp. 109-
110). The number of multiple Zen images is infmite. The second developmental
stage, logic, is incapable of subsuming such a multiplicity of images. In fact, in
Zen philosophy, the fmal function of the intellect is that it "enables us to com-
prehend at last that all is incomprehensible" (Humphreys, p. 14). This compre-
hension, and the need to encompass the logical conflicts of our middle range
theories, leads to the third Zen stage of nonsense (or super-sense). Here, like
Steinbeck, there is no tension between opposing images since "things are what
they seem to be and that is that" (p. 85) and there is "oneness of cause and
effect." The state of mind called satori is an awareness of such wholeness and
the simultaneity of conflict and harmony. At the third stage, one is willing to
experience all phenomena and to believe in an intuition that is beyond descrip-
tion and logic. For Zen, "the intellect may argue and debate; it may learn and
teach a vast amount about almost anything; it can never know. Zen is a matter
of experience" (Humphreys, pp. 13-14). At this third stage, knowledge is not
measurable, for it is experienced in its wholeness and totality of images at an
intuitive, qualitative level.
It is suggested here that Campbell's (1974) notion of "qualitative knowing"
is congruent with the above Zen experiential mode of knowing. This is apparent
in types of data discussed by Campbell (e.g., remembered experiences of partici-
pants, recorded participant experience, participant and nonparticipant observa-
tion, and the experience interview). For Campbell, "qualitative, common-sense
knowing of wholes and patterns provides the enveloping context necessary for
the interpretation of particular quantitative data" (p. 11, italics added). Thus,
the wholeness of our research is couched in qualitative, experiential knowing.
We now tum to a fmal exemplar, which suggests a state of mind necessary for
the encompassment of multiplicities.
Herman Hesse
The main character in Hesse's Steppenwolf, Harry Heller, believes himself to be

a victim of the struggle between two internal selves-Heller as Steppenwolf and
Heller as a member of the bourgeoisie. The Steppenwolf is an impulsive, emo-

tional being who is characterized as "a chaos of potentialities, instincts, and
impulses" (Hesse, 1929/1963, p. 129). To Hesse, the bourgeois self is typified by
one who "believes in the value of mere knowledge and its acquisition" (p. 78),
and by one whose "harvest is a quiet mind ... [which prefers] pleasant tempera-
ture to that deathly inner consuming fire" (p. 52). Heller learns, however, that
his struggle is not between two selves, but among a multitude of selves (multiple
images). Even though there is a Zen wholeness to Heller's existence, Hesse indi-
cates that the multiplicity is not resolved by rational, scientific processes:
Science is in this so far right as no multiplicity may be dealt with unless there
be a ... certain order and grouping. It is wrong insofar as it holds that one
only and binding and lifelong order is possible for the multiplicity .... We
demonstrate to anyone whose soul has fallen to pieces that he can rearrange
these pieces of a previous self in what order he pleases, and so attain to an
endless multiplicity of moves in the game of life (p. 192).
There may be at least as many ways of structuring the set of middle range
theories, leads to the third Zen stage of nonsense (or super-sense). Here, like
Hesse, it is humor- "laughter without an object, ... simply light and lucidity"
(p. 154). It is important to note that this laughter does not arise from the
futility of explanation or a nihilistic philosophy. On the contrary, Hesse's humor
is symbolic of a loving acceptance of the existence of our multiplicities. The
result of this acceptance is a life "in the atmosphere of ... magic possibilities. A
mere nothing suffices-and the lightning strikes" (p. 56). The symbolism in this
quote has arisen in more recent literatures. Regarding "possibilities," Sampson
(1977) indicated that differing assumptions could provide alternative solutions
to social problems. The "mere nothing" has become "a blank sheet of paper
enclosed, by way of explanation" (Salinger, 1965, p. 92). Finally, Weick (I 975)
has considered laboratory experiments as ''lightning marks" that leave experien-
tial traces that are "etched into the sensing apparatus of researchers" (p. 10).
Thus, these lightning marks affect the images that we bring to bear on our
research.
IMPLICATIONS FOR THE DEVELOPMENT OF THEORY
"Artifacts"
To reiterate, the notion of middle range theory has utility in that it leads to a
heightened awareness that multiple images of organizations coexist. These
multiple images are based upon differing normative assumptions (stipulations)
underlying the research (Israel, 1972; Ritzer, 1975; Sampson, 1977) which,
in turn, are a function of traits of the individual researcher, values of the current
culture, or criteria imposed by the organization that requests the study. The
nature of these normative assumptions suggests that it may prove useful to
consider the effect of: (1) organizational and cultural demand characteristics
on theory formulation as well as (2) personality traits and expectancies of
the researcher.
In their review of laboratory experimentation, Fromkin and Streufert (1977)
consider demand characteristics, experimenter expectancy, and experimenter
personality under the rubric of "artifacts" and note that such notions are also
applicable to field research settings. It is likely that both normative cultural
assumptions and images of working man set forth by organizations operate as
demand characteristics in the theory development process. Differences in de-
mand characteristics will lead researchers to differing images of individuals,
organizations, or their interrelationships. For additional support of this point,
consider Sherwood and Nataupsky's (1968) rmding that theoretical conclusions
based on empirical data are related to demographic variables of the individual
researcher. It is therefore suggested that a researcher's expectancies about
empirical outcomes are affected by both demand characteristics of the situation
and by researcher personality profiles. Such exp ;tancies affect the type of
questions asked, the modes of data collection, and the interpretation of empiri-
cal results-leading to a self-fulfilling prophecy about the "truth" of the original
normative assumptions. Rather than dismiss the above processes as "artifactual,"
it is suggested that the effects of demand characteristics and personality variables
on theory development provide, in themselves, phenomena that are deserving
of study.
Humor and Creativity
All exemplars clearly suggest that we accept the existence of multiple images and
the resultant sets of apparent contradictions. However, these contradictions are
apparent only at the rational, quantitative level of thought. The acceptance of
contradictions (by Steinbeck and Zen philosophy) as necessary reflections of a
greater All implies that "contradictions" are no longer contradictory, for they
are all reflections of the same universe of human behavior. That is, the circum-
ference of the circle is at the center. How are we to accept these maddening
notions? Hesse and Zen philosophy state very clearly: laughter and humor. That
is, humor becomes a capacity (perhaps a loving capacity) to laugh at/with our-
selves and the simultaneity of multiple images. The humor is of harmonious, not
nihilistic, origins. In this marmer, we now become aware of the normative nature
of our assumptions, their effect on the development of theory, and their multi-
plicity. For von Neuman (1958) the beauty of the brain, unlike the computer,
is exactly this capacity to cope and function in "noisy," sometimes, chaotic,
environments. Only through this humor within ourselves can we effectively dare
to leap among the points of the Zen circle.
Such realizations are important for a creative, investigative science. All exem-
plars have suggested research approaches that are considered to be preconditions
for creativity: i.e., humor, acceptance of multiplicity, novel viewpoints (from
the circle), tolerance for ambiguity, and divergence (cf. Barron, 1969; Gowan,
Demos, and Torrance, 1967). Thus, if we accept the implications of multiple
fundamental assumptions, organizational research may become a truly cre-
ative endeavor.
Causality
Steinbeck suggests that causal (teleological) explanation is neither the sole nor
the major mode of explanation of organizational phenomena. Indeed, both Zen
philosophy and Rozeboom's comment on factor analysis depicted the circularity
of explanation. In general linear model theory, recent attention to the notion of
circularity has led to the analysis of structural equations that consider feedback
loops, changes in parameters, and circularity of behavioral processes (Heise,
1975; Hanushek and Jackson, 1977). Finally, it may be fruitful to accept the
extreme notion of synchronicity (Jung, 1955/1973) and assume that since space
and time are relative, many theoretical explanations are necessarily acausal.
Quality
Epistemologically, the above exemplars suggest that acceptance of multiple

images implies acceptance of Zen experience and Campbell's qualitative know-
ing. The etymology of "qualitative" is the Latin word qualitas, meaning quality.
It is the quality dimension that is used by Pirsig (1975) in Zen and the Art of
Motorcycle Maintenance to solve the dialectic between the pseudo-dimensions
of "Classic Knowledge" (reason) and "Romantic Knowledge" (feelings). For
Pirsig, the criteria for quality vary among individuals and quality is undefinable,
although one knows quality when one experiences it. Pirsig states:
Quality, value, creates the subjects and objects of the world. The facts do
not exist until value has created them. (p. 304)
This is a reaffirmation of the fact that normative assumptions create our mul-
tiple images. The factor analysis exemplar indicates that a strict return to
quantitative and rational techniques does not avoid the existence of multiple
imagery. That is, the "objective" multivariate techniques, such as factor analysis,
yield results that are a function of subjective, value-laden decisions (such as the
choice of "objective" rotation schemes in factor analysis).
If multiple images are created from multiple value structures (individual,
organizational, or cultural), research results are comprehensible only in the
light of their respective value bases. Perhaps each journal article should contain
a statement, by the author(s), indicating reasons (both personal and otherwise)
for choosing the particular research questions, criteria, and modes of analysis.
Additionally, authors might include personal growth experiences as a result of
the research inquiry, by indicating how this growth affected implementation of
the study and interpretation of the data (Smith, 1978). In this way, social
science knowledge becomes a process of the interplay between value structures
and the collection of research data.
"Progress Is Our Most Important Product"
There is an important implication for the notion of progress of organizational

theorizing. If points on the circumference of our Zen circle are also at the center
of that circle (Le., if our multiple images are at the center), then our theories do
not progress towards any fmal "truth" or unification of theory, for they are
already part of that truth. The multiple images do add richness to our under-
standing of what is All. As noted previously, acceptance of the existence of
multiplicities provides a regained freedom to consider alternative ways of con-
structing the SOcial/organizational environment. We are changing, the All is
changing, and the resolution of our multiple images is that there is no single
unifying resolution.
The position of this paper has been that the rubric ''middle range theory"
provides an awareness of the multiple imagery of organizational research. In
contrast, some researchers argue that middle range theories provide piecemeal
solutions that await a unified combination (e.g., Weick, 1974; Pinder and Moore,
1977). If investigations under the rubric of "ecological fallacy" (Hannan, 1971)
are correct, this wait may be forever. These investigations of ecological fallacy
indicate that theoretical formulations at one level of analysis (Le., level of
imagery) do not necessarily conform to formulations at other levels of analysis
and that translation between the two levels is often impossible. Once again, these
levels are "simply" a subset of the multiple reflections about the circumference
of the circle. An attempt at a theory of unification for these multiple images
leads immediately to a form of Russell's paradox, "Is the set of all sets a set?"
(Whitehead and Russell, 1927). That is, "Is the set of all middle-range theories
a theory?" For Whitehead and Russell, the answer is "no":
The subjects of our predications in fact fall into an infinite hierarchy of

"logical types" -individuals, classes of individuals, classes of classes of indi-
viduals ... and so on -and what can be intelligibly said about objects of one
type cannot be said about objects of another type (Prior, 1962, p. 284).
Middle range theories provide a multilectic, not dialectic, outlook. The dialec-
tical constructs of thesis and antithesis are trappings of dualistic, bipolar, Western
thought. These constructs are two endpoints of a straight line. In contrast, the
multilectic approach considers all points on the circle as alternative reflections of
what is. Thus, the value of middle range theory is not a unified synthesis but a
comprehension of multiple, alternative, simultaneous images. As Blyth (1942,
p. 84) states, ''Truth lies beyond the extremes, not in the middle."
REFERENCES
Barron, F. Creative Person and Creative Process. New York: Holt, Rinehart, and
Winston, Inc., 1969.
Blyth, R.H. Zen in English Literature and Oriental Classics. San Francisco:
Heian International Pub. Co., 1942.
Campbell, D. T. Qualitative Knowing in Action Research. Paper presented at the
meeting of the American Psychological Association, New Orleans, September,
1974.
Cronbach, L.J. The two disciplines of scientific psychology. American Psychol-
ogist, 1957,12,671-684.
Francis, 1. Factor analysis: Fact or fabrication. Mathematical Chronicle. 1974,
3,9-44.
Fromkin, H.L., and Streufert, S. Laboratory experimentation. In M.D. Dunnette
(Ed.), Handbook of Industrial and Organizational Psychology. Chicago: Rand
McNally Publishing Co., 1976.
Gowan, J.C.; Demos, G.D.; and Torrance, E.P. (Eds.). Creativity: Its Educa-
tional Implications. New York: John Wiley and Sons, Inc., 1967.
Hannan, M. T. Problems of aggregation. In H.M. Blalock (Ed.), Causal Models in
the Social Sciences. New York: Aldine-Atherton, Inc., 1971.
Hanushek, E.A., and Jackson, J.E. Statistical Methods for Social Scientists. New
York: Academic Press, 1977.
Heise, D.R. Causal Analysis. New York: John Wiley and Sons, Inc., 1975.
Hesse, H. Steppenwolf. (B. Creighton, trans.) New York: Holt, Rinehart, and
Winston, Inc., 1963. (Originally published, 1929.)
Humphreys, C. Zen Buddhism. New York: MacMillan Co., 1968.
Israel, J. Stipulations and construction in the social sciences. In J. Israel and
H. Tajfel (Eds.), The Context of Social Psychology: A Critical Assessment.
New York: Academic Press, 1972.
Jung, C.G. [Synchronicity] (R.F.C. Hull, trans.). Princeton, New Jersey: Prince-
ton University Press, 1973. (Originally published, 1955.)
Kuhn, T. S. The Structure of Scientific Revolutions (2nd ed.). Chicago: Univer-

McKelvey, B. Guidelines for the empirical classification of organizations.
Administrative Science Quarterly, 1975,20,509-525.
Nord, W. The failure of current applied behavioral science: A Marxian perspec-
tive. Journal of Applied Behavioral Science, 1974,10, 557-578.
Pinder, C. C. Facets of a Movement Toward Middle Range Theory Development
in Organizational Behavior Science. Unpublished Manuscript, 1977. Faculty
of Commerce and Business Administration, University of British Columbia,
Vancouver, Canada.
Pinder, C.C., and Moore, L.F. The resurrection of taxonomy for the develop-
ment of middle-range theories of organizational behavior. Administrative
Science Quarterly, March 1979.
Pirsig, R.M. Zen and the Art of Motorcycle Maintenance. New York: Bantam
Books, 1975.
Prior, A. N. Formal Logic (2nd ed.). Oxford: Oxford University Press, 1962.
Ritzer, G. Sociology: A multiple paradigm science. The American Sociologist,
1975,10, 156-157.
Rozeboom, W. W. Foundations of the Theory of Prediction. Homewood, Illinois:
Dorsey Press, 1966.
Salinger, J.D. Raise High the Roof Beam, Carpenters. New York: Bantam Books,
1965.
Sampson, E. E. Psychology and the American ideal. Journal of Personality and
Social Psychology, 1977,35,767-782.
Sherwood, J.J., and Nataupsky, M. Predicting the conclusions of negro-white
intelligence research from biographical characteristics of the investigator.
Journal of Personality and Social Psychology, 1968,8,53-58.
Smith, K. Personal communication, January, 1978.
Steinbeck, J. The Log from the Sea of Cortez. New York: Viking Press, 1962.
(Originally published, 1941.)
University of British Columbia announcement of a conference entitled, "Toward
middle-range theories of organizations and organizational behavior." Sent in
1977. Faculty of Commerce and Business Administration, University of
British Columbia, Vancouver, Canada.
Von Neumann, J. The Computer and the Brain. New Haven, Connecticut: Yale
University Press, 1958.
Weick, K. E. Middle range theories of social systems. Behavioral Science, 1974,
19,357-367.
- - . Laboratory Experimentation with Organizations: A Reappraisal. Paper
presented at the National Academy of Management meetings, New Orleans,
August, 1975.
Whitehead, A.N., and Russell, B. Principia Mathematica, (Vol. 1, 2nd ed.).
Cambridge: Cambridge University Press, 1927.
9 MIDDLE RANGE THEORIES
Clusters of Clusters
of Organizational Phenomenon
Randall S. Schuler,*
The Ohio State University
INTRODUCTION
The study of organizations and organizational behavior generally focuses on

single constructs of organizational phenomenon, e.g., leadership, task design,
motivation, organizational design, and so on. For many of these phenomenon,
theories have been developed and/or hypotheses have been stated and examined.
For some phenomenon, however, theories are still to be articulated. Recently,
suggestions for integrating theories of a few of these single phenomena have been
made and specific statements of their theoretical relationships explicated (House,
1971; Hackman and Oldham, 1976; Schuler, 1977a). It is suggested here that
what is occurring in the study of organizations and organizational behavior, and
what is necessary, is efficient and effective integration of those theories of single
organizational phenomenon into fewer theories spanning these single theories.
These occurrences are descriptive of development, use, and function of middle
range theory in organizational behavior and theory. To advance our understand-
ing of all organizational phenomena, it is important to continue to develop more
middle range theories, to discuss issues related to the occurrence of these middle
*The author wishes to express his thanks to R.l. House for several critical comments,
which added substantially to this paper.
113
range theories, and also to search for middle range theory(ies) to integrate those
middle range theories. Middle range theories that integrate other middle range
theories are called trans-middle range.
Before advancing an example of a trans-middle range theory, let's look at
middle range theory, including reasons for and properties of middle range theory
in the study of organizations and organizational behavior. It is argued that some
organizational phenomenon are currently described by middle range theories but
that these middle range theories are unconnected and unintegrated.
MIDDLE RANGE THEORY
Middle range theory is not uncommon in the study of organizational phenomena

(e.g., March and Simon, 1959; Romans, 1961). Its systematic application and ac-
ceptance as a deliberate vehicle for epistemological advancement have, however,
been less common (Merton, 1957). A systematic application of middle range
theory to organizational phenomena, nevertheless, is feasible and necessary for
greater advancement in the study and understanding of organizational behavior
and theory.
The use of the term middle range theory suggests a singular treatment and
construction of organizational phenomena.! Such treatment and construction
at this time would not be a sufficient paradigm by which to understand, explain,
and predict all the several diverse, though interconnected, patterns of complex
organizational phenomenon within the total cluster of organizational behavior
and theory phenomena. Instead, several middle range theories would probably
be more appropriate at this time to articulate and describe clusters of organiza-
tional phenomenon, which include a limited set of variables that have a greater
degree of relationship within each cluster than across clusters. Each single cluster
represents a single organizational phenomenon such as leadership. Phenomenon
such as leadership should represent a cluster of variables with construct and em-
pirical similarity and theoretical relationship. All variables within a cluster
should thus be highly related. The identification of many clusters has implicitly
been done in organizational behavior and theory, as evidenced by theories of
leadership, task design, motivation, and so forth. These theories, by some defini-
tions, may even be only hypotheses, but are described here as examples of nar-
row range theory. Then integrating or joining some of these single clusters into
larger clusters, on the basis of the degree of similarity (again, empirical and con-
struct) and theoretical relationship among single clusters forms the basis for a
middle range theory. Greater understanding of these multiple clusters of orga-
nizational phenomenon by their own theory, appropriately labelled middle
range, is a prerequisite to a trans-middle range theory, broader than, but yet
as crystalized, testable, and consistent as any single middle range theory.
MIDDLE RANGE THEORIES 115
It is argued here that the advancement and understanding of organizations

and organizational behavior is best served by conceptualizing organizational
phenomenon in integrative, middle range terms and by developing and using
middle range theories for clusters of these phenomenon and then tying them
together. Better middle range theories would be the ones that can integrate (tie
together) several other diverse middle range theories, that is, trans-middle
range theories.
Merton suggests that support for the development and importance of middle
range theory for the advancement of knowledge is not new. Bacon, Mill, and
others indicated the importance of a gradual series of empirically confirmed
intermediate theories. These "theories (those of middle range) are empirically
grounded, that is, with sets of confirmed hypotheses, not merely organized
descriptive data or empirical generalizations or hypotheses that remain logically
disparate or unconnected" (Merton, 1957, p. 61). Merton, although discussing
sociological theory, indicates that middle range theories lie between ''the minor
but necessary working hypotheses (these may be examples of narrow range the-
ory) that evolve in abundance during day-to-day research (that is, the common-
sense procedure used by all of us every day) and the all-inclusive systematic
efforts to develop a unified theory that will explain all the observed uniformities
of social behavior, social organization ... " (Merton, 1957, p. 39). The use of
"narrow range theory" here focuses Merton's working hypotheses notion on
descriptions of relationships of variables within single clusters only. Middle range
theory involves abstractions and multiple clusters, but they are close enough to
observed data to be incorporated into propositions that permit empirical testing.
Middle range theory appears to be a necessary though not entirely sufficient
level of theory to explicate organizational behavior and theory (Merton, 1957).
It is necessary because a narrow range theory does not incorporate a sufficient
number of variables for practical application (BenniS, Benne, and Chin, 1962)
and because a broader range (grand) theory would have problems going beyond
general terms, definitions, typologies, and descriptive considerations of broad
topics in organizational analysis (Zaleznik, 1967). Middle range theory is not
entirely sufficient, however, because no single middle range theory can by defmi-
tion incorporate all of the variables within organizational behavior and theory,
although use of trans-middle range theories begins to approach sufficiency.
Weick (1974) and Merton (1957) also argue for the conceptualization of
organizational phenomena within discrete, delimited, therefore, manageable,
and falsifiable boundaries. Without these boundaries, the conceptualization of
organizational phenomena is not only too abstract to be useful for the research,
but also misrepresentative of the phenomena being conceptualized. The implicit
analogies made by these writers is that conceptualizations without boundaries
describe grand theories of organizational phenomena and that grand theories (for
example, in the social sciences) tend to be premised on systems properties, but
that while certain phenomenon within organizations closely approximate sys-

tems properties, that is, are tightly coupled, other phenomenon do not contain
these properties, are less tightly, more loosely, coupled (Glassman, 1973; Weick,
1974; 1976).
So while Merton primarily argues that grand theories are too abstract to be
useful for research and better understanding of organizations and organizational
behavior, Weick argues primarily that they misrepresent some phenomenon with-
in organization. Specifically, they fail to distinguish the degree of coupling
(Le., intensity and direction of relationship) across phenomenon. Both of these
arguments represent two major thrusts against development of grand theory and
at the same time two major reasons for development and use of theories of the
middle range for studying organizational phenomena. Middle range theories can
recognize the degree of coupling of the relationships among variables in organiza-
tions by dealing with clusters of variables having more in common with each
other than other variables that can be grouped in other clusters and initially
examined separately.
Filley, House, and Kerr (1976) enumerate the characteristics of a good
theory: internal consistency, test of empirical validity, operationality, and
generalizability. Bennis, Benne, and Chin (1962) suggest that for a practical
application, a good theory should also account for interrelated levels in a con-
text (e.g., individual, group, and organization levels). These characteristics thus
require that relationships among variables be explicitly hypothesized, their
relationships consistent, and congruent with reality, the hypothesized relation-
ships verifiable and testable, and the relationships and results applicable across
situations or conditions. In addition, methods to measure and control the hy-
pothesized variables must be available in order that the hypotheses derived
from the theory can be tested and replicated in public settings. Merton suggests
that these characteristics of good theories are the essence of theories of middle
range. Middle range theories are not derived from grand theories, although they
may be consistent with them; therefore, the proof of the middle range theory is
not in comparison to ideas derived from a more general theory, but in its appli-
cability (Le., its testability, empirical validity, and generalizability).
Arguments against middle range theory are generally made in favor of grand
theory. These arguments, which Merton rejects or argues away, relate to its lack
of acceptance by the ancestors of the field of inquiry (e.g., sociology, its lack of
generalizability, and its tendency to induce fragmentation in the total area of
inquiry). Merton suggests, to the contrary, that middle range theories were
anticipated by sociology's intellectual ancestors, and their characteristics are
generalizable across clusters of phenomenon and can tie together clusters of
phenomenon, not produce division among them.
Those two characteristics of middle range theory identified by Merton are
important because they comport with Weick's suggestion that organizations be

seen as having many variables among which there are varying degrees of coupling.
This argument suggests that phenomenon within organizations are not so tightly
coupled as to be unique from each other, nor so uniformly coupled as to be
treated as the total organizational phenomena at this time. Rather, the total
organizational phenomena are clusters of organizational phenomenon with vary-
ing interdependencies over time, which are themselves also common threads by
which these phenomenon may be connected with other phenomenon but yet
maintain their uniqueness.
As mentioned above, each cluster of organizational phenomenon may have
its own theory to explain its events or variables_ The development of middle
range theories that can incorporate single clusters of organizational phenomenon,
and those integrative theories that can incorporate other middle range theories
are desirable goals in the development and use in middle range theory. Middle
range theories should be able to explain more organizational phenomenon
more efficiently and effectively and perhaps reconcile inconsistencies in re-
search in single phenomenon. These goals, in fact, are necessary for middle range
theory to advance our knowledge of organizational behavior and theory.
There are several important middle range theories of clusters of single phe-
nomenon in organizational behavior and theory. Unfortunately, integrative
middle range theory(ies), though necessary is/are unusual in organizational
theory and behavior. It is important, therefore, that for these middle range the-
ories to continue to be useful they be integrated with or by other theories. To
illustrate the concepts of middle range theory developed above, the task design
theory of Hackman and Lawler (1971) and the leader behavior theory of House
(1971), as examples of middle range theories, will be examined. Then an ex-
ample of the necessary step beyond middle range theories to trans-middle
range theory, which incorporates the task and leader behavior theories will be
discussed.
MIDDLE RANGE THEORY: STAGNATION OR GROWTH

FOR ORGANIZATIONAL BEHAVIOR AND THEORY
Thus, characteristics of middle range theory(ies) and relevance to the criteria

of good theory indicate the importance of this method for scientific inquiry
and accumulation of knowledge. Neither grand theory nor narrow range theory
appears to meet the requirements of good theory nor do they either fit or de-
scribe the reality of organizations. However, in order for the application of
middle range theory to be fruitful for understanding organizational phenomena,
it must avoid the tendency to dust bowl empiricism (theorism), that is, the
production of middle range theories, which don't go or haven't successfully gone

beyond themselves and/or fail to integrate other clusters of phenomenon within
an organization that have systemic characteristics different from their own. The
next two examples of middle range theories represent important clusters of
phenomenon in organizational behavior. These are only two of several theories
that could have been chosen, but each is a good example of a middle range
theory because they integrate two or more single clusters of phenomenon. The
task design theory incorporates task dimensions and individual differences to
predict employee responses and the leader theory incorporates leader behavior,
individual motivation, and task design to predict employee responses. Although
they are good examples of middle range theory, it is argued here that they
remain to be sufficiently integrative with other middle range theories or even
with each other. Further examination of the leadership theory has, however,
provided substantial inspiration for an integrative, trans-middle range theory
that can incorporate both of these middle range theories.
Hackman and Lawler (1971) developed a theoretical model of the relation-
ships between task design, individual differences (growth need strength) and
employee responses. 2 Even within this cluster of two single clusters of organi-
zational phenomenon (task and individual differences) there are choices to be
made that reflect or describe the nature of the middle range theory for those
phenomenon. For example, task design was described by five components:
identity, variety, significance, feedback, and autonomy. Other dimensions of
task were not included; for example, task variability and difficulty (Van de Ven
and Delbecq, 1974). The task dimensions chosen by Hackman and Lawler have
been useful, at least conceptually, in integrating task design theory with indi-
vidual differences but have been less useful in integrating with middle range
theories such as the Path-Goal Theory of leadership.
The Path-Goal Theory of leadership (House, 1971) represents a theory of
leader behaviors and employee responses. Based upon empirical data, House
hypothesized relationships between leader behaviors and employee responses
that would better explain these relationships by including task design charac-
teristics. These task design characteristics, chosen by House, however, had more
in common with task variability and difficulty than with Hackman and Lawler's
(1971) five task dimensions. Again, the Path-Goal Theory represents a particular
middle range theory of a cluster of single clusters of organizational phenomenon
(i.e., leader behaviors, motivation, and task design). By choice and necessity of
hypothesis generation and testing, only certain leader behaviors, task characteris-
tics, and employee responses have been examined in the Path-Goal Theory.
Although each of these separate middle range theories represents several
single clusters of organizational phenomenon, neither has explicitly joined other
clusters of organizational phenomenon nor has either middle range theory been
successful in consistently explaining the clusters of phenomenon it incorporates.

The individual difference variable of growth need strength, for example, appears
not to be a consistent moderator of Hackman and Lawler's task characteristics
(Schuler, 1977b); therefore, it is not able to reconcile or explain the conditions
in which task design will be more effective or explain more variance in employee
responses to task design. Examinations of the Path-Goal Theory of leadership
have not consistently supported the hypothesized relationships between leader
initiating structure behavior and employee responses even when accounting for
the nature of the task.
Will the usefulness and/or development of these two important middle range
theories end? Each of these middle range theories, represents an important and
necessary step in the use of middle range theory to provide an understanding
of some phenomenon in organizational behavior and theory. The usefulness of
middle range theories such as the Hackman and Lawler task design and the Path-
Goal leader behavior theories, however, may end if their development ends.
Their development will end if they are not integrated with other middle range
theories. Thus, they are susceptible to falling into the trap of dust bowl theorism.
But how can the tendency to dust bowl theorism be avoided? In addition, and
more generally, how can clusters of organizational phenomenon be recognized
and defined to suggest new middle range theories of organizational phenomenon
that can be integrated or can serve as integrators for other new or existing
middle range theories?
TRANS-MIDDLE RANGE THEORY
Generally, dust bowl theorism can be avoided by developing middle range

theories of/for middle range theories, that is a trans-middle range theory. As
discussed, each middle range theory represents a cluster of single clusters of
organizational phenomenon in the total cluster of organizational phenomena.
Single clusters will be identified by the relative degree of coupling (referred to
above as empirical similarity) among variables (Weick, 1974), their construct
similarity and theoreticallinkages. 3 The degree of coupling, the construct sim-
ilarity, and theoretical linkages among variables within a single cluster of organi-
zational phenomenon should be greater than the theoretical linkages and the
degree of coupling and construct similarity of those variables with variables in
other clusters of phenomenon. Combining single clusters will constitute larger
clusters that can be described by middle range theory. These larger clusters
should also be identified by the degree of coupling and construct similarity and
theoretical linkages. The trans-middle range theory will then integrate across
these larger clusters starting again with those clusters with greatest degree of
coupling and construct similarity and theoretical relationship. It is argued that

the degree of coupling and the construct similarity and theoretical linkage are
important because they are necessary conditions for efficient integration and
reconciliation of inconsistent results of relationship among variables within
clusters, across single clusters in larger clusters, and eventually across larger
clusters. Greater degrees of coupling and more similarity are also associated with
more explained variance of the variables in the single clusters of phenomenon
and in the single clusters in the larger clusters being integrated. Trans-middle
range theories that span larger clusters of phenomenon are thus developed by
attention to the degree of coupling and construct Similarity among variables in
single clusters, single clusters in larger clusters, and their theoretical relationships.
An example of a trans-middle range theory will now be developed and dis-
cussed. In addition, some general issues that arise in the application of middle
range theory and development of middle range theories will be presented. 4
This illustration, using components of the Path-Goal Theory, focuses on only
one specific leader behavior hypothesized to be related to employee responses
because research results using that leader behavior have been inconsistent. This
example illustrates how the use of middle range theory can be used to expand
understanding of organizational phenomenon and offer reconciliation of incon-
sistent results when clusters of phenomenon are grouped by a trans-middle range
theory. Note that this example of trans-middle range theory, developed post
hoc, is based on empirical and construct similarity and theoretical relationships.
Systematic development of trans-middle range theories, however, may also pro-
ceed by identification of a priori construct similarity and theoretical relationship
and then empirical analysis of the degree of coupling among the variables.
The Path-Goal Theory hypothesized that leader initiating structure is posi-
tively related to employee satisfaction when tasks are complex and unstructured
and negatively related when tasks are simple and structured. Research examining
these relationships has provided inconsistent results. The task variable failed to
reconcile the inconsistencies. Another possible route for reconciliation was
another cluster of organizational phenomenon.
The Theory implicitly suggested a second possibility. That is, it suggested that
the construct of role perceptions be used. Since leader initiating structure be-
havior describes clarifying, initiating, and directing behaviors and role percep-
tions describe the clarity of demands and the conflict among those demands
perceived by employees, they have some construct similarity. These two clusters
were also linked because of the degree of coupling among these clusters was
empirically determined by Schuler (1978) in the correlation matrix containing
the variables in the two clusters.
The results of one study suggested that this integration of leader behavior
and role perceptions is useful in describing employee responses (Schuler, 1978).
In that study, however, another cluster of organizational phenomenon, struc-

ture, was also incorporated. It was also included because it had construct and
empirical similarity to role perceptions (Schuler, 1978). Thus, the integrative
framework for examining these three clusters of organizational phenomenon was
supported by data, with one of the three clusters, role perceptions, used as the
integrating cluster for the other two. Each cluster had its own theory, and the
role perception cluster theory became a trans-middle range theory by providing
a means to integrate two specific leader and structure cluster theories.
This example of an integrative framework not only used several existing
middle range theories and created a trans-middle range theory from one of them,
but it also provided specific hypotheses, measures of the clusters of phenomenon
and empirical test results (Schuler, 1978). The results provided direction for
increased understanding of organizational phenomenon by going across clusters
of organizational phenomenon as suggested by Merton (1957). Although this is a
useful example of the development and use of middle range theory and theories,
there are still a number of important general, though abstract, issues to be ad-
dressed in using middle range theory for the improvement of our understanding,
prediction, and integration of organizational phenomenon.
GENERAL ISSUES AND SUMMARY
1. How can middle range and trans-middle range theories be developed and
used for advancing the understanding of and defining the total area of
organizational behavior and theory? As seen from the discussion of mid-
dle range theories such as the Path-Goal Theory of leadership the question
arises, "Can the boundaries of any inquiry be naturally determined or
must they be arbitrary?" The analysis of leadership, for example, could
be regarded as an area of inquiry in political science and the use of role
perceptions as an artifact of sociology. The total area of organizational
behavior and theory has, of course, drawn extensively from these other
grand areas mentioned. Although organizational behavior and theory has
developed some uniqueness from these areas, it continues to take advan-
tage of aspects of those areas which serve its own uses, help reconcile
inconsistencies, and/or explain more variance in the phenomena currently
within the purview of organizational behavior and theory. Thus our major
area of inquiry (Le., organizational behavior and theory) in essence is also
a cluster-though perhaps grand-within the universe of epistemology,
seeking other grand clusters, which may have common theoretical and
empirical threads, with which to integrate (Driggers, 1977). For the
moment, any specific middle range theory of organizational phenomenon
within organizational behavior and theory may be several levels removed

from this universe of epistemology, but its area of inquiry can be defined
and developed from those grand areas which "directly" serve to advance
the total area of organizational behavior and theory as well as from the
total area of organizational behavior and theory itself.
The scope of epistemology, however, from which understanding of the
total organizational phenomena can be augmented should be regarded as
transient, for what is "directly" today may be "indirectly" tomorrow.
Furthermore, investigation of other areas of epistemology, previously
overlooked because of their nonsystemic relationships with the study of
the total organizational phenomena, may reveal their importance for the
study of organizational behavior and theory. Consistent with this, it
might also be argued, using the coupling paradigm of Weick (1974), that
what constitutes a particular scope of investigation for organizational be-
havior and theory may change by the changing degree of coupling among
related grand phenomena and their theoretical relationships. In addition,
what constitutes a system may not be a functional way to delimit a grand
area of inquiry either, because all the grand phenomena now defined as
being part of organizational behavior and theory possibly do not meet all
the requirements of systems but nonetheless have important relationships
with the total phenomena of organizations and organizational behavior
(Weick, 1974).
2. Within this total area of organizational behavior and theory, how are clus-
ters of phenomenon defined or identified? Defining or identifying clusters
is necessary because they establish the parameters for middle range the-
ories. As mentioned above, clusters are identified by the degree of the
coupling, the construct similarity, and theoretical linkages among variables
to constitute a single cluster. Although the interrelationships among vari-
ables within a cluster may be greater than across clusters, the amount of
variance in a dependent variable explained by a single cluster may prove
inconsistent or insufficient. It is necessary, therefore, to seek a larger clus-
ter that can integrate several single clusters, becoming in essence a cluster
for which a middle range theory can be developed. Then these larger
clusters, each with its own middle range theory, are integrated by a trans-
cluster. Identification of trans-clusters is also a function of the construct
similarity and theoretical relationships. In addition, these clusters are de-
termined empirically by investigating the degree of coupling. Thus, the
actual clusters in a trans-cluster as well as the actual variables within a
cluster may vary. Each trans-cluster may then be described by a middle
range theory of one of the larger clusters, which upon being used to inte-
grate several larger clusters becomes a trans-middle range theory as in the
case with role theory. Schuler (1978) integrated a middle range theory of
task design, a middle range theory ofleader behavior, and a middle range
theory of organizational structure by a middle range theory of role per-
ceptions, which then became a trans-middle range theory. The same clus-
ters of phenomenon could have been integrated with another middle
range theory and also with other theories of middle range for task, leader
behavior, and structure.
For example, Silverman (1971) and Collins (1975) have attempted to
integrate two unique clusters of phenomenon in sociology: conflict and
action. Each attempt is based upon a different middle range theory of
conflict and action and, therefore, has a different framework for the de-
velopment of the trans-middle range theory. Silverman and Collins' the-
ories show the feasibility and perhaps necessity for several theories,
perhaps even conflicting theories, within each cluster or across clusters
of phenomenon (Benson, 1977).
Thus, the determination of the best middle range theory or best trans-middle
range theory may be difficult, although not impossible. This determination of
the best middle range theory and the best trans-middle range theory, for exam-
ple, could be made within the framework or rule of Theoretic Blockage ex-
plained by Driggers (1977). What is important, however, is that regardless of
the middle range theories used, they must be examined both theoretically and
empirically within themselves and across other clusters seeking an integrating
cluster. Determination of the middle range theory for a cluster depends upon its
ability to explain the phenomenon and/or resolve inconsistencies, but also it
depends upon the efficiency with which it can be integrated with other clusters
or be an integrator for them. Thus, utilization of middle range theory as dis-
cussed here, by recognizing the differing degrees of coupling among variables,
their theoretical linkages and construct similarity, transcends specific approaches
(e.g., structural functionalism) and becomes trans-theoretic in orientation
(Spinner, 1973). It is, therefore, suggested here that generation of efficient
middle range and trans-middle range theories is the best hope for increased under-
standing and advancement of knowledge of the total organizational phenomena.
CONCLUSION
This paper has discussed the importance of middle range theory within the con-
text of increasing our understanding of organizational phenomena. Grand
theory and narrow range theory appear to be inefficient or incapable to advance
understanding. The middle range theories, of which there are now several in
organizational behavior and theory, must be developed, utilized, and linked

together or integrated to avoid the label and fate of dust bowl theorism. Sys-
tematic development of new middle range theories for organizational phe-
nomenon would increase the likelihood of developing several trans-middle range
theories. Systematic analysis of the current middle range theories may also in-
crease the likelihood of developing trans-middle range theories. For example, a
middle range theory of role perceptions was offered here as a trans-middle range
theory to integrate the existing middle range theory of leader behavior described
by the Path-Goal Theory and a theory of organizational structure. Empirical
results support such a conceptualization of these clusters of phenomenon and
show the fruitfulness of middle range theories when they are linked by trans-
middle range theories to help resolve inconsistencies, increase variance explained
in dependent variables, and expand understanding within and across clusters of
organizational phenomenon.
Finally, the middle range theory method is important for organizational
behavior and theory because it requires an integrative mode of thinking and
conceptualizing by the researcher. Although each researcher may not be success-
ful in developing a trans-middle range theory, at least the researcher will be
aware of the need to recognize possible relationships among phenomenon. This
recognition may assist the researcher in a more cautious, but richer interpreta-
tion of data, a more thorough search for reconciliation of inconsistent findings,
and a desire to explain more variance in the phenomenon under investigation.
NOTES
I. Although singular, the term phenomenon will be used when referring to one, two or
more single clusters of variables, or larger clusters. The term phenomena will be reserved for
reference to the total set of clusters that describe organizational behavior and theory. Thus,
leader behavior and task design are two clusters of variables and are referred to as two phe-
nomenon within the total set of clusters referred to as the total organizational phenomena.
2. The remaining discussion uses employee responses as the dependent variable for
which maximizing variance is desired. Ouster identification is usually made on the basis of
construct and empirical similarity and theoretical relationships of the independent modera-
tor, and intervening variables.
3. Clusters would ideally be identified fIrst by analysis of the theoretical linkages and
construct similarity of the current universe of variables in organizational theory and be-
havior. This identification process is similar to that used in the development of construct
validity for any scale or test. The second step in cluster identification is the determination
of empirical similarity or the degree of coupling. Since the degree of coupling refers, in part,
to intensity and direction of the relationship between variables, the determination of the
degree of coupling can be done by examining the correlation matrix of the variables iden-
tified by construct similarity.
4. It is recognized that research in organizational behavior and theory is subject to

operationalization problems in addition to the theoretical problems discussed here, par-
ticularly in developing reliable and valid measures of the constructs used. Although oper-
ationalization is an important step in the advancement of understanding organizational
behavior and theory, it is regarded here as a step that precedes the theoretical step. Due
to the limited space, this paper limits itself to this theoretical step only.
REFERENCES
Bennis, W.G.; Benne, K.D.; and Chin, R. The Planning of Change: Readings in
the Applied Behavioral Sciences. New York: Holt, Rinehart and Winston,
1962.
Benson, J. K. Innovation and crisis in organizational analysis. The Sociological
Quarterly, 1977,18, 3-16.
Coilins, R. Conflict Sociology: Toward an Explanatory Science. New York:
Academic Press, 1975.
Driggers, P.F. Theoretical blockage: A strategy for the development of organiza-
tional theory. The Sociological Quarterly, 1977,18, 143-159.
Filley, A.C.; House, R.J.; and Kerr, S. Managerial Process and Organizational
Behavior (2nd ed.). Glenview, lllinois: Scott, Foresman and Company, 1976.
Glassman, R.B. Persistence and loose coupling in living systems. Behavioral
Science, 1973,18,83-98.
Hackman, J.R., and Lawler, E.G., III. Employee reactions to job characteristics.
Journal of Applied Psychology Monograph, 1971,55, 259-286.
Hackman, J. R., and Oldham, G. R. Motivation through the design of work: Test
of a theory. Organizational Behavior and Human Performance, 1976, 16,
250-279.
Homans, G.C. Social Behavior: Its Elementary Forms. New York: Harcourt
Brace and World, 1961.
House, R.J. A path-goal theory of leader effectiveness. Administrative Science
Quarterly, 1971,16,321-339.
Katz, D., and Kahn, R. L. The Social Psychology of Organizations. New York:
John Wiley & Sons, Inc., 1966.
March, J.G., and Simon, H.A. Organizations. New York: John Wiley & Sons,
Inc., 1959.
Merton, R.K. On Theoretical Sociology. New York: The Free Press, 1957.
Rudner, R. S. Philosophy of Social Science. Englewood Cliffs, New Jersey:
Prentice-Hall, Inc., 1966.
Schuler, R.S. Role conflict and ambiguity as a function of task-structure tech-
nology interactions. Organizational Behavior and Human Performance,
1977a,20, 67-78.
- - . Moderating effects of job involvement and growth need strength on task-
outcome relationships. Journal of Business Research, 1977b,5, 293-309.
- - . Interaction of task and structure as determinant of the effectiveness of

leader initiating structure: A test of the path-goal theory. Working Paper, The
Ohio State University, 1978.
Silverman, D. The Theory o/Organizations. New York: Basic Books, 1971.
Spinner, H. Science without reduction. Inquiry, 1973,16, 16-94.
Van de Ven, A., and Delbecq, A. A task contingent model of work unit struc-
ture. Administrative Science Quarterly, 1974,19, 183-197.
19, 357-367.
- - . Educational organizations as loosely coupled systems. Administrative Sci-
ence Quarterly, 1976,21, 1-19.
Zaleznik, A. Book Review of Katz and Kahn's Social Psychology 0/ Organiza-
tions. Administration Science Quarterly, 1967,11,690-694.
10 NOMOTHETICAL NETS
AND HIGHER ORDER FACTOR
ANALYSIS IN MIDDLE RANGE
THEORY DEVELOPMENT
Uma Sekaran, Thomas N.
Martin, Richard S. Trafton.
and Richard Osborn,
Southern Illinois University
at Carbondale
The purpose of this paper is to provide a conceptual and methodological approach

that will contribute to the development of middle range theory. Nomothetical
nets and higher order factor analysis represent these respective approaches.
Essentially, we argue that middle range theory development should proceed
as a merger of the two currently available but complementary methods for scien-
tific theory building. The first is the traditional analytic method, which calls for
reductionism, repeatability, and refutation (Checkland, 1976). The classic labo-
ratory investigation where the complexity of the real world is reduced to testable
hypotheses, which are then validated by their repeatability, is an example of this
method. Here, theory building occurs by refuting hypotheses. The second is the
systems method as developed by von Bertalanffy, Weiss, and Whiteland (Laszlo,
1975). Underlying this method is the belief that it is necessary to examine the
interrelatedness of elements and to search for the holism of the relationships
in the real-world problem.
Analytical researchers criticize systems researchers for imprecise measurements
at best, construction of untestable propositions, and a faulty understanding of
science. However, the analytical view is not without its own problems. Too often
analytical researchers seem to forget they are working within a narrow discipline
and tend to generalize concepts and partially tested models across disciplines.
127
Also, strict adherence to the analytical method may entice researchers into de-
veloping single causal models where there is reciprocal causation and premature
closure (Dubin, 1975). Recurring patterns not originally identified within the
domain of the researcher's discipline are rarely considered or discussed. An ex-
ample of this is the relationship between leader behavior and employee satisfac-
tion. Recent studies indicate that leadership may only be important in some
types of units in some types of organizations. Yet, typologies of organizations
are not normally found in the psychological literature that forms the base for
most leadership research (Kerr, 1976; Hunt, Osborn, and Schriesheim, 1978).
To encourage middle range theory development, a compromise between, and
a meeting ground of, the analytical and systems methods must be found. Specifi-
cally, as Weick (1974) denotes, this development process would be limited to
patterns of relationships that can be confirmed within limited ranges of data,
would represent consolidation of segregated hypotheses, and would be available
for consolidation with other middle range theories.
What is needed is a view that recognizes both the measurement of concepts
and patterns of relationships. The notion of a nomothetical net incorporates
this view. A nomothetical net is simply a system of relationships among con-
structs. If, however, a single pattern continues to emerge across subjects and set-
tings (organizations), then one begins to think of validated nets that can be
subjected to more rigorous analytical testing procedures. Following Marx and
Hillix (1973) and Windelband (1921), analyses leading to validated nets help
isolate general laws (nomothetical relationships). In contrast, most organiza-
tional research has been idiographic in character. That is, there has been a search
for explaining maximum variance across particular events or individuals.
If the notion of a nomothetical net is combined with higher order factor
analysis, there is a simultaneous search for both concepts and patterns of rela-
tionships across concepts. Normally in organizational analysis, factor analysis
is most often used to identify patterns of covariation across items on a ques-
tionnaire to derive concepts (Gorsuch, 1974; Neter and Wasserman, 1974).
However, a higher order factor analysis can be used to examine relationships
among patterns, particularly where the initial identification of the constructs
allows them to be intercorrelated. Rather than factoring correlations among
items, higher order factor analysis factors correlations among constructs to
delineate patterns of covariation. The patterns of covariation across concepts
appear to be the statistical analog of a nomothetical net. Only when a particular
pattern keeps recurring across subjects and organizations, does one begin to
think of a validated net.
While mapping nomothetical nets using higher order factor analysis may be
the approach, the most critical question that still remains is, "What body of
constructs should be selected for analysis?" In our opinion the work system
NOMOTHETICAL NETS AND HIGHER ORDER FACTOR ANALYSIS 129
design literature abounds with lists of dimensions, which should yield constructs
and which should lend themselves to the middle range theory development pro-
posed by Weick. Our particular selection was guided by researchers attempting
to integrate individual-psychological views with organizational-sociological views.
Specifically, we were interested in twenty-five dimensions that attempted to
tap employee perceptions of the organization, job characteristics, and selected
psychological dimensions researched by those interested in this aspect of orga-
nizational analysis. We have used the term dimension since there is some ques-
tion as to whether the dimensions listed by the researcher are constructs. While
many claim they are, there is mixed evidence for most of the measures.
Our previous discussion leads to two research questions. First, "Do employees
perceive the twenty-five dimensions in the same way as evolved by previous
researchers so that they can be labelled as concepts?" Second, "What are the
relationships among the concepts identified? Do they conform to existing
theoretical notions such that we may begin to see the emergence of a validated
nomothetical net?" If so, then the work system design theory building process
may be advanced beyond the basic analytical perspective into the realm of a
To summarize, our conceptual approach to develop a middle range perspec-
tive of the work system design literature is the nomothetical net. The analytical
tool to develop the net is higher order factor analysis. This exploratory study
should be a prelude to developing a middle range theory of work systems design.
We expect that future research with other employees in different organizational
settings would identify similar and associated nomothetical nets. If, however,
different types of nomothetical nets are mapped, these differences would be
equally important information contributing to the development of a middle
range theory of work systems design.
METHOD
Sample
A sample of 1123 white-collar clerks and first-level supervisors in twelve nation-

alized banks in three metropolitan cities in India was randomly selected. All
twelve banks operated in a munificent setting, used a mediating technology,
shared the same mechanistic formal structure, were large (6,000-10,000 em-
ployees), and operated under the same common policy umbrella set by the gov-
ernment. Work units in all twelve banks ranged in size from 6 to 20 with a mean
size of 10. Analysis of variance did not suggest significant differences across
banks or units, and thus the 1123 employees constituted the N. Questionnaires
Table 1. Dimensions Intended to Be Tapped in the Survey.
Number
of Items
in the
Dimensions Originally Question-
Intended to Be Tapped naire Origin of the Measure
Job Involvement 6 Lodahl and Kejner (1965)
Motivation to Work 5 Landy and Guion (1970)
Intrinsic Motivation to Work 3 Hackman and Oldham (1974)
Sense of Competence 5 Wagner and Morse (1975)
Satisfaction with:
Pay 2 Hackman and Oldham (1974)
Security 2 Hackman and Oldham (1974)
Social Interactions 3 Hackman and Oldham (1974)
Growth 4 Hackman and Oldham (1974)
Supervisor 3 Hackman and Oldham (1974)
Opportunities to learn, use
knowledge and skills,
advancing 6 Pelz and Andrews (1966)
Job Characteristics:
Skill Variety 3 Hackman and Oldham (1974)
Task Identity 3 Hackman and Oldham (1974)
Task Significance 3 Hackman and Oldham (1974)
Autonomy 3 Hackman and Oldham (1974)
Feedback from Job 3 Hackman and Oldham (1974)
Dealing with Others 3 Hackman and Oldham (1974)
Challenge 3 Michigan Organizational
Assessment Package (1975)
Task Interdependence 3 Michigan Organizational
General Task Characteristics Michigan Organizational
Not Included Above 7 Assessment Package (1975)
Self-Esteem from Job 4 Quinn and Shepard (1974)
Organizational Factors:
Warmth and Support 5 Litwin and Stringer (1968)
Structure 4 Litwin and Stringer (1968)
Standards 3 Litwin and Stringer (1968)
Freedom and Opportunity to Michigan Organizational
Interact 3 Assessment Package (1975)
Supervisory Behavior 7 Michigan Organizational
Training On and Off Job 3 Michigan Organizational
TOT AL NUMBER OF ITEMS 99
130
were administered to groups of 30 to 60 people in banks' conference rooms in

the fall of 1976 and were collected immediately after completion. Full details
concerning the organizations and the sample are available from the senior author.
Measures
The twenty-five dimensions (comprising 99 items) used in this study were drawn
from a comprehensive review of the organizational behavior literature and per-
tain to the domains of the organization, the job, work-related attitudes, selected
psychological dimensions and work-related satisfaction. Table 1 provides these
dimensions and their origin. All instruments were reported to have adequate
reliability and validity for exploratory research.
Data Analysis
All 99 items were first factor analyzed using Rao's canonical model with squared
multiple correlations as initial communality estimates. A scree test of the eigen-
values indicated twelve Significant factors. Various oblique factor rotations were
made to arrive at simple structure to detect the constructs empirically identified
in this sample (Gorsuch, 1974). Next, higher order factor analysis was performed
on the correlations among the twelve factors using Rao's canonical model with
squared multiple correlations as an initial communality estimate. The scree test
was used to determine the number of factors to retain. Several orthogonal and
oblique rotations were applied to approach simple structure. This was done to
trace the nomothetical net.
RESULTS AND DISCUSSION
The first research question was whether respondents saw the concepts as the
researchers did. The 99 items in the questionnaire should have loaded on 25
different factors. Instead, the scree test of the eigenvalues indicated only twelve
factors. Forty-one percent of the total variance was explained by these twelve
factors. While some of these factors correspond closely to the originally devel-
oped instruments, others seem to be apparently a conglomeration of several.
Using the most interpretable oblique rotation (Delta = -3) and incorporating all
items with factor loadings of .30 and above with no cross-loadings on the other
factors (a total of 53 items), twelve factors were found. The twelve factors and
example items and Cronbach alpha for each, are presented in Table 2.
In the initial factor anlaysis of all 99 items, Factor I (eigenvalue = 15.44)
represented "Satisfaction with Growth." All three items from the Hackman and
w
w Table 2. Example Items and Cronbach Alpha of the 12 Constructs Evolved from Factor Analysis of 99 Items.
N
--
Factor Number Cronbach
Number Construct Example Items of a
Satisfaction with Growth To what extent you are satisfied with: 7 .89
1. amount of personal growth and
development you get in doing the
job?
2. amount of challenge in the job?
II Satisfaction with Pay and To what extent you are satisfied with: 3 .79
Job Security 1. amount of job security you have?
2. amount of pay and fringe benefits
you receive?
III Job Involvement The most important things that happen 5 .70
to me involve my work.
I live, eat, and breathe my job.
IV Supervisory Behavior My supervisor is warm and friendly. 7 .75
The supervisor helps subordinates
with their personal problems
V Informality at the Work I get a lot of chances to make friends. 4 .60
Place I have the freedom to talk to people
during working hours.
VI Job Complexity My job requires the use of all the 7 .76
knowledge and skills I have.
The job requires me to use a number
of complex or high level skills.
VII Self-Esteem from the Job How do you see yourself in your work? 4 .73
Successful - - - - - - Unsuccessful
Important - - - - - - Unimportant
VIII Sense of Competence My job offers me a chance to test 3 .73
myself and my abilities.
Mastering this job meant a lot to me.
IX Satisfaction With Work- Satisfaction with: 3 .60
Related Interactions 1. people I talk to and work with.
2. the chance to get to know other
people on the job.
X Organizational Climate You get a lot of support and encourage- S
5 .70
ment for trying something new in this
organization.
Red-tape is kept to a minimum in this
organization.
XI Motivation to Work I invariably start the next assignment 3 .69
the moment the previous one is done.
I really want to work hard most of
the time.
XII Work Interdependencies To do my job well, I have to work 2 .51
closely with others.
w
w
w
--
Oldham's (1974) instrument and four items from Pelz and Andrews (1966)
denoting satisfaction with learning new skills and advancing on the job loaded on
this factor. Factor II, "Pay-Security Satisfaction," was comprised of four Hack-
man and Oldham items with two items tapping satisfaction with pay and the
other two, tapping satisfaction with job security. The two separate satisfactions
collapsed into one factor and is reminiscent of the satisfaction of "Existence
Needs." Factor III, "Job Involvement," was essentially the short version of
Lodahl and Kejner's (1965) job involvement scale. Factor IV, "Supervisory
Behavior," consisted of seven leadership items from the Michigan Organizational
Assessment Package (1975). Factor V, "Informality," was composed of two
items from Litwin and Stringer's instrument of structure and the "freedom
to talk" item from the Michigan Organizational Assessment Package. Factor VI,
"Job Complexity," did not appear to be rooted in anyone particular instru-
ment. The seven items that loaded on this factor dealt with job descriptions, and
the loadings suggested that simple versus complex were the two ends of this con-
tinuum. Factor VII, "Self-Esteem," was the self-esteem index from the Michigan
Organizational Assessment Package. Factor VIII, "Sense of Competence," had
three of the five items used as a measure from the Wagner and Morse (1975)
instrument and the "personal satisfaction from job" item from the Hackman and
Oldham Intrinsic Motivation to Work item. Factor IX, "Satisfaction with Work
Interactions," contained Hackman and Oldham's "Satisfaction with Social
Interactions" items. Factor X, "Organizational Climate," consisted of five items
from the Litwin and Stringer measures, while Factor XI, "Motivation to Work,"
was the Landy and Guion (1970) motivation measure. The last factor, Factor
XII, "Work Interdependence," contained two items from Hackman and Old-
ham's "Dealing with Others" measure.
Overall, it appears that respondents identified some of the dimensions as
delineated by researchers. Yet, they also saw configurations of items that are
hard to describe. These results may reflect poor instrumentation. However, they
also may reflect the perceptions of employees as they see themselves, their orga-
nizations, and their jobs.
The second research question concerned the presence of a nomothetical net.
Using the higher order factor analysis, the twelve concepts were collapsed into
four higher order factors as determined by the scree test. Fifty-two percent of
the common variance was explained by these four factors. The eigenvalues and
percentage of variance explained are presented in Table 3. The factor loadings on
the four higher order factors are in Table 4.
There are two important components of the outcome of the higher order fac-
tor analysis. One, the interpretation of the higher order factors themselves, and
two, the examination of the correlation among the higher order factors.
Table 3. Eigenvalues of the Twelve Factors Generated by the Higher Order

Factor Analysis.
Percentage Cumula-
Eigen- of tive
Factor value Variance Percentage
I = Satisfaction-Growth 2.5 20.5 20.5

II = Satisfaction-Pay and Job Security 1.4 12.1 32.6
III = Job Involvement 1.2 10.1 42.7
IV = Supervisory Behavior 1.1 9.2 51.9
V = Informality at the Work Place 0.9 7.4 59.3
VI = Job Complexity 0.9 7.3 66.6
VII = Self-Esteem 0.8 6.9 73.4
VIII = Sense of Competence 0.7 6.0 79.4
IX = Satisfaction-Work Interactions 0.7 5.6 85.0
X = Organizational Climate 0.6 5.4 90.3
XI = Motivation to Work 0.6 4.9 95.3
XII = Work Interdependence 0.6 4.7 100.0
Table 4. Factor Loadings on the Four Higher Order Factors.

Higher Higher Higher Higher
Order Order Order Order
Factor Factor 1 Factor 2 Factor 3 Factor 4
1 Satisfaction-Growth -0.15 -0.25 -0.04 -0.35

2 Pay and Security Satis-
faction 0.05 -0.30 0.08 0.23
3 Job Involvement -0.15 -0.04 0.30 -0.14
4 Supervisory Behavior -0.02 0,01 0.02 0.37
5 Informality 1.01 -0.05 -0.03 -0.14
6 Job Complexity 0.00 0.91 -0.02 0.22
7 Self-Esteem 0.06 0.02 -0.08 0.61
8 Sense of Competence -0.04 0.06 -0.02 0.35
9 Sa tisfaction -Work
Interactions 0.04 0.02 0.03 -0.58
10 Organizational Climate 0.16 -0.09 0.05 0.41
11 Motivation 0.12 0.03 0.92 0.08
12 Work Interdependencies -0.12 -0.18 -0.08 -0.13
Interpretation of the Higher Order Factors
The first higher order factor consisted primarily of the original Factor V and was
thus labelled "Informality" -denoting a "loose" structure. The second higher
order factor consisted of the Job Complexity factor (originally Factor VI) and
the Pay and Security Satisfaction factor (originally Factor II) and was thus
labelled "Job." The emergence of this higher order factor suggests a close asso-
ciation between the characteristics of the job and the satisfaction that goes along
with job pay and job security. These two satisfactions seem to have their roots in
the job itself. The loadings on the third higher order factor indicate the inclusion
of the original Factors III and XI, representing the job involvement and motiva-
tion to work factors, respectively, and was thus labelled "Work Commitment."
The fourth higher order factor includes the original Factors I, IV, VII, VIII, IX,
and X, denoting growth satisfaction, supervisory behavior, self esteem, sense of
competence, satisfaction with work interactions, and organizational climate.
This complex factor appears indicative of the attributions regarding: (1) self as
the actor in the organization, (2) the superior as the actor, (3) the supportiveness
of the organization, and (4) the work interactions. Thus, this factor was labelled
"Organizational Support."
The last of the original twelve factors -work interdependency -did not get
caught in the nomothetical net. Possible explanations might include poor mea-
surement or its being a part of another network.
Overall, the factor structure of the higher order factor analysis suggested that
some of the original factors tend to get integrated into consistent wholes or
gestalts. These gestalts seem to represent partially consistent groupings of per-
ceptions that appear to have a common base. This is also evident from the nature
of the clusterings of variables in the higher order factors.
Correlations among the Higher Order Factors
A most interesting aspect of the results of the higher order factor analysis is the
relationship among the factors. The nomothetical net as captured by the higher
order factor analysis and the correlation among the different pockets in the net
is depicted in Figure 1.
Informality, Job, and the Work Commitment factors are very weakly corre-
lated with each other. Only the Organizational Support factor has a modest
correlation with the Informality factor. Each of the higher order factors appears
to be differentiated from the others by employees. This is evidenced by the
relatively low correlations among the factors. The employees appear to integrate
the pieces, or groups of pieces, even though all the pieces do not fit together into
one gestalt.
FIGURE 1
THE PERCEPTUAL NOMOTHETICAL NET
INFORMALITY
Informality
(1.01 )
ORGANIZATIONAL SUPPORT
Satisfaction-Growth (-.35)
2 Sel f Esteem ( .61)
JOB Sense of Competence (.35)
Supervisory Behavior ( .37)
Job Complexity (.91) .13 Satisfaction with Work
Satisfaction-Pay and Interactions (.58)
Security (-.30) Organizational Climate (.41 )
.01
WORK
COMMITMENT
Job Involvement (.30)
Motivation to Work (.92)
Figures in parentheses indicate higher order factor loadings, while the correlations among the
"constructs" are indicated on the connecting lines.
Generalized job attitudes as represented in Work Commitment seem to reflect

an overall orientation to work. This factor is posited for future research to be
carried as an individual characteristic by the employee from job to job, rather
than being determined by factors in the work situation, and is reminiscent of the
protestant ethic.
As previously mentioned, informality has the highest correlation with orga-
nizational support. This suggests that informality and social interaction at the
work place may be only modestly associated with one's self-esteem, sense of
competence, and the like.
Job complexity was weakly correlated with the other factors. It is important
to remember, however, that for this factor job difficulty (complexity) combined
with pay and security satisfaction.
From a systems perspective, analysis of the pattern of correlational associa-
tions among the four higher order factors should reveal higher interrelationships.
Since this is not the case, the system's theorists might recommend that a holistic
approach has not been followed in the work systems design literature. Addi-
tional nets are entirely feasible if a larger range of variables is included. Perhaps
the work system design researcher might incorporate the potential importance of
nonpsychological factors such as physical location, state of the economy, and
the type of product produced by the organization as part of their future research.
While some might argue that all that has been done is a Fishbein type of simu-
lation of numerous organizational variables, the mapping of nomothetical nets
does facilitate integration of knowledge over and above piecemeal analytical and
predictive studies and is faster and more reliable. Mapping the perceptual do-
mains of respondents during pilot study phases would seem to be a useful strat-
egy for refming theoretical models before launching into full-fledged research
endeavors. We suspect that several nomothetical nets could be mapped into an
organization's work system. Thus, researchers might expand their research into
numerous nomothetical nets encompassing limited but interdisciplinary ranges
of data and hypotheses. This is a step beyond the fragmented, disciplinary
boundaries of the pure analytical perspective and yet, not encompassing the
holistic universe of the systems perspective. Ukewise, the higher order factor
analysis represents a useful tool of analysis for middle range theory develop-
ment. This tool begins to help the investigator to understand how constructs are
integrated to form nomothetical nets.
This paper has departed substantially from current studies by examining con-
structs in terms of the nomothetical net of the respondent instead of analyzing
data to fit a priori theoretical models of the investigator. This study suggests that
the search to identify nomothetical nets may be a fruitful middle range field of
endeavor in our attempts at understanding what goes on in organizations. Add-
ing more constructs and mapping the nets in different organizations may be
useful in explaining higher levels of variance and also in understanding the inves-
tigated systems. Perhaps this is what Merton (1957) and Weick (1974) had in
mind when they called forth the category of theories of the middle range!
REFERENCES
Checkland, P.B. Science and the systems paradigm. International Journal of

General Systems, 1976,3, 127-134.
Dubin, R. Causality and social systems analysis. International Journal of General
Systems, 1975,2, 107-113.
Gorsuch, R.L. Factor Analysis. Philadelphia: W.B. Saunders, 1974.

Hackman, J.R., and Oldham, G.R. The job diagnostic survey: An instrument for
the diagnosis of jobs and the evaluation of job redesign projects. Yale Univer-
sity, Department of Administrative Sciences, Tech. Rep. 4.
Hunt, J.G.; Osborn, R.N.; and Schriesheim, C.A. Neglected aspects of leadership
research. Proceedings of the 21st Annual Conference, Midwest Academy of
Management, 1978,364-375.
Kerr, S. Substitutes for leadership: Their meaning and measurement. In H.E.
Schneider (Ed.), American Institute of Decision Sciences Proceedings, 1976.
Landy, F.J., and Guion, R.M. Development of scales for the measurement and
of work motivation. Organizational Behavior and Human Performance, 1970,
5,93-103.
Laszlo, E. The meaning and significance of general system theory. Behavioral
Science, 1975,20, 9-24.
Litwin, G.H., and Stringer, R.A. Motivation and Organizational Climate. Cam-
bridge: Harvard University Press, 1968.
Lodahl, T.M., and Kejner, M. The definition and measurement of job involve-
ment. Journal of Applied Psychology, 1965,49, 24-33.
Marx, M.H., and Hillix, W.A. Systems and Theories in Psychology (II edition).
McGraw-Hill, 1973.
Merton, R.K. Social Theory and Social Structure. Glencoe, Ill.: Free Press,
1957.
Michigan Organizational Assessment Package: Progress Report Il Institute for
social research, Michigan, 1975.
Neter, J., and Wasserman, W. Applied Linear Statistical Models. Homewood:
Irwin, 1974.
Pelz, D., and Andrews, F. Scientists in Organization: Production Climate for
R&D. New York: Wiley, 1966.
Quinn, R.P., and Shepard, L.J. The Quality of Employment Survey. Survey Re-
search Center, Michigan, 1974.
Wagner, F.R., and Morse, J.J. A measure of individual sense of competence.
Psychological Reports, 1975,36, 451-459.
19,357-367.
Windelband, W. An Introduction to Philosophy (Translated by J.C. McCabe).
London: T. Fisher, Unwin, 1921.
11 ON THE POTENTIALITIES
OF MIDDLE RANGE THEORY
J. Kenneth Benson,
University of Missouri-Columbia
All of the papers under consideration 1 advocate middle range theory in some
form. I will attempt first to explicate the conceptions of middle range theory
that are developed in the papers. This will involve an exploration of the am-
biguities and differences of view in the several statements. Then, I will tum
to an assessment of middle range theory and its possibilities.
THEORIES OF THE MIDDLE RANGE
The advocacy of middle range theory, in sociology and in organization theory,

is born of frustration with general theory. Merton's original proposal (1949)
was a reaction against the position of Parsons, who through. the 19408 was a
major advocate of general theory in sociology and who through the 1950s was
a major formulator of general theory. Merton's phrase "theories of the middle
range" became a rallying cry for abstract empiricists who were committed more
to research methods than to theory. It is worth noting, however, that Merton
was advocating theory of the middle range, not merely opposing general theory.
Further, what Merton took to be the middle range was a good deal broader than
the conceptions put forward in the present papers. As Bobko points out, there
140
ON THE POTENTIALITIES OF MIDDLE RANGE THEORY 141
is considerable ambiguity about what constitutes the middle range. Clearly,

in Merton's work the term referred to the whole of substantive fields such as
political sociology, deviant behavior, and the sociology of science. A compre-
hensive theory of organizations would be a middle range theory for Merton.
General theory, in his formulation, referred to theories of whole societies or of
all social behavior.
The present advocates of middle range theory share with Merton a dissatis-
faction with general theory. The object of this dissatisfaction appears primarily
to be systems theory in the form of open systems, and functional theory. There
is also some dissatisfaction with rational model theorizing, which assumes the
organization can be understood as a rationally constructed tool. Instead, orga-
nizations are seen as "loosely coupled" systems, following Weick (1976), which
have semiautonomous components linked together in partly nonrational ways.
Given this nonrationalloose coupling, it becomes feasible and perhaps necessary
to develop strands of theory on specific components of organizational behavior
and to leave the manner of their coupling as an open question to be resolved
later. While both Bobko and Pinder and Moore appear to distance themselves
from this view, the other authors seem to work well within it; and even Bobko
and Pinder and Moore appear to share it to a degree.
TASKS IN THE CONSTRUCTION OF MIDDLE RANGE THEORY
I will examine next some tasks that would be important in a program of middle
range theory building, as seen in the papers at hand. Both Schuler and the
Sekaran group are constructing middle range theory. Both papers take middle
range theory to be something more than empirical generalizations, and both face
the problem of how to put pieces of theory together. They confront two closely
connected tasks.
First, there is the task of achieving conceptual comparability between a series
of distinct inquiries. This problem arises because much organization theorizing
is in the form of narrow-gauged empirical generalizations. To construct middle
range theory is to build up from such empirical generalizations to more general
statements. This requires finding a common conceptual framework. (Note that
Schuler uses the term "trans-middle range theory" to refer to such integrative
formulations.)
Second, empirical division points must be found that separate clusters of
variables closely connected to each other and that are differentiated from other
clusters. Theories of the middle range presumably would isolate those units of
organizational behavior which hang together empirically and which have a degree
of autonomy from other units. If such units cannot be identified, then the
claims for middle range theory quickly evaporate.
The authors use quite different strategies for dealing with these tasks. Schuler
uses a strategy of theoretical reduction, that is, rmding sets of theoretical argu-
ments and reducing two or more sets to the terms of a third. Such a reduction
involves abstracting general concepts from specific concepts and constructing
an encompassing framework of propositions based upon and accounting for the
several sets of empirical generalizations. Specific concepts and generalizations
are converted to instances of a more general theoretical scheme. Thus, for
Schuler the first step in building middle range theory is conceptual rather
than empirical.
Sekaran et al., by contrast, rely upon a strategy of empirical observation and
testing. They use factor analysis of indicators used in previous research on re-
lated problems to search for "nomothetical nets." Their strategy is basically
empirical. The ultimate answer is to see how the several measures correlate and
then to correct conceptualizations. They even assume that a scientific mapping
of the social world should correspond closely to the cognitive maps of organiza-
tional participants.
A serious problem with the Sekaran paper is that it rests upon the assumption
of the existence of an unambiguous empirical reference point, which can even-
tually make sense of our conceptual and explanatory efforts and resolve differ-
ences between the proponents of different constructs, theories, and paradigms.
The paper exemplifies a form of what Pinder and Moore call the realist position.
The exploration of the problem by Pinder and Moore and by Bobko should be
sufficient to show how problematic the position is.
In addition, the Sekaran paper involves an important conceptual step prior
to the factor analysis, which is not sufficiently articulated. The authors choose
a population of indicators on the basis of the assumption that comparability
should exist across certain categories of research. This involves establishing a
claim of comparability across different groups or scientific communities. The
authors offer us no guidelines for including some such groups and excluding
others. The usefulness of their proposed approach depends upon explicating
this step. This is a point that Bobko also makes about factor analysis; and deal-
ing with the problem would open a whole range of issues regarding scientific
paradigms, which Bobko and Pinder and Moore take up but which the Sekaran
group leave implicit in their paper.
Schuler does not escape this difficulty either. He encounters it at the point of
selecting clusters for inclusion in a theoretical reduction. That is, both the
Sekaran and Schuler papers encounter a basic difficulty in dividing the social
world into sets of variables or clusters that form the starting point for analysis.
In both papers the clusters are formed in a very conventional way. That is, they
rely upon conventionally defined clusters of variables corresponding to estab-
lished areas of research such as leadership or job design. What are these clusters?
How are they bounded? Are the boundaries social? Do they represent theoretical
traditions? Do they represent common tasks of organization theorizing, practi-
cal, or applied tasks of administrations? One would need to ask these questions
very early in the game; but neither paper pays sufficient attention to them.
The problem of dissensus on such issues is the focus of the papers by Pinder
and Moore and by Bobko.
FOUNDATION PROBLEMS OF MIDDLE-RANGE THEORY
Pinder and Moore and Bobko address the epistemological underpinnings of

middle range theory and draw implications for the ongoing practice of theory
construction and for scholarly communication.
Pinder and Moore argue that mUltiple, incompatible paradigms are inevitable
in organization theory. They anticipate continued dissensus on foundational
issues growing out of fundamentally different images of organizational phe-
nomena, opposing exemplars, and the like. This conclusion is reached both on
realist grounds and on subjective idealist grounds. They put forward these two
arguments as mutually supportive a~ if jointly they lead to a firmer conclusion
than either one produces separately. They seem more committed, however, to
the realist position (i.e., the argument that paradigm diversity is a consequence
of the variety and changeability of real organizations).
Bobko, by contrast, is clearly committed to a version of the subjective idealist
position. He too argues that paradigm diversity is to be expected indefmitely. He
makes this argument on the basis that organization scientists inevitably look at
organizations from a particular perspective. Different "images" of organizations
produce incompatible theories. Contrary to Pinder and Moore, Bobko suggests
that the problem is more complex than organizational variation (the realist
argument) and thus that it cannot be solved by taxonomic analysis-producing
homogeneous sets of organizations, each of which might be the appropriate
object of a middle range theory. Rather, the multiplicity of images is a result
of opposing "basic assumptions about the nature of man and society."
Bobko is much less optimistic than the other authors that middle range
theories might "add up" to something more comprehensive or even that a num-
ber of separate inquiries might be combined to form strands of middle range
theory. Work may be additive within a paradigm, that is, within a set of inquiries
sharing a stance toward the nature of man, society, and knowledge. It is not
additive across such paradigmatic differences. Thus, Bobko advocates middle
range theory for different reasons from those shared by Pinder and Moore,
Schuler, and the Sekaran group.
There are some similarities in the practical implications developed by Pinder
and Moore and by Bobko. Both papers favor diversity of approach; both en-
courage middle range theory; both advocate tolerance for differences; and both
discourage the formulation of general theory. There is even a tendency, espe-
cially in Pinder and Moore, to condemn cross-paradigm debate and to declare a
moratorium on fundamental questions in the interest of encouraging a variety
of middle range developments.
There are basic differences between Bobko and the other authors on what
should come from the pursuit of middle range theory. Pinder and Moore, for
example, envision a gradual building up of linkages between separately devel-
oped strands of theory. Eventually it may be possible to construct general the-
ory by bringing together enough separate strands of middle range theory. Pinder
and Moore even suggest, in their explication of the realist argument, that we may
be able to construct a theory of organizational variation that would tell us why
organizational realities change and vary and thus by implication tell us why we
have theoretical diversity. For this purpose they resort to a brief statement of
population-ecology theory (Le., that diversity and change result from coping
with different and changing environments). Pinder and Moore seem to envision
putting the field together again in the sense of constructing a series of empiri-
cally verifiable statements about organizations that would be widely affirmed
and that would mediate the remaining differences.
For Bobko, by contrast, the implications are largely contemplative. He is con-
cerned not with the state of the field but with a state of mind. He would have
us learn from several exemplars-Steinbeck, factor analysis, Zen-to value diver-
sity, to adjust to it with humor, to be creative and reflexive, to contemplate with
love our multilectic, anarchistic field. This state of mind would have an uncer-
tain bearing on the state of the field. There is no indication of what a loving
acceptance of diversity would contribute to our capacity to construct an inte-
grated set of statements about organizations that could stand the scrutiny of
other scholars. In fact, it might lead us to accept diversity where it should be
challenged with techniques such as theoretical reduction (Schuler'S strategy) or
factor analysis (as in Sekaran).
Bobko's argument should not be dismissed lightly. In fact, it is a very en-
couraging sign. It might contribute to the formulation of a theory of knowledge
production specific to the organizations field. Such a theory would account for
the shape of the field by reference to the interests, relevance structures, social
positions, and so on, of theory groups. The result might be a sociology of orga-
nization science that could make empirically verifiable statements about the
production of organization theories. This would take us beyond the contempla-
tion advocated by Bobko and beyond the mere recognition of the importance
of standpoints, which is the endpoint of Pinder and Moore's explication of the
idealist position. It might also be linked to a theory of the production of
organizational realities, giving us an understanding of how the production of

organizations and the production of organization theories are interrelated.
I see two quite different things in the Pinder and Moore paper. One is a
theory of organizational diversity; the other is a theory of the diversity of orga-
nizational theories. There should ultimately be some connection between those
theories; but Pinder and Moore do not offer a connection. Bobko's paper at least
suggests that we might attempt to formulate the connections by examining the
images, relevance structures, and even infrastructures that underlie organization
theories. Following this path, we might begin to construct theories of how the
production of organizational differences is related to the production of varia-
tions in organizational theories. Such theories would tie into the clusters and
nomothetical nets that are the concerns of the Schuler and the Sekaran papers.
More specifically, the nets or the clusters may be seen as outcroppings of par-
ticular research programs often based on applied concerns (e.g., improving
leadership or task design). Such programs may have linkages to the interests of
administrative elites having fairly clear ties to the larger institutional arrange-
ments and dominance structures of the society. We could then look at clusters
of empirical work and nomothetical nets not merely as sets of variables having
an independent existence but as the expressions of underlying paradigms or
forms of inquiry conducted by particular groups who have ties to the interest
and power structures of the society. Observing the social practices of these
groups we may find similarities that transcend their differences of theoretical
terms and variables. Pursuing this line of inquiry does not imply a rejection of
science but a new understanding of science (see Bernstein, 1977).
CRITIQUE OF MIDDLE RANGE THEORY
Next I will outline some reservations about the proposals for theories of the
middle range in the papers at hand. To begin, however, I must record that I find
the interest in middle range theory a welcome development since it represents
a departure from the dominance of rational model and systems theorizing in the
organizations field. This said, I tum to a few critical remarks.
First, all four of the papers overestimate the diversity in the field. This over-
estimate occurs because of superficial readings of the field and the failure to
look for its deep structures. The problem is clearest in Pinder and Moore's
slippery usage of the concept of paradigm. If we look at surface differences in
the names of concepts and theories, the field appears to be enormously diverse.
Such differences, however, probably should not be termed paradigmatic. To
identify a paradigm requires looking for underlying assumptions, commitments,
or images that run through a series of works without necessarily being stated.
Georgiou's fonnulation (1973) of the "goal paradigm" is a case of appropriate

usage; it isolates a reasoning structure characteristic of the field as a whole,
which is so pervasive that it is taken for granted.
The problems of theoretical diversity that concern Schuler and the Sekaran
group are also on a relatively superficial level. Their problem is to effect a mer-
ger of concepts, propositions, and measurements between specific theories or
research traditions. From a larger perspective concerned with more basic affm-
ities, these efforts may be seen as minor consolidation and communication
problems among theory groups sharing a common basic approach. Bobko recog-
nizes the possibility that multiple theories may cohabit within a paradigm. How-
ever, he then succumbs to an extreme subjectivism that also results in a great
overestimate of diversity. One of the striking failures through the papers is the
omission or underdevelopment of concern with institutional forces and interest
structures that shape the organizations field. There is actually a large degree of
coherence to the field, I would argue, because of the continuity and integration
of the interest structures shaping it. Attention to these structural ties could help
to counter Bobko's flight into subjectivism.
We should also be interested in concrete practices that characterize the field
as a whole and that unite scholars of avowedly different imageries and models.
Attention to the link between observable practices and structured interests
would move us toward a theory of the social production of organization theories,
while countering the sense of unlimited variability and changeability of organiza-
tion theorizing.
Second, there is a tendency to legitimate the status quo in organization
theory. The field is nominally divisible into multiple theory groups each pur-
suing a small cluster of empirical issues such as leadership, job design, technology,
and interorganizational relations. The call for middle range theory could become
a legitimation of these specialized research problems as the best we can do now
or in the foreseeable future. This may forestall efforts to look at the macro-
structure issues involving organizations. This would parallel the history of
Merton's advocacy of middle range theory in sociology, where the phrase became
a slogan justifying the continuation of narrow research programs distinguished
more by methodological rigor than by theoretical grasp. In fact, Merton's
fonnula contributed to a form of repressive tolerance in which issues of small
scope were admitted to debate while larger questions concerning the fundamen-
tal structure of societies were ruled out of order.
NOTE
1. The papers included are those by Bobko; Pinder and Moore; Sekaran, Martin, Trafton,
and Osborn; and Schuler.
REFERENCES
Bernstein, Richard J. The Restructuring of Social and Political Thought. New

York: Harcourt, Brace and Jovanovich, 1977.
Georgiou, Petro. The goal paradigm and notes toward a counter paradigm. Ad-
ministrative Science Quarterly, 1973,18,291-310.
Merton, Robert K. Social Theory and Social Structure. New York: The Free
Press, 1949.
Weick, Karl E. Educational organizations as loosely coupled systems. Adminis-
trative Science Quarterly, 1976,21, 1-19.
12REPLY TO BENSON'S
COMMENTS IN
"ON THE POTENTIALITIES
OF MIDDLE RANGE THEORY"
Randall S. Schuler,
Benson raised several crucial issues regarding the application of middle range
theory in organizational behavior and theory, which address statements I made
in my paper "Middle Range Theories: Clusters of Clusters of Organizational
Phenomenon." I wish to comment on the points he raised and further clarify my
position regarding the conceptualization and operationalization of middle range
theory in the study of organization.
I think that Merton provided a useful thrust to explore and explain organiza-
tional phenomena with his popularization of middle range theory. Benson points
out, however, that for Merton middle range theory was used to describe "the
whole of substantive fields like political sociology .... " For Merton the field of
organization would be described by a middle range theory. Thus, the reference
to Merton in the application of middle range theory is most appropriate only at
the conceptual level. His intent was not to use middle range theory within sub-
stantive fields; however, ours is. Therefore, we leave Merton at the gate of the
field of organization. What we do inside is up to us.
Benson points out that two important issues in the use of middle range
theory in organization are differentiation and integration. That is, how can we
divide the field of organization into parts and then how can the parts be com-
bined? Benson suggests that there are many ways each of these issues can be
148
REPLY TO BENSON'S COMMENTS 149
addressed. Benson indicates that "the puzzle (field) is extemely complex because
of the large number of possible clusters (ways to divide the field)," and thus the
prospect is for many middle range theories of minimal support. I agree with
Benson and suggest that there be competing middle range theories for the same
clusters. The identification of these clusters in the field of organization (i.e., the
differentiation) should be done both conceptually and empirically. My illustra-
tion of where we could start used three clusters: task design, leadership and
roles. My purpose in using these three was not to maintain the status quo in the
investigation of organization but to provide an illustration of middle range
theory development. I appreciate Benson's concern for divergent approaches to
organization understanding and suggest that middle range theory is an excellent
vehicle to accommodate both divergent and more conventional approaches.
The issue of integration of clusters or parts of the field of organization is
important. It is not until this integration occurs that I recognize middle range
theory use and development. Each cluster or part before integration is best
described by narrow range theory. Thus, the focus of middle range theory is
delimited at one end. At the other end, when all the clusters or parts in the field
are encompassed in one theory it is called grand theory. Between the extremes,
lie middle range theories. There are many middle range theories. Further-
more, when I use the term middle range theory, I refer to the integration of
only two clusters and when there is an integration of more than two clusters
I use the term trans-middle range theory. Again, there are many trans-middle
range theories.
I see the use of middle range and trans-middle range theories as spurs to the
development of our understanding of organization. I see lively debates and com-
petition among researchers for their middle range theories or trans-middle range
theories. The winners will be difficult, if not impossible, to identify. Indeed,
winning implies criteria to judge. What criteria can we use to judge the best
middle range theory or best trans-middle range theory? First of all, the competi-
tion needs to be based on the entries theorizing about the same clusters-or does
it? Who's to say that a middle range theory using task design and employee
growth-need strength is better than a middle range theory using roles and em-
ployee growth-need strength? How could these two theories be compared?
The competition in the field of organization should be directed toward seeing
which researcher can incorporate the most clusters in a single theoretical scheme
most efficiently. The concern for which issues or clusters are covered will be in
the hands of the researchers, not determined by the use of the middle range
theory. Thus, "micro" and "macro" issues may be addressed in the use of
middle range theory. Which issues are addressed more frequently will, in part,
reflect the concerns of the researchers. The concerns of the researchers, in turn,
may reflect their ability to empirically test their middle range or trans-middle
range theories.
A final concern here for middle range theory and trans-middle range theory
is what are their (our) purposes? Are the purposes to reconcile apparent diver-
sity, to increase predictability, to improve practicality and applicability, to
improve conditions for employees in organizations, or to identify and accept real
diversity? I would suggest all of these are legitimate purposes. I think that real
divergencies in the field must be identified, but this requires more than the use
of middle range theorizing to approach the study of organization. It perhaps
takes researchers such as Benson, with divergent ideas and conceptualizations
using middle range theory, to identify these real divergencies.
This identification of purposes also suggests the great number of middle range
and trans-middle range theories that are possible. For example, if we want to
increase predictability, several theories can be constructed; each would have dif-
ferent criteria for which they are trying to increase predictability (e.g., employee
satisfaction, performance, group cohesion, etc.).
Thus, the suggested use of middle range theory for the study of organization
incorporates recognition of numerous clusters of phenomena within the field of
organization, the numerous ways of incorporating those clusters into larger
groupings, the real divergencies that may exist among the clusters, and the vital-
ity from the competition among researchers in conceptualizing and testing their
middle range or trans-middle range theories.
A final caveat. This push for the use of middle range theory by no means
implies that we should ignore grand range theorizing nor the development of
narrow range theories. Grand range theorizing may provide vision or may con-
ceptualize a guide for assemblying middle range theories, and narrow range
theories are the important stepping stones for the construction of middle
range theories.
13 ZEN AND SCIENCE
A Reply to Benson
Philip Bobko,
University of Maryland
For the most part, I think Professor Benson has captured the spirit of the four
articles that he has reviewed. In particular, I believe he responsibly articulates
my multilectic approach to social scientific knowing. However, I would like to
indicate a few subtle misinterpretations-due, I believe, to Benson's unwilling-
ness to accept the Zen paradigm in toto. Secondly, I would like to expand on
the possibility of an integration/coupling of our organizational research en-
deavors. Finally, I reiterate my conclusions and heartily support Benson's con-
clusions regarding the epistemological nature of social science.
BIPOLARITY, EVALUATION, AND ZEN
In the Zen paradigm, there is no tension between opposing images. The clear
implication is a negation of the necessity for dualistic notions and singular
dimensions with bipolar opposites as endpoints. Unfortunately, Benson's analy-
sis of my paper maps his Western, dialectical (Benson, 1977) schema onto a
structure that has a circular geometric metaphor. For example, Benson mis-
takenly classifies my position as subjective idealist "in contrast" to the realist
position of Pinder and Moore. However, the multilectic view of my paper
151
accepts both positions as possibly fruitful starting points for inquiry, that is,
as points on a circle and not as opposing endpoints in a thesis-antithesis para-
digm. As a second example, consider Benson's statement that I am "concerned
not with the state of the field, but with a state of mind." Again, field and mind
are not opposing, or even separate, entities. It is the processual interplay of field
and mind which creates the organizational totality of interest. (Otherwise, one
could envision a "mindless state of the field" -much to the delight of critics of
organizational theorizing!)
Additionally, negation of the necessity of bipolarity implies a negation of the
necessity for evaluation. Thus, I do not "advocate middle range theory" (as
Benson suggests). Rather, I simply note the existence of multiple imagery in
organizational theorizing and the importance of that awareness for creative,
investigatory social science. Similarly, I am not dissatisfied with open systems
theory (as Benson suggests). I just note that open systems theory may be
considered as a type of middle-range theory and, as such, is subject to the con-
siderations and implications of my multilectic approach.
INTEGRATION AND DIVERSITY
Benson suggests that my position "results in a great overestimate of diversity."

The problem with this statement is a failure to distinguish between "what is"
and "what could be." The fact that the diversity of organizations in the status
quo may be bounded by "institutional forces and interest structures" leads
precisely to the need for an expanded set of viewpoints about what could be. I
do not believe my paper overestimates the possibilities.
An important question becomes, "How can these diverse viewpoints be inte-
grated?" As Benson notes, the solution to this question is not directly attended
to by any of the four papers. However, one implication of my position is clear:
If we focus on the basic assumptions (stipulations) of each viewpoint, we might
at least know on which dimensions it even makes sense to ask the above ques-
tion. Perhaps there truly are viewpoints that can be reduced through empirical
techniques such as factor analysisl. However, this can be known only through
an investigation of the underlying stipulations of our imagery.
SCIENCE
Benson suggests that one implication of my position is the need for a sociology
of organizational theorizing that would go "beyond the contemplation advo-
cated by Bobko." I heartily agree with the need for such a sociology and, in
ZEN AND SCIENCE 153
fact, indicated the further need for both a sociology and psychology of organiza-
tional theorizing in my section on artifacts (thus disagreeing with Benson's use
of the term contemplation).
Finally, it would seem crucial to accept Benson's interpretation of Bernstein
(1977). That is, a multilectic inquiry "does not imply a rejection of science but
a new understanding of science." Etymologically, the word science is derived
from the Latin word scire, meaning "to know." Thus, science is the totality of
knowing, and not just one aspect of knowledge bounded, for example, by
quantitative or positivistic thought. In this original defmition, science is not
wedded to the choice of an endpoint on a particular epistemological dimension.
Rather, it is the union of all choices on the Zen circle (or sphere, or hyper-
sphere). It is this union and enhancement of our knowledges through multiple
imagery that provides a new understanding of science.
NOTE
1. However, techniques such as factor analysis may give a false sense of precision. For
example, assume the Sekaran et al. measurement paradigm to be true. In their study, an
initial factor analysis captured 41% of the variance. Their higher order factor analysis cap-
tured 52% of the 41%-for a total of .41 X .52 =.21, or 21% of the variance in the original
sample. Also, the sample size in their study was N = 1123. A conservative estimate of the
number of parameters estimated in their study is p = 99 + 12 = III (i.e., 99 initial variables
and 12 factors in the higher order factor analysis). Using the Stein formula (see Darlington,
1968) to estimate shrinkage of the explained variance in a new sample with R2 = .21,
N = 1123, and p = 111 gives a shrunken R2 of less than .02. Thus, the Sekaran et al. factor
structure would probably explain less than 2% of the total system variance in a cross-
validation sample -hardly a figure of practical significance!
REFERENCES
Benson, J.K. Organizations: A dialectical view. American Science Quarterly,

1977,22, 1-21.
- - . On the potentialities of middle range theory: Commentary on Bobko,
Pinder and Moore, Sekaran et aI., and Schuler. In C. Pinder and L. Moore
(Eds.), Middle Range Theory and the Study of Organizations. Martinus
Nijhoff, 1979.
Bernstein, R.J. The Restructuring of Social and Political Thought. New York:
Harcourt, Brace & Jovanovich, 1977.
Darlington, R. B. Multiple regression in psychological research and practice.
Psychological Bulletin, 1968,69, 161-182.
Pinder, C., and Moore, L. The inevitability of multiple paradigms and the resul-
tant need for middle range analysis in organization theory. In C. Pinder and
L. Moore (Eds.), Middle Range Theory and the Study of Organizations. Mar-
tinus Nijhoff, 1979.
Sekaran, U.; Martin, T.; Trafton; R., and Osborn, R. Nomothetical nets and
higher order factor analysis in middle range theory development. In C. Pinder
and L. Moore (Eds.), Middle Range Theory and the Study of Organizations.
Martinus Nijhoff, 1979.
IIIGENERAL
FRAMEWORKS
FOR MIDDLE RANGE
THEORIZING
14 EVOLUTION AND MIDDLE
RANGE THEORIES
Toward a Matrix
of Organizational Modes
John D. Bigelow, Oregon State University
To date, organizational theorizing and behavioral science practice have developed

largely as two distinct disciplines, with more communication within diSciplines
than between. This is a problem that has been recognized by a number of theo·
rists and practitioners (e.g., Friedlander and Brown, 1974; Kahn, 1974; Walter·
Busch, 1975; Burke, 1976), but none have proposed specific remedies by which
the gap between the two disciplines could be bridged.
Recently, an approach has been put forth that addresses this problem. Pinder
and Moore (1978) state that there is an overemphasis on general theories of orga·
nizations which are overly distanced from organizational realities, and conse·
quently difficult to apply. Their response is to propose moving to a "multiple
paradigm" approach to organizational theorizing, in which a set of middle-range
theories is developed in relation to a defined set of specific types or aspects
of organizations.
Pinder and Moore's approach appears to have much merit as a pragmatic solu-
tion to the problem of developing applicable theories of organizations. However,
a preliminary question is raised: To what extent do organizations actually fall
into classifiable categories? In considering classificatory guidelines, McKelvey
(1975) points out:
To maintain as much homogeneity within a class as possible, one would like
157
to be able to organize the classes around one or more dominant modes along
a given attribute dimension, with each class encompassing the modal organiza-
tions and a few on either side of the mode (p. 521).
In developing organizational classifications, to what extent can we rely on the
existence of such modes? An alternative possibility is that organizations exist in
every configuration, and a continuum of intermediate types can be identified for
every pair of classifications. Such a possibility is implied in Wilson's (1975) con-
cept of "ecological release." Noting that there exists an extraordinary variety of
cultural forms, Wilson asks:
How can such variation in social structure persist? The explanation may be
lack of competition from other species, resulting in what biologists call eco-
logical release. During the past ten thousand years or longer, man as a whole
has been so successful in dominating his environment that almost any kind of
culture can succeed for a while ... (p. 550).
Of course, the existence of modes is a question addressable by empirical
investigation, as McKelvey later indicates. However, before investing the consid-
erable time and energy that such an investigation would undoubtedly require, it
seems worthwhile to first attempt to come to terms with the question theoreti-
cally. Even modest success in such an attempt may prove worthwhile by focusing
subsequent research into areas of high potential payoff.
Recently, the author has been working with a theoretical perspective that
addresses this question (Bigelow, 1977). He began with a concern similar to
Pinder and Moore's: that organizational theories lack relevance for organizational
practice. His view is that inattention to the diachronic dimension of organiza-
tions is a major obstacle to application of many current theories of organizations.
His response has been to develop an evolutionary perspective of organizations,
which has a greater potential for application.
The two views presented above are somewhat different from one another, and
each points to a somewhat different response to the problem. By the same token,
the two have potential for cross-fertilization, which can lead to mutual enrich-
ment. The multiple paradigm model can provide a perspective in which evolu-
tionary contributions can be located and understood. Conversely, evolutionary
concepts can help clarify what is meant by a multiple paradigm model, and how
this might be developed.
The purpose of this paper is to explore the implications of an evolutionary
perspective for the multiple paradigm model. A shortcoming seen in many cur-
rent theories of organizations will first be discussed, which led the author to
move to an evolutionary perspective. Some specific aspects of evolutionary
theory that are pertinent to middle range theorizing are then presented. Follow-
ing this, attention is turned to implications of this perspective for the multiple
paradigm model and middle range theories.
EVOLUTION AND MIDDLE RANGE THEORIES 159
A SYNCHRONIC TENDENCY OF CURRENT

ORGANIZATIONAL THEORIZING
The difficulty in relating many current organizational theories to organizational

practice was noted above. The author believes that a major factor underlying this
difficulty is the synchronic tendency of these theories. Many view organizations
in terms of the dynamics of their current situation, and less so in terms of their
diachronics, or time-related dynamics. This is most clearly the case with struc-
tural theories of organizations, but is also true of information processing,
contingency, and general system theories as well. These theories recognize the
importance of interactions among organizational components, and of environ-
mental influences, but do so in a framework of the present or near-present.
There are a number of reasons why organizational theorists may prefer to
focus on organizational synchronics. Organizations are highly complex, and a
decision to defer for the time being the organization's history can constitute
an important simplifying assumption. Another reason may be that synchronic
theories lend themselves well to empirical investigations which can be com-
pleted in a practicably short period of time. Since a theorist's career path may
ideally involve a series of empirical investigations, there is pressure to compress
the time span of each.
While their synchronics are undoubtedly important in understanding orga-
nizations, there is a growing opinion that synchronics alone do not provide a
complete picture. For example, Greiner (1972) states that" ... the future of an
organization may be less determined by outside forces than it is by the organiza-
tion's history" (p. 38). Stratton (1974), in a case study of organizational change,
criticizes the non diachronic quality of systems theory:
In several instances, the variables which explained behavior were found to be
rooted in the historical development of the organization. Systems Theory,
regardless of the level of analysis, was found to suffer from the limitation of
not being able to properly evaluate these legacies from the past that continue
to influence behavior (p. iii).
Other scholars have also pointed out the importance of historical processes in
organizations (e.g., Pettigrew, 1973, 1976; Starbuck, 1971; Clark, 1972; Strauss,
1974).
The importance of diachronics in understanding organizations is a particularly
crucial point with reference to theory application, since application by its nature
is a time-related process. Moreover, application in complex organizations can be
highly time-consuming. Some organizational development projects, for example,
may take years to consummate. Consequently, it would seem only logical that
synchronic organizational theories are difficult to apply; they lack insight into
the temporal dimension, in which application takes place.
A clear implication of the above discussion is that in order to become more

applicable, theories of organizations must take more into consideration than
organizational diachronics. This was an important consideration for the author
in choosing evolutionary concepts to apply to organizations (Bigelow, 1977).
Evolution is a diachronic theory and focuses on processes by which living sys-
tems develop, maintain, and transform their form. It should be pointed out,
however, that evolutionary concepts have been and are being conceptualized and
used in a number of different ways. Consequently, before discussing implications
of organizational evolution for middle range theory, it is first necessary to make
explicit some elements of the author's conceptualization of organizational evolu-
tion. This is done in the following section.
AN EVOLUTIONARY PERSPECTIVE OF ORGANIZATIONS
The first point to be made is that social evolution, rather than biological evolu-
tion, is the central focus. While a number of authors have considered effects of
biological evolution on social and organizational systems (e.g., Adrian, 1970;
Eibl-Eibesfeldt, 1970; Maclay and Knipe, 1972; Trivers, 1972; Wilson, 1975),
these effects are essentially static in the time frames of most organizational
theorists. Social evolutionary processes, on the other hand, are likely to be in
a more rapid state of change, and the process is more relevant to organizational
theorists and practitioners.
A second point is that the organization is seen as existing in an environment
to which it must adapt-reactively and/or proactively-in order to survive. This
view is consistent with contingency and open system perspectives of organiza-
tions (e.g., Thompson, 1967). It is important to note that "adaptation" is not
necessarily a reactive process, but can be proactive as well: that is, an orga-
nization may adapt by acting on and changing its environment, as well as by
changing itself to fit its environment. It should also be noted that in some in-
stances an organization's survival is not necessarily independent of societal sur-
vival at some higher level. For example, a poorly operating school system may
not be allowed to fail (as might a poorly operating private enterprise) apart from
the governmental system that supports it. In such instances, the higher level
system is an appropriate unit of evolutionary analysis.
A third point is that organizational practices are seen as a principal means of
adaptation. While evolutionary theory shows that at least three other kinds of
entities may be evolving in organizations (Le., knowledge systems, kinds of peo-
ple included, and physical artifacts of the organization), organizational practices
appear to be the central means of adaptation. Other kinds of evolution are
pertinent to organizational adaptation, but tend to follow, rather than lead,
organizational practices (cf. Weick, 1969).
Having stated these three points, it is possible to derme adaptation as a pro-

cess involving: (a) development of a selective process by which successful and
unsuccessful practices can be recognized; (b) generation of a variety of practices;
and (c) continuation or discontinuation of practices based on selection. Here,
evolutionary theory goes beyond open systems and contingency perspectives of
adaptation by derming it as a process of variation, selection, and retention.
Adaptation is not simply an outcome of this process: It is the process. The
nature of this adaptive process in organizations is described below.
Selection occurs through the application of indicators predictive of survival,
rather than through actual survival. An important distinction is made between
natural selection, or survival of the fittest, and vicarious selection (Campbell,
1965), which takes place prior to, and acts to forestall, natural selection. Natural
selection is active in biological evolution and in ecologies of competing organiza-
tions, but is rarely seen within organizations. Here, indicators of organizational
success, such as profit, productivity, job satisfaction, withdrawal, and so on, are
identified and monitored. Practices that appear to enhance the indicators are
continued, while others are discontinued.
Organizations generate variations in practices by borrowing from other orga-
nizations, making incremental variations in existing practices, by creating original
variations, and by adopting "blind" variations. Organizations typically monitor
the practices of other, similar organizations, and tryout practices that appear to
be more successful than their own. Organizations also tryout minor variations of
their own ongoing practices in attempts to boost their indicators. "Original"
variations are sometimes generated (e.g., through a suggestion box), and tried
out. Finally, random variations in practices sometimes lead to improvements,
and if this is noticed, the variation may be institutionalized.
Retention of practices occurs through translation into memories and records,
and through systems that ensure appropriate retranslation back into practice.
"Practices" have no intrinsic existence apart from their doing, and consequently
retention must involve translation of practice into some more durable media.
This in itself is no assurance of retention (Le., continuation of practice, as op-
posed to mere storage). Consequently, organizations devise policies, rules, regula-
tions, training programs, reward systems, and so forth, which regulate the
retranslation into practice.
A final point to be made about organizational evolution is that it does not
occur in a vacuum: Organizations develop concrete, enduring mechanisms
through which evolutionary process occurs. While organizational members may
not be explicitly following an evolutionary model, organizations do develop and
institutionalize mechanisms for variation (e.g., creativity training, R&D units,
norms encouraging innovation), selection (e.g., economic and social indicators,
assessment centers), and retention (e.g., policies, rules, regulations, guidelines,
training). Through these mechanisms, organizations seek out, identify, and retain
organizational practices adaptive to their current situation.
In this section, an evolutionary perspective of organizations has been briefly
overviewed. We see that this is a dynamic and diachronic view, in which orga-
nizations develop mechanisms that regulate the process of variation, selection,
and retention, and thereby adapt to their situation. Let us now consider what
implications this perspective has for middle range theories.
EVOLUTION AND MIDDLE RANGE THEORY
Recall that earlier a question was raised about a preliminary assumption of mid-
dle range theorizing, that organizations will fall into some number of relatively
discrete categories. It is now possible to address this question, using an evolu-
tionary perspective.
Initially, this perspective would appear to be discouraging to the possibility of
organizational categories. EvolUtionary theory ties organizational form to envi-
ronment by an adaptive process, and the lack of environmental categories might
be interpreted as implying the same for organizations. Moreover, the popular
usage of evolution is synonymous with slow and continuous change, which
would not appear to produce the categorical breaks required in middle range
theorizing.
Recently, however, it has been shown that evolutionary models can in fact
include the possibility of rapid changes occurring as a consequence of evolution-
ary process. Waddington (1975), drawing on the work of the topologist Rene
Thorn, proposes a "catastrophe theory of evolution," in which rapid evolution-
ary changes are brought about by the movement of evolving systems from one
evolutionary "canal" to another. Since this model is based on topological theory,
it is as applicable to organizational change as it is to biological change. 1
The possibility of rapid change and discontinuities arising from evolutionary
processes has an important implication for middle range theorizing: These
changes can serve to demark organizational forms into a discrete set of categories
or adaptive modes, each of which (or subsets of which) may be an appropriate
topic for a middle range theory.
In this view, what kinds of discontinuities might we expect to see in organiza-
tions? The literature on organizational change indicates at least three factors that
may lead to discontinuity: organizational niche, the impact of culture, and
chronological stages within single organizations. Each is discussed as follows:
1. Organizational niche. McKelvey makes the interesting observation that
biologists, unlike organization and management researchers, don't need contin-
gency theories, since biological speciation provides a clear classification system
in which to apply biological knowledge. Underlying this observation is the sig-

nificant biological discovery that species are distinct from one another, and
smooth gradations between species do not exist. It was the somewhat startling
discovery of this break point among species which made biological taxonomies
possible. Why does this break point exist? A widely accepted view in biology is
that species evolve in adaptation to a particular set of environmental circum-
stances, or niche. It may be possible for species adapted to two different niches
to interbreed, but the offspring will not be as well adapted to either niche as are
the parents. Thus, the offspring have a lower likelihood of survival, and it may
even be that inhibitions to interbreeding will evolve. In this way, species become
distinct from one another.
Does a similar dynamic exist between different types of organizations? Cer-
tainly we would expect that social evolution, and not biological evolution,
would be the driving force, and evolutionary change would occur at a corre-
spondingly faster rate. Nonetheless, it would appear that organizations often do
evolve in relation to particular semistable environmental circumstances that can
be considered as an organizational equivalent of biological niches. Thompson
(1967) gives the name of "domain" to the characteristic environmental circum-
stances in which an organization exists. Also similar to the biological dynamic,
we know that an organization cannot be all things to all people. A profit-making
organization cannot at the same time be nonprofit, nor can a private organiza-
tion at the same time be public. There is some indication too that organizations
tend to develop characteristic technological domains. A company making soap
may eventually expand into a region of similar technologies, say shampoos and
perfumes. A charitable institution may expand into a number of related charita-
ble causes: However, a soap-making company is unlikely to incorporate charities
into its business, and neither is likely to go into government or steel making.
The effect of niche or domain on organizational form is especially apparent
in organizations that exist in different environmental circumstances at different
times. The United States Army provides an excellent example of an organization
whose mode of adaptation changes with its niche. The army exists in either a
peacetime or a wartime niche. In peacetime, the army is likely to adapt through
a bureaucratic structure, economizing, and public relations. In wartime, the
army moves to a more fluid form, with emphasis on rapid communication, and
greater autonomy (via "command decisions") at lower levels. Thus, as its niche
changes, so does the army's mode of adaptation.
2. The impact of culture. A second factor affecting modes of organizational
adaptation is the cultural system, of which organizations are subsystems. Orga-
nizations are an outgrowth of cultures, which have a strong influence on how
organizations are conceived and implemented. For example, Miller (1978) sug-
gests that organizations can be classified according to the critical functions they
carry out in their society. Different societies, however, may distribute functions
to organizations in different ways. If so, two organizations in two different
cultures could have similar niches, but adapt to them in ways characteristic of
their particular culture. For example, during the 1800s in Ethiopia, the functions
of both government and defense were carried out by a single mobile organiza-
tional unit. Other societies at that time (and presently) distinguished government
and defense into two different organizational units. The way production is car-
ried out in free enterprise and socialistic societies is another example of how the
same function may be carried out in different ways in different societies.
This is an area in which more research is needed. Presently, greater attention
has been given to means by which Western organizations can be transplanted to
other cultures, than to ways by which Western technologies can be uniquely
incorporated by them. Schumacher's (1972) treatise on Buddhist economics is
a needed step in this direction.
3. Organizational stages. There is considerable evidence that organizational
change over time is not necessarily continuous. As Starbuck (1971) states:
The basic proposition advanced is that change in organizational structure is
occasionally punctuated by abrupt, major transformations which sharply
distinguish one period of organizational history from another (p. 275).
A number of scholars have considered discontinuous changes of this type. For
example, Greiner (1972) distinguishes five distinct phases in organizational
growth. Strauss (1974) distinguishes the stages of organizational "youth" and
"adolescence" as organizations grow. Clark (1972) sees organizational life as
characterized by "sagas," each with its own "understanding of a unique ac-
complishment based on historical exploits." In each case, relatively smooth
underlying changes were adapted to by a succession of one or more discrete
adaptive modes.
These successive stages of organizations serve as another possible point of
demarcation for middle range theories. It may be, for example, that a theory
of "young" organizations should be distinguished from a theory of "adoles-
cent" organizations.
The possibility that organizations may transform from one adaptive mode to
another raises a critical issue for middle level theoreticians: when can an organi-
zation be relied on to dependably exhibit a particular adaptive mode, and thus
be an appropriate topic of the corresponding theory? Clearly, if a middle-level
theory is to be applied, there should be some assurance that the organization will
continue to exemplify that theory, at least in the near future.
This issue suggests a possible way of looking at evolutionary theory in a mid-
dle range theory context. It should first be noted that the evolutionary theory
presented here is itself a middle range theory. While some authors, such as
Bajema (1972) and Wilson (1975) view evolutionary theory as an overarching

theory akin to general systems theory, it is presented here more specifically as a
theory of organizational change processes. As such, its scope excludes change
processes in other types of social systems, as well as the results (i.e., actual adap-
tive modes) of organizational evolution.
As a middle range theory, however, evolutionary theory does not itself fall
into any of the classifications which it generates. Rather, its scope includes one
aspect of all organizations. Thus, the image which evolutionary theory portrays
of middle range theorizing is not so much one of a taxonomy (a unidimensional
hierarchical ordering of categories), but one of a matrix, involving more than one
classification dimension. The advantage of this image is that, as indicated above,
it deals with an issue of organizational classification which a taxonomy does not:
namely, the question of how organizations come to be in one classification at
one point in time, and in another at another point. A middle range theory of
organizational change can deal with not only the processes by which organiza-
tions transform from one category to another, but also with larger issues of
utilizing an organizational classification approach. Evolutionary theory indicates
how discrete organizational modes may come into being, and some of the factors
which affect the character of modes. It may be possible to identify regions in
which midrange theorizing is appropriate, and regions in which it may not be;
i.e., situations in which dynamics of discontinuous change do not apply, and
consequently in which discrete modes do not exist. It may be possible to iden-
tify "null set" modes that are possible but not occupied by any organizations,
and emerging modes that are not yet occupied. These are some of the possible
uses to which an evolutionary theory of adaptive modes may be put.
SUMMARY MODEL AND CONCLUSIONS
Evolutionary theory appears to be consistent with a middle range theory ap-

proach. It suggests that organizations may in fact manifest discrete modes of
adaptation, subsets of which may serve as appropriate topics for middle range
theories. Evolutionary theory itself can be considered a midlle range theory,
suggesting a matrix view of middle range theories as opposed to a taxonomy. In
this view, evolutionary theory serves as a theory of modes and organizational
transformations in which a set of middle range theories can be arrayed.
Four dimensions pertinent to classifying modes were discussed, and these are
portrayed in Figure 1. The figure portrays a four-dimensional matrix of cells,
each enclosing a mode of organizational adaptation as subjects of a middle range
theory. The cells are produced by differentiation along four dimensions: (1)
according to the particular niche the organization occupies, (2) according to the
......
0\
0\
Mode Mode Mode Mode
~
ID Ir-------~~------_+--------~--------~
~
.c
~
....uU Mode Mode Mode Mode
Z
.-t·
3. Stages
Figure 1. A Hypercube Model of Middle Range Adaptive Modes.

effects of culture; (3) according to organizational stages; and (4) according to

the change processes involved. As a possible example, one might elect to study
organizations that (1) occupy a healthcare niche (treatment and/or prevention
of a specified set of maladies), (2) are in an African traditional setting, (3) are at
a mature state of development, and (4) are competing with Western medical
technology. One can then develop a midrange theory concerning the organiza-
tional structures, technologies, and practices associated with this mode, as well
as address the question of whether the organization will move to another mode
(e.g., by adopting a Western medical approach, or by changing its niche to treat-
ment of more psychogenic types of maladies).
A turning from attempts to develop general theories of organizations to more
middle-level considerations may seem to some to be a kind of retreat, instead of
an admission that theorists have been unable to generate a truly applicable gen·
eral theory of organizations. This is not the view of the author. The existence of
a general theory of organizations cannot be disproved; it can only fail to be con·
firmed by its existence. An emphasis on middle range theory is a good strategy
for addressing problems of applicability. Such an emphasis does not decrease the
likelihood of eventually developing higher level theory; in fact, it may increase
the likelihood, by making possible a sound middle·level bridge.
NOTE
1. The author and Lynne Markus (Case Western Reserve University, Department of
Organizational Behavior) go into detail on this topic in a forthcoming paper. For more on
catastrophe theory, see Woodcock (1978) and Zeeman (1978).
REFERENCES
Adrian, C. Ethology and bureaucracy. Paper read at the Eighth World Congress
of the International Political Science Association, held at Munich, 1970.
Bajema, C.J. Transmission of information about the environment in the human
species: A cybernetic view of genetic and cultural evolution. Social Biology,
1972, 19(3).
Bigelow, J. Evolution in Organization. Dissertation, Case Western Reserve Uni-
versity, Cleveland, 1977.
Burke, W. W. Organization development in transition. Journal of Applied Be-
havioral Science, 1976, 12( 1),23-43.
Campbell, D. T. Variation and selective retention in socio-cultural evolution. In
H.R. Barringer, G.T. Blankston, and R.W. Mack (Eds.), Social Change in
Developing Areas. Cambridge: Schenkman Publishing Co., 1965.
Clark, B. R. The organizational saga in higher education. Administrative Science

Quarterly, 1972, 17, 178-184.
Eibl-Eibesfeldt, I. Ethology: The Biology of Behavior. New York: Holt, Rinehart
and Winston, 1970.
Friedlander, F., and Brown, L.D. Organization development. Annual Review of
Psychology, 1974, 25, 313-341.
Greiner, L.E. Evolution and revolution as organizations grow. Harvard Business
Review, July-August, 1972,37-46.
Kahn, R. L. Organizational development: Some problems and proposals. Journal
of Applied Behavioral Science, 1974,10, 485-502.
Maclay, G., and Knipe, H. The Dominant Man: The Pecking Order in Human
Society. New York: Dell Publishing Company, 1972.
McKelvey, B. Guidelines for the empirical classification of organizations. Ad-
ministrative Science Quarterly, 1975,20(4),509-525.
Miller, J.G. Living Systems, New York: McGraw-Hill, 1978.
Pettigrew, A. M. Occupational specialization as an emergent process. The Socio-
logical Review, 1973, 21 (2).
~-. The Creation of Organisational Cultures. Paper presented to the joint
EIASM-Dansk Management Center Research Seminary on Entrepreneurs
and the Process of Institution Building, Copenhagen, May 18-20, 1976.
Pinder, C.C., and Moore, L.F. The Inevitability of Multiple Paradigms of Orga-
nizational Behavior. Faculty of Commerce and Business Administration
Working Paper No. 513, University of British Columbia, Vancouver, B.C.,
1978.
Schumacher, E. F. Small is Beautiful: Economics as if People Mattered. New
York: Harper Colophon Books, 1973.
Starbuck, W.H. (Ed.) Organizational Growth and Development. England: Pen-
guin Books, Ltd., 1971.
Stratton, W. E. Organizations and Environment: A Case Study in Adaptation to
Change. Dissertation, Case Western Reserve University, Cleveland, 1974.
Strauss, G. Adolescence in organization growth. Organization Dynamics, Spring,
1974,3-17.
Thompson, J.D. Organizations in Action. New York: McGraw-Hill Book Co.,
1967.
Trivers, R. L. The evolution of reciprocal altruism. Quarterly Review of Biology,
1977,46(4),35-57.
Waddington, C.H. The Evolution of an Evolutionist. New York: Cornell Univer-
sity Press, 1975.
Walter-Busch, E. Problems of the theory of science in the methodology of social
research. Zeitschrift fur Sociologie, 1974,4(1),46-69 (Ger.).
Weick, K. The Social Psychology of Organizing. Reading: Addison-Wesley, 1969.
Wilson, E.O. Sociobiology: The New Synthesis. Cambridge: Belknap Press,
1975.
Woodcock, A., and Davis, M. Catastrophe Theory. New York: Dutton, 1978.
Zeeman, E.C. Catastrophe Theory: Selected Papers. Reading: Addison-Wesley,
1978.
15 ORGANIZATIONAL
SPECIATION
Bill McKelvey, University of California, Los Angeles
1. INTRODUCTION
Organizational systematics has been defined as the science of diversity (McKelvey,

1978a). It includes three component subjects of inquiry: (a) Taxonomy, the de-
velopment of a theory of differences (speciation) among organizational forms,
together with a theory of classification; (b) Evolution, the tracing out of the
historical development of different lineages of organizational forms; and (c)
Classification, the identification and assignment of organizational forms to
formally recognized classes. A move away from grand theories, treating organi-
zations as one homogeneous group of phenomena, toward middle range theories
(Merton, 1957) requires not only vertical differentiation, such as Parsons' (1956)
institutional, managerial, and technical levels, but also horizontal distinction
among kinds of organizations. However, successful accomplishment of the latter
distinction awaits the development of organizational systematics.
This paper pursues the "natural selection" line of inquiry taken recently by
Campbell (1969), Weick (1969), Hannen and Freeman (1977), and Aldrich
(1978, 1979) by looking more closely at the shifts in theories of causality and
classification implied by natural selection theory. The main intent is to propose
a conceptual framework and theory of the organizational speciation process,
169
thereby elaborating the taxonomic subcomponent of organizational systematics.

Such a theory is needed to fully understand: (1) how environmental forces
actually produce differences in organizational form, (2) how a given form remains
stable over successive generations of organizational members in the face of subse-
quent environmental changes and the appearance of new forms, (3) how a group
of organizations remains similar in form but different from other groups, and (4)
how adaptative changes in form take place in the long term.
Organizational speciation is that process which results in the creation of dif-
ferent organizational/onns, the latter defined as follows:
Dl. Fonn: the appearance, condition, and nature of an organization and its
subunits as well as the configuration, interrelationship, and manner of
coordinating its subunits (McKelvey, 1978a).
It is important to note that the inquiry will focus on the creation of a form,
which is embodied in the several members of an organizational specie or group-
ing, not on the life or death of a particular organization as a legal entity. A
specific definition of an organizational specie appears later in the paper.
2. CAUSALITY AND THEORIES OF CLASSIFICATION
The development of organizational taxonomy requires the selection of a theory

of classification and within the latter the identification of a specie concept,
which is at the root of organizational differences. The choice of a theory of
classification in tum depends upon one's approach toward causality. A useful
framework with which to view causation is that supplied by Schwab (1960) after
a review of over 4000 studies in the physical, biological, and behavioral sciences.
He termed them "principles of enquiry." Briefly, they are as follows:
1. Reductive Principle. Scientists search for the cause of an entity's behavior

by lowering the level of analysis to the study of its constituent parts.
2. Rational Principle. Explanation of an entity's behavior is pursued by
raising the level of analysis higher to the larger system or environment in
which it is embedded.
3. Holistic Principle. An object is viewed as a multileveled system of inter-
dependencies, many of which contain two-way causal flows. The focus is
on the pattern of relations among the parts themselves and their relation
with the whole.
4. Antiprinciple. Scientists avoid accepting and being guided by any precon-
ceived theories or hypotheses and try to "let the facts decide." Assump-
tions about causality are avoided as guides to inquiry.
ORGANIZATIONAL SPECIATION 171
5. Primitive Principle. Common sense notions are used to guide inquiry,

including whatever views of causality are implicit in them.
Not surprisingly, theories of classification are rooted in different principles of

enquiry. Over the past two centuries biologists have tried out several theories of
classification with varying success (Mayr, 1969: 65-77). For our purposes the
most important are:
1. Essentialism. All objects are treated as totally analyzable entities that may
be defmed in terms of a fundamental reality or essence giving rise to prop-
erties that are inevitable consequences of the essential definition. Whether
taking biological organisms or organizations, essentialists attempt to trace
all observable attributes back to a few essential characteristics. All exist-
ing typologies of organizations, such as those by Etzioni (1961), Blau and
Scott (1962), Parsons (1956), Katz and Kahn (1966), Perrow (1967), and
Thompson (1967), among others, are based on the identification of a few
essential attributes.
2. Empiricism. This view holds that there are naturally occurring groupings
of objects and that if investigators carry out enough empirical studies, the
groupings will eventually be discovered. They do not posit theories about
how or why the groupings occur. Instead, they form groupings strictly on
the basis of similarity. Some examples of this approach to organizational
classification, using numerical taxonomic methods, are studies by Haas et
al. (1966), Pugh et al. (1968, 1969), Goronzy (1969), and Pinto and
Pinder (1972). Methodological comments are given by McKelvey (1975)
and Warriner (1977a).
3. Evolutionism Adherents of this theory not only try to classify the ob-
jects in question but also attempt to explain their origin and subsequent
differences by looking to natural selection processes which result in the
survival of some forms but not others. Environmental variation, selection
processes that create species, and retention mechanisms that preserve
species once they are created, are the main elements of evolutionary
theory. The result is not only a classification of presently existing entities
but also a development of lineages tracing the entities' descent from com-
mon ancestors. The best example of such a theory is the biological classi-
fication scheme now in wide use. The evolutionary approach has not been
used in classifying organizations. But a recent development is a paper by
McKelvey (1978b) that discussed the evolution of organizational forms in
ancient Mesopotamia.
Table I shows the relation between theories of classification and principles of

enquiry. Ignoring the primitive principle on the assumption that the field has
Table 1. Theories of Classification Related to Principles of Enquiry
Theories Principles
Essentialism Reductive
Evolutionalism Rational
Holistic
Empiricism An tiprinciple
Primitive
progressed beyond that, the only principle that has been left out of existing the-
ories of classification is holism. Though it, in the form of systems theory, has
been on the rise in organization theory, it has not yet led to a paradigm that
has fostered theoretical propositions and subsequent empirical inquiry on a
wide scale. 1
Since organizations, like biological organisms, are not totally analyzable
entities, organizational systematicists probably ought to dismiss essentialism and
its implicit reductive principle as their biological counterparts have already done.
Though there may not be as many different kinds f organizations as there are
biological organisms, organizations may be more complex in that more taxo-
nomic characters may be needed to adequately describe their total behavior
As Burns (1967), Hall et al. (1967), and McKelvey (1978a) have noted, the
simple essentialist typologies have proved inadequate in face of the richness of
total organizational behavior, leading to their rejection by researchers needing a
workable classification for use in identifying meaningful subpopulations of
organizations for empirical study.
In biology, systematicists have tended to adhere either to evolutionism (Mayr,
1969) or to empiricism (Sneath and Sokal, 1973), with each largely rejecting the
opposing theory. McKelvey (1978a) has argued for an approach combining both.
Thus evolutionary theory would be used a priori to suggest possible classes
which are then tested out by the more objective, broader based, numerical tax-
onomic methods. As noted before, without a solid a priori theory the empirical
methods run into difficulties. And without empirical testing there is no check on
biased interpretations of evolutionary development due to poor and occasionally
unavailable historical data. In short, each theory benefits the other.
3. THE RATIONAL PRINCIPLE AND THE EVOLUTION OF

ORGANIZATIONAL FORM
The missing link in the combined evolutionary-empiricist approach advocated by

McKelvey is the lack of an evolutionary theory of organizational classification,
and utimately an evolutionary theory of organizational speciation. The combi-

nation of the rational principle of inquiry and evolutionism has not yet been
widely accepted in the study of organizational differences. But there are three
streams of thought suggesting that a full-fledged environmentally responsive
evolutionary approach is close at hand: the natural selection, contingency, and
resource-dependency theories.
Natural Selection. Campbell (1969) noted a reawakening of interest in evolu-

tionary theory in social science. Drawing on the theory of natural selection in
biology, he identified what appear to be the necessary and sufficient conditions
of sociocultural evolution. They are as follows:
1. The occurrence of variations: heterogeneous, haphazard, 'blind,' 'chance,'

'random,' but in any event variable (the mutation process in organic
evolution, and exploratory responses in learning).
2. Consistent selection criteria: selective elimination, selective propagation,
selective retention, of certain types of variations (differential survival of
certain mutants in organic evolution, differential reinforcement of certain
responses in learning).
3. A mechanism for the preservation, duplication, or propagation of the pos-
itively selected variants (the rigid duplication process of the chromosome-
gene system in plants and animals, memory in learning) (Campbell, 1969:
73).
Campbell's broadly defined interest in the evolution of all kinds of socio-

cultural units has been applied more recently to the process, as opposed to the
outcome, of evolution in organizations, by Weick in his book on the process of
organizing (1969). The natural selection model as applied to organizations re-
ceived a preliminary review by Aldrich and Pfeffer (1976) and a more extended
treatment is found in Aldrich (1979). The natural selection model has recently
been applied by Hannen and Freeman (1977) to explain organizational differ-
ences through use of the biological population ecology model.
Contingency Theory - The Technology Hypothesis. This approach was best

described by Perrow (1967) as the following chain of events:
1. Because of the environments organizations choose or are forced to oper-

ate in, their goals, outputs, and inputs are specialized and differentiated.
2. These attributes create certain problems in the workplace and organi-
zations develop appropriate technologies to solve them.
3. These technologies in turn create problems that require specialized

managerial systems and organizational structures for their solution (not
a quote).
Following the work of Lawrence and Lorsch (1967) this view came to be la-
belled "contingency theory." Over the years there has developed strong em-
pirical support for the central ideas that organizational technology and structure
are contingent on the kind of environment prevailing (Woodward, 1958; Burns
and Stalker, 1961; Chandler, 1962; Udy, 1964; Duncan, 1971; Negandi and
Reiman, 1972; Child and Mansfield, 1972; Galbraith, 1973; Khandwalla, 1974;
Lorsch and Morse, 1974; and Child, 1974, 1975; among others).
Resource-Dependency Theory- The Power Hypothesis. The chain of events

based on power relations is as follows:
1. The environment provides many of the constraints, uncertainties, and con-

tingencies because of the necessity for transacting with the environment.
2. These contingencies affect the distribution of power and influence within
organizations, providing some subunits with more power and others with
less.
3. Power is used in determining organizational social structures, particularly
to the extent that there is uncertainty and the decisions concern critical
issues (Aldrich and Pfeffer, 1976: 89).
The theoretical and empirical work in support of this approach has recently been
exhaustively reviewed by Aldrich and Pfeffer (1976) and Pfeffer (1977), so need
not be discussed here.
Without entering into a lengthy discussion, let it be noted that the natural
selection model does not require acceptance of a view of environmental deter-
minism in which managers are denied choice, as is implied by Hannen and
Freeman (1977). Instead, the environment may be seen as a set of objective con-
straints, which managers may or may not correctly discern or enact (using
Weick's (1969) term) and chose to respond to. Thus managers are never denied
the option of not adapting to the prevailing environment. The fact that a man-
ager chooses not to respond or does so poorly does not undermine the long-run
viability of the form of an organizational population or specie best suited for
its niche 2 anymore than the choice of a rabbit to jump in front of a truck under-
mines the long-run viability of its specie.
4. CONSTRAINTS FOR A THEORY OF ORGANIZATIONAL

SPECIATION
The central element in a theory of organizational speciation is the concept of an

organizational specie. The development of such a concept proceeds within a set
of constraints posed by a model of what an ideal classification might look like
and by the model of natural selection presented earlier. There are four condi-
tions a specie concept must satisfy to become the foundation of an effective
evolutionary classification:
1. The concept should lead to observation of sharp discontinuities among

the organizational forms being classified. The sharper these demarcations
are, the easier to develop a classification scheme. Without observed dis-
continuities, classification becomes arbitrary.
2. It should result in high levels of homogeneity within classes in addition to
sharp demarcations between them. Without homogeneous classes a classi-
fication loses its value as a means of ider.tifying subpopulations of organi-
zations for empirical investigation and as a means of codifying knowledge
about their total behavior.
3. The specie concept should result in groupings that are stable over a period
of time long enough to make the effort of classifying the groupings
worthwhile.
4. Since an evolutionary approach is advocated here, the specie concept
should explain how the mechanism assuring stability in the short run,
which satisfies condition 3, also operates to allow evolutionary changes
in the long run.
The constraints imposed by natural selection theory stem directly from the
three essential processes: variation, selection, and retention. Taking a cue from
the way in which the biological specie concept (Mayr, 1969) handles these con-
straints, an organizational specie concept probably should have analogs to:
1. A Generational Mechanism. In biological organisms the intercommuni-

cating gene pool, helped along by occasional mutations, is the root process
assuring variations. Because the total number of gene combinations avail-
able within a specie is very much greater than the combinations held by a
single member of a specie the result is, except in exceedingly rare in-
stances, that each member is somewhat different, and with mutations,
quite different. Coupled with isolating mechanisms (discussed shortly),
the gene pool also is the means by which the specie form is passed on
from generation to generation. Thus, the intercommunicating gene pool

is responsible both for assuring stability within the specie in the short run
(because some gene combinations are more likely) and for allowing long-
run evolution due to less likely gene combinations and mutant forms
occurring that are beneficial to survival.
2. An Ecological Mechanism. Among biological organisms the Darwin-
Wallace theory of the survival of the fittest explains why some variations
persist while others die out. As a result of conditions imposed on the
members of the specie by their niche, some members, and the gene com-
binations they hold, survive, mate, and otherwise pass on their particular
combinations. Over a period of time the most successful combinations
come to characterize the specie.
3. Isolating Mechanisms. In most biological organisms, reproductive isolation
is the principal means by which a limited set of gene combinations comes
to be passed along from generation to generation. Without isolating mech-
anisms a specie would not persist even if it were created in the first place.
Even though it mingles with other organisms it is prevented from inter-
breeding with them. Thus species, once formed, remain separate.
The biological specie concept is not perfect in its handling of all possibilities
of biological speciation, though it does handle the mainstream of sexually re-
producing organisms. Ross (I974) concluded that rather than argue whether or
not the main specie concept handled all situations, it would be preferable to have
additional specie concepts to handle special situations. Such will undoubtedly be
the case for organizational specie concepts as well.
5. A CORE TECHNOLOGY SPECIE CONCEPT DEFINED
Following Monod (1971), Hannan and Freeman (1977) argue for identification
of an organizational blueprint analogous to the genetic code of biological species.
Such a blueprint would "consist of rules and procedures for obtaining and acting
upon inputs in order to produce an organizational product or response" (Hannan
and Freeman, 1977: 935). After the contingency theorists in general and Perrow
(1967) in particular, the view taken here is that what will shortly be defined as
an organization's core technology constitutes the relevant body of knowledge,
skills, rules, and procedures for obtaining and converting inputs into outputs.
Defmitions of key concepts are collected in this section along with brief ex-
amples. An explanation of how the concepts combine to form a viable organiza-
tional specie concept is given in the following section.
D2. Organizational species: polythetic, technology-sharing populations of

organizations isolated from each other because their tech-pools are not
easily learned or transmitted.
D3. Tech-pool: a population of mutually compatible technological elements
that may be combined variously to form a particular core technology.
04. Polythetic group: one where: (a) each member possesses many proper-
ties, p, of a set of properties, P (b) each p in P is possessed by many
members; and (c) no p in P is possessed by all members (after Beckner,
cited in Sneath and Sokal, 1973: 21). In a monothetic group all mem-
bers possess the same set of properties.
The specie concept is based on two key concepts drawn from biological tax-
onomy, polythetic groups and the intercommunicating gene pool. Biologists
have long since given up on the idea of forming a classification around mono-
thetic groups-the phenomena are too varied to allow this. Instead, they have
found the notion of polythetic grouping fully satisfactory. Since organizations
are also highly varied, it is suggested that organizational systematicists will want
to adopt the polythetic group concept. 3
Though there is always risk in adopting analogs from another science, it will
be argued that the tech-pool concept solves similar conceptual problems in
dealing with organizational stability and adaptation. As outlined in the follow-
ing section, the concept of the intercommunicating tech-pool solves a number
of problems related to conceptualizing and explaining how organizational
species remain stable across generations in the short term and adapt in the long
term. Furthermore, it will be argued that technological elements, via manager
and employee choices, intercommunicate and combine very much like genes do
to form the technological base of individual organizations.
In thinking about technology it is important to distinguish between the sub-
set of technical and managerial elements used by managers in a particular organi-
zation and the broader pool of technological elements from which they are drawn.
This pool, containing all known elements of a technology, is defined as the tech-
pool. Thus, the tech-pool of founding consists of all skill and knowledge ele-
ments pertinent to operating the technical and managerial aspects of a foundry
organization. A particular foundry would not have within its grasp all elements
of the broad tech-pool, though each foundry would have many elements of the
tech-pool. And, each element of the tech-pool would be present in many foun-
dries. Thus, the members of an organizational specie comprise a poly the tic group
sharing many elements of their tech-pool. The relation between the tech-pool
and the organizations that are its members is depicted in Figure 1.
Technical Elements
abcdefghijklmn
Alpha
~ abcdefgh k n abcdef hijk
___________n ~ Beta
~-----"-----....
abc ef hijklmn ....- - -_ _ Gamma
Figure 1. Relation between Tech-Pool and Organizations that are Members. In this example
the tech-pool comprises 14 technological elements (real organizations have vastly greater
numbers of elements). Each member has a subset of the elements, none of which are quite
the same, though overlap is considerable. The number of elements held by a member may
vary; in this case, Gamma has one more element than Alpha or Beta. The tech-pool is shown
to be intercommunicating with its members. The elements within a member are held tempo-
rarily by its employees, with no employee holding all elements.
The technological base of organizations is defmed as follows:
DS: Core technology: the knowledge and skills used in the primary task
workplace and the direct management of that workplace.
D6: Primary task: comprises the set of activities that bear directly on the
conversion inputs into those outputs critical to an organization's
survival (Miller and Rice, 1967):
(a) there is only one primary task per specie;
(b) it is possible for an organization as a legal entity to have several
primary tasks (a conglomerate, for example);
(c) peripheral service or staff units in larger organizations are not ele-
ments of the primary task; however, they may be seen as having
their own primary tasks.
D7: Workplace management task: comprises the set of activities that bear
directly on carrying out boundary transactions between the workplace
unit and other organizational subunits or entities comprising the organi-
zation's niche.
Space precludes full development of the pros and cons of the primary task con-
cept. In most organizations it is the production, marketing, or fmancing func-
tion, singly or in combination. Thus, the primary task of a foundry is pouring
molten metal into molds; the primary task in an elementary school is education,
not serving hot lunches. In BIC Pens, Ltd_ the primary task is marketing. In
Procter and Gamble, Inc., it is a synergistic combination of production and

marketing. In a university it is a synergistic combination of research and teach-
ing; in RAND Corp. it is research; in a small college it is teaching. In large con-
glomerate or vertically integrated organizations there may be several primary
tasks and associated forms. The divisions each operate one or more primary
tasks such as ship building, entertainment, mining, construction, and so on, and
the corporate headquarters organization is yet another form of organization with
a different primary task, usually fmancial control.
The workplace-management task pertains to the managerial activities directly
concerned with assuring the effective design and operation of the hardware or
technical system of the workplace. Managers try to protect the technical core
from uncertainty or to design it to be more flexible. They try to control the
activities to foster goal attainment. They try to differentiate the activities
to correspond with a heterogeneous environment and to take advantage of
specialized skills. They attempt to integrate all the activities into a unified
whole. Generally, these activities seem to fall into three broad categories:
measuring effectiveness; mitigating environmental contingencies such as threat,
change, heterogeneity, and uncertainty; and coordinating interdependencies
(McKelvey, 1977). Note that the core technology concept is not a hardware-
oriented concept at all, but in fact is much more broadly concerned with many
aspects of managerial information processing and organizational behavior knowl-
edge and skills as well.
6. EVALUATION IN TERMS OF THE CONSTRAINTS
To be viable, the core technology specie concept, coupled with the tech-pool
concept, should include generational, ecological, and isolating mechanisms.
Generational Mechanisms. It is customary to think of a biological specie, an

animal or plant, as a physical object. Instead, think of a specie as a set of highly
probable gene combinations that are passed through time by being held tempo-
rarily in the body of a plant or animal. An organizational form may be thought
of in the same way-it is a set of highly probable combinations of technological
elements that: (a) are passed around to the several organizational members of
the specie and (b) are passed on in time by the employees of the organizations.
Each organizational member, and each employee within that member, knows
only a subset of the total tech-pool. Thus, there is a unique combination of
elements within each organization. Since each organization possesses many of
the elements, its version of the form implicit in the tech-pool varies somewhat
from the form of the tech-pool embodied in another organizational member of
the specie. This is a source of variations in organizational form due to each
180 MIDDLE RANGE THEORY AND THE STUDY OF ORGANIZATION,)
organization's choice (via its employees) to use a certain unique subset of the
tech-pool.
As does its biological counterpart, the tech-pool also offers a way of concep-
tualizing short-term stability and long-term evolutionary change. The short-term
stability of an organizational specie obtains because all members of the specie
draw from the same tech-pool of successful, and therefore highly probable, com-
binations of technological elements. Organizations tend to keep on using ele-
ments that are effective. They reward employees for learning the elements of the
tech-pool that work. They hire from other members of the specie employees
who are known to have mastered needed technological elements. As long as
there are individual and organizational rewards for mastering the elements of the
tech-pool, the tendency will be toward stable combinations of elements closely
reflecting, but not identical to the composition of the tech-pool.
Long-term evolutionary change in the form of incessant, usually small changes
(they mayor may not be improvements) is also covered by the tech-pool specie
concept. During the course of doing his or her job an employee may discover a
new way of doing something. This is a variation. If it is successful it probably
will be spread around the organization (though there are numerous case studies
showing that certain kinds of incentive systems mitigate against this, at least in
the short term-strike another note for stability) and eventually to other organi-
zations making up the specie. It is also possible that a new employee, though
being taught by an old master about to retire or leave, brings in a new way or
soon discovers one. This is a variation that eventually, if successful, may be
passed on into the tech-pool. Another possibility is that a new employee fails
to learn from the truly skilled old master and so the organization loses part of
its grasp on the tech-pool. This variation, if it happens often enough, may lead
to the failure of the organization or it may lead to the discovery of other suc-
cessful variations that replace the old elements and eventually take their place in
the tech-pool.
Ecological Mechanisms. To be useful the core technology concept must offer a

way of conceptualizing how environmental forces operate to select some varia-
tions and dampen down others. Khandwalla's (1977) study offers ample evidence
that niche attributes such as hostility, turbulence, heterogeneity, uncertainty,
and restrictiveness affect organizations. In terms of the core technology concept,
technological elements are seen to be differentially effective in the face of these
and other niche attributes. Those elements that are effective tend to persist
while those that are not disappear, at least in effective organizations in the long
term.
Isolating Mechanisms. Isolating mechanisms serve to keep a specie differentiated

from other species. There seem to be several mechanisms that play this role in
keeping one tech-pool from merging with another over fairly long periods of
time. Organizations, for example aerospace companies, tend to hire employees
who know certain parts of the "aerospace tech-pool." They tend not to hire
miners, nurses, teachers, or airline pilots, and so forth, who have knowledge of
technological elements from different tech-pools. In this way, the aerospace
companies remain isolated from other tech-pools. Even though an organization
might hire a member from a totally different tech-pool (such as a business school
hiring a business person as dean) the "strange" member knows only a small
portion of the tech-pool from which he or she came.4 Since the tech-pool is
typically widely diffused among many people it is very difficult, if not im-
possible, for an organization to quickly change from one tech-pool to another
simply by hiring people. Organizations also tend to try to retain people who
they see are valuable tech-pool holders. Often the elements of a tech-pool are
held by artisans; learning a craft takes a long apprenticeship or an on-the-job
training program - these increase the difficulty of an organization quickly
picking up a different tech-pool. Many tech-pools are held by professional
people such as professors, engineers, and scientists. Because the training is
long and often difficult, the job of learning even a part of the tech-pool of an
organization such as an aerospace company is very difficult. Furthermore,
employees holding part of a tech-pool tend to mix with other people holding
part of the same tech-pool- they attend the same professional conferences
and so forth. Warriner (1977b) noted that accreditation agencies, federal regu-
latory agencies, professional associations, and the like, also acted as isolating
mechanisms.
One of the advantages of drawing heavily on the biological specie concept
as a metaphor is that the various mechanisms it entails also work to satisfy the
conditions imposed by an ideal classification model. The isolating mechanisms
work to assure homogeneity within classes and sharp demarcations between
them. The tech-pool notion, which interrelates with the generational and eco-
logical mechanisms is the device that allows the core-technology concept to
handle the problem of short-term stability and long-term evolutionary change.
In conclusion, it seems reasonable to go forward with the view that the core
technology concept is a viable specie concept. Preliminary acceptance of this
concept does not preclude the development of alternative specie concepts, how-
ever. It is far too soon to assume that the core technology concept, or any other
soon to be presented, will be the concept gaining eventual broad acceptance.
7. A MODEL OF THE ORGANIZATIONAL SPECIATION PROCESS
The proposed model attempts to represent the process by which organizational

forms come to be different and remain that way. It is offered below in outline
form. It is patterned after a model of the biological speciation process suggested

by Ross (1974: 80-84).
Stage 1. Relevant technological elements are present in isolated pockets. In
the case of American railroads, the relevant technological elements had been ac-
cumulated by the British locomotive builders, and American mining tramway
operations, steamship operations, foundries, small steam engine builders, over-
land stage lines, and canal operations. Also included here would be what
Stinchcombe (1965) labelled the organizational capacity of a society. Additional
societal precursors to the creation of new organizations have been suggested by
Aldrich (1978).
Stage 2. An invention or discovery or a change in environmental circum-
stances prompts the mixing of previously unmixed technological elements. In
the case of the railroads, the success of the Erie Canal, coupled with a war with
England, which dried up federal canal building funds, and the hilly region west
of Baltimore and Charleston, prompted people to turn their interest toward rail-
roads and the steam locomotive. It does not seem that a new tech-pool and
organizational form result simply from the invention of a new technological
element such as the steam locomotive, airplane, assembly line, computer, or
linear programming, and so forth. Instead, once a new element is recognized
there is a slow mixing of previously unmixed elements. In some cases the new
technological element is added on to a previous mix, such as assembly lines in
automobile organizations, and hence there is an evolution within an organi-
zational form rather than actual creation of a new form.
Stage 3. During the period the new technological mix has an advantage, there
is the gradual development of a tech-pool of unique character. rhus, during the
first 50 years or so of the railroads, there was the crystallization of a tech-pool
uniquely associated with difficulties in running railroads across vast stretches of
land and in operating heavy equipment. During this time the tech-pool becomes
more "ingrown" and increasingly insensitive to other outside developments. rhe
personnel holding technological elements come increasingly from other organi-
zations using the same tech-pool, as do new developments in technology.
Stage 4. The new core technology evolves to the point where elements from
other tech-pools prove inferior, or are clearly not applicable. In biology this
stage is analogous to the point in speciation where two species may mate but
produce inferior or sterile offspring. At this stage, people "in the business"
would usually feel little to be gained by bringing in outsiders or ideas from other
businesses, and there would be a tendency to reject things "not invented here."
Insiders would feel outsiders do not understand how things need to be done, and
so forth. Business schools at this stage might bring in businesspersons as deans
but the message would be "keep your hands off the running of faculty and
academic affairs." ("But it would be fine if you would go outside and raise
money for us!")
Stage 5. The separation of the tech-pool is reinforced by the development of

isolating mechanisms. Training in the core technology would be time-consuming
and expensive. The technology would be complex enough to preclude an organi-
zation picking it all up easily by having existing employees learn it or by hiring
small numbers of new employees. Federal regulations, professional associations
and the like also would have emerged. In the railroad business problems with
unions, the ICC regulations, customer preferences, not to mention the problems
of making efficient use of expensive equipment, would accumulate to make it
difficult for outsiders to come in and master the core technology, and it would
be equally difficult for another organization with a different core technology
to easily become effective in the railroad business. One of the anecdotes asso-
ciated with the failure of the new rapid transit line in the San Francisco Bay area
(BART) was that in all their enthusiasm for new technology they never got
around to hiring someone who could "run a railroad." BART was developed in
the space-age, using much of its tech-pool. They forgot how important the rail-
road tech-pool was.
CONCLUSION
After reviewing notions of causality implicit in several theories of classification,

an argument was made in support of an evolutionary approach to thinking about
how differences among organizations are created. A further argument was made
that effective classification of organizations cannot proceed until a workable
specie concept is developed. Pursuant to this, the core technology concept, to-
gether with the tech-pool concept, were presented. The paper ended with a brief
outline of major stages in the organizational speciation process.
The ideas presented in this paper are not meant to end discussion on organi-
zational speciation but rather to begin it. There are several logical next steps.
One is to pursue the development of specie concepts alternative to the core
technology notion. There needs to be fuller consideration of the argument made
here and elsewhere (McKelvey 1978a) that numerical methods will remain in
the doldrums until viable a priori theories of organizational differences are
accepted. The evolution of organizational forms, as opposed to management
thought, is a largely untapped subject. Finally, the development of tighter, more
empirically grounded, middle range theories tied to different species of organi-
zations needs to be pursued. These might include theories of organization
design, change, motivation and leadership, and planning and control systems, to
name but a few. Until such theories are focused and compartmentalized, except
at the most abstract levels, it seems likely that organization science will have
little solid impact on practical management.
NOTES
1. For further critical review of systems theory see Melcher (1975).

2. See Hannan and Freeman (1977) for a discussion of niche theory.
3. Decision rules bearing on the classification of organizations into polythetic groups
are the subject of classification method, not dealt with in this paper.
4. Shetty and Perry (1976) found that chief executives brought into an organization
from outside performed at a lower level than those already inside.
REFERENCES
Aldrich, Howard. Social Structure and the Creation of Organizations. Paper

presented at the meetings of the Southern Sociological Society, New Orleans,
March 3 I-April 1, 1978.
- - . Organizations and Environments. Englewood Cliffs, N.J.: Prentice-Hall,
1979.
Aldrich, Howard E., and Pfeffer, Jeffrey. Environments of organizations. Annual
Review of Sociology, 1976,2,79-105.
Blau, Peter M., and Scott, W. Richard. Formal Organizations. San Francisco:
Chandler, 1962.
Burns, T. The comparative study of organizations. In V. Vroom (Ed.), Methods
of Organizational Research. Pittsburgh University Press, 1967.
Burns, T, and Stalker, G.M. The Management of Innovation. London: Tavistock,
1961.
Campbell, Donald T. Variation and selective retentIOn in socio-cultural evolution.
General Systems, 1969, 14, 69-85.
Chandler, Alfred. Strategy and Structure. Cambridge: The MIT Press, 1962.
Child, John. Managerial and organizational factors associated with company
performance, Part I: A contingency analysis. The Journal of Management
Studies, 1974,11, 175-189.
- - . Managerial and organizational factors associated with company perfor-
mance, Part II: A contingency analysis. The Journal of Management Studies,
1975,12,12-27.
Child, John, and Mansfield, Roger. Technology, size and organization structure.
Sociology, 1972,6, 368-393.
Duncan, Robert B. The implementation of different decision making structures
in adapting to environmental uncertainty: An expansion of contingency
theories of organization. Proceedings of the Academy of Management, 1971,
39-47.
Etzioni, Amitai. A Comparative Analysis of Complex Organizations. New York:
Free Press, 1961.
Galbraith, Jay. Designing Complex Organizations. Reading, Mass.: Addison-
Wesley, 1973.
Goronzy, FriedheIm. A numerical taxonomy of business enterprises. In A.J.

Cole (Ed.), Numerical Taxonomy. London: Academic Press, 1969,42-52.
Haas, J. Eugene, Hall, Richard H., and Johnson, Norman J. Toward an em-
pirically derived taxonomy of organizations. In R. V. Bowers (Ed.), Studies
on Behavior in Organizations. Athens: University of Georgia Press, 1966,
157-180.
Hall, Richard H., Haas, J. Eugene, and Johnson, Norman J. An examination of
the Blau-Scott and Etzioni typologies. Administrative Science Quarterly,
1967,12, 118-139.
Hannan, Michael T., and Freeman, John. The population ecology of organiza-
tions. American Journal of Sociology, 1977,82, 929-964.
Katz, Daniel, and Kahn, Robert L. The Social Psychology of Organizations. New
York: Wiley, 1966.
Khandwalla, Pradip N. Mass output orientation of operations technology and
organizational structure. Administrative Science Quarterly, 1974,19,74-97.
- - . The Design of Organizations. New York: Harcourt Brace Jovanovich, Inc.,
1977.
Lawrence, Paul R., and Lorsch, Jay W. Differentiation and integration in com-
plex organizations. Administrative Science Quarterly, 1967, 12, 1-47.
Lorsch, Jay W., and Morse, John J. Organizations and Their Members. New
York: Harper & Row, 1974.
Mayr, Ernst. Principles of Systematic Zoology. New York: McGraw-Hill, 1969.
McKelvey, Bill. Guidelines for the empirical classification of organizations.
Administrative Science Quarterly, 1975,20, 509-525.
- - . Organizational Systematics. Graduate School of Management, UCLA.
1977.
- - . Organizational systematics: Taxonomic lessons from Biology. Management
Science, 1978a, 24, 1428-1440.
- - . The Evolution of Organizational Form in Ancient Mesopotamia. Paper
presented at the meetings of the Academy of Management, San Francisco,
August 9-12, 1978b.
Melcher, Arlyn J. Theory and application of systems theory: Its promises,
problems and realizations. Organization and Administrative Sciences, 1975,
6,3-9.
Merton, Robert K. Social Theory and Social Structure (2nd ed.). Glencoe, 111.:
Free Press, 1957.
Miller, E.J., and Rice, A.K. Systems of Organization. London: Tavistock Insti-
tute, 1967.
Monod, Jacques. Change and Necessity. New York: Vintage, 1971.
Negandi, Anant R., and Reimann, Bernard C. A contingency theory of organiza-
tion re-examined in the context of a developing country. Academy of Man-
agement Journal, 1972,15, 137-146.
Parsons, Talcott. A sociological approach to the theory of organizations. Admin-
istrative Science Quarterly, 1956,1, 63-85 and 2, 225-239.
Perrow, Charles. A framework for the comparative analysis of organizations.

American Sociological Review, 1967,32, 194-208.
Pfeffer, Jeffrey. Power and resource allocation in organizations. In B. Staw and
G. Salancik (Eds.), New Directions in Organizational Behavior. Chicago: St.
Clair Press, 1977.
Pinder, Craig C., and Moore, Larry F. The resurrection of taxonomy for the
development of middle range theories of organizational behavior. Administra-
tive Science Quarterly, March 1978.
Pinto, Patrick R., and Pinder, Craig C. A cluster-analytic approach to the study
of organizations. Organizational Behavior and Human Performance, 1972,8,
408-422.
Pugh, D.S., Hickson, D.J., and Hinings, C.R. An empirical taxonomy of struc-
tures of work organizations. Administrative Science Quarterly, 1969, 14,
115-126.
Pugh, D.S., Hickson, D.l., Hinings, C.R., and Turner, C. Dimensions of organiza-
tional structure. Administrative Science Quarterly, 1968,13, 65-105.
Ross, Herbert H. Biological Systematics. Reading, Mass.: Addison-Wesley, 1974.
Schwab, Joseph J. What do scientists do? Behavioral Science, 1960,5, 1-27.
Shetty, Y.K., and Perry, Newman S., Jr. Are top executives transferable across
companies? Business Horizons, 1976,19, 23-28.
Sneath, Peter H.A., and Sokal, Robert A. Numerical Taxonomy. San Francisco:
Freeman, 1973.
Stinchcombe, Arthur L. Social structure and organizations. In J.G. March (Ed.),
Handbook of Organizations. Chicago: Rand McNally, 1965, 142-193.
Thompson, James D. Organizations in Action. New York: McGraw-Hill, 1967.
Udy, Stanley. Administrative rationality, social setting, and organizational devel-
opment. In W.W. Cooper, H.J. Leavitt, and M.W. Shelley II (Eds.), New
Perspectives in Organization Research. New York: Wiley, 1964, 173-192.
Warriner, Charles K. Empirical Taxonomies and the Comparative Study of
Organizations. Paper presented at the American Sociological Association
Annual Meeting, Chicago, 1977a.
- - . Personal Communication, 1977b.
Weick, Karl E. The Social Psychology of Organizing. Reading, Mass.: Addison-
Wesley, 1969.
Woodward, Joan. Management and Technology. London: HMSO, 1958.
16
THE RESURRECTION
OF TAXONOMY TO AID THE
DEVELOPMENT OF MIDDLE RANGE
THEORIES OF ORGANIZATIONAL
BEHAVIOR*
Craig C. Pinder and Larry F. Moore,
Since the publication of Katz and Kahn's The Social Psychology of Organizations
(1966) and Thompson's Organizations in Action (1967), the open systems
model has dominated the thinking of scholars interested in organization theory.
Although a number of authors, such as Scott (1961), had discussed the systems
view of organizations before 1966, Katz and Kahn's book was the first major
exemplar of the systems model, and the one probably most often cited since
then in connection with the systems paradigm. For Katz and Kahn, the stated
purpose of applying the systems model to organizations was to facilitate the
integration of so-called "macro" and "micro" concepts, thereby fostering some
commonality of terms and concepts. Further, they hoped that the systems
model would escape a commonly alleged fault of earlier approaches to the study
*Reprinted from "The Resurrection of Taxonomy to Aid the Development of Middle

Range Theories of Organizational Behavior" by C. C. Pinder and L. F. Moore in Administra·
tive Science Quarterly, vol. 24, no. 1, by permission of Administrative Science Quarterly.
© 1979 by Cornell University. We are grateful to Hari Das, Peter Frost, David Hayes, George
Gorelik, Bill McKelvey and, Louis Pondy for their helpful comments on earlier drafts of this
article. We also thank Brian Scrivener of the University of British Columbia Press for his
invaluable copyediting assistance before the article was accepted for publication.
187
of social organization, namely, a tendency to rely on analogies and metaphors

that were not entirely appropriate (Katz and Kahn, 1966: 9).
The suggestion by Katz and Kahn that organizations be viewed as open sys-
tems was based largely on their observation of a number of characteristics of
organizations that are common to many other phenomena and physical entities
normally viewed as "systems." Since then, these commonalities have seemed so
apparent that most organization theorists and researchers have reinforced the use
of the system metaphor by fleshing it out, by exploring its implications for the
management of organizations, and by working toward methodological tech-
niques suitable for testing theories rooted in the systems tradition.
Recently, a number of critics have implied that through its widespread adop-
tion as the dominant organizational "image" (Ritzer, 1975), the systems model,
ironically, itself has been fitted with analogies and metaphors that are not always
entirely appropriate to organizational reality. The earliest example of this criti-
cism is found in Zaleznik's (1967) review of Katz and Kahn's first edition. More
recent attacks are found in the comments of Kast and Rosenzweig (1972),
Thayer (1972), Weick (1974), Burack (1975), Melcher (1975), Milburn et al.
(1975), and Pondy and Mitroff (1979).
Aside from the appealing similarities seen by many theorists between orga-
nizations and other systems, the systems metaphor has at least one other major
feature that has made it popular in recent years-its universality. It has been
seen as applicable to all organizations, thereby offering an element of theoretic
parsimony not offered by less general frameworks. The cost of parsimony in
social theory, however, is often predictive inaccuracy for many individual cases
(Friedman, 1953), and this problem underlies many of the criticisms of the sys-
tems approach cited above. The recent criticism of the widespread use of the
systems metaphor is based, in part, on the suspicion that maybe the imagery it
entails is not completely justified and, in part, on a rejection of the universality
it presupposes.
The recent popularity of "contingency theories" can be seen as a recognition
of the problems underlying general theorizing and universalistic management
techniques. Although contingency theories are general in the sense that they
help to make sense of all (or most) organizations, they limit the generality of the
specific descriptive or prescriptive statements they make to particular subsets of
the total domain of study (Fiedler, 1967; Lawrence and Lorsch, 1969; Vroom
and Yetton, 1973). Contingency theorists compromise parsimony and hope to
gain more predictive accuracy and understanding of individual cases by intro-
ducing one or more contingency or moderator variables into the analysis of
organizations, so that the treatment accorded anyone case depends upon its
position with regard to the contingency variable(s). In many ways, contingency
theories, although they are not inconsistent with the systemic image of systems
THE RESURRECTION OF TAXONOMY 189
theory, are in sympathy with the sentiments of critics such as Weick (1974),
who favor a shift away from the search for general theory toward the develop-
ment of "theories of the middle range" (Merton, 1968).
In contrast to general theories that purport to apply to all organizations,
theories of the middle range attempt to predict and explain only a subset of all
organizational phenomena. As such, each midrange theory makes different sets
of assumptions about organizations, considers different parameters to be impor-
tant, and leads to entirely different prescriptions for practice from other mid-
range theories. In fact, using Ritzer's (1975) paradigm concepts, each midrange
theory may be based on a unique set of images of what constitutes organiza-
tional behavior and on unique research strategies and tactics.
There are at least three ways to develop middle range theories. One approach
is to develop more or less specific theories to deal with each particular phe-
nomenon of interest (e.g., communication). This approach has been underway
for a number of years, as is apparent from even a cursory view of the tables of
contents of textbooks in the field.
A second approach is to limit the sample frame of analysis by sorting indi-
viduals, groups, or organizations into categories for subsequent analysis. This
strategy of forming typologies or taxonomies l of organizations has also been
attempted a number of times in the past. A third strategy is to combine the first
two, developing limited range theories concerned with particular phenomena in
the context of limited classes of organizations.
An examination of early attempts to form typologies or taxonomies of orga-
nizations reveals that they have had several things in common. First, as noted by
McKelvey (1975), most of them were based on only one or two parameters (cf.
Etzioni, 1961; Blau and Scott, 1962; Woodward, 1965; Becker and Gordon,
1966; Katz and Kahn, 1966; Van Riper, 1966; Perrow, 1967; Thompson, 1967).
Second, they have tended to group entire organizations into categories on the
basis of interorganizational similarities and differences in terms of parameters
that are seen as characterizing whole organizations. For example, technology,
size, beneficiary served, and primary basis of authority and compliance have
been treated as organization-wide characteristics. Consequently, organizations
have been sorted so as to maximize differences between categories and to mini-
mize differences within categories on these dimensions. This strategy has char-
acterized much of the so-called "comparative organizations" approach (cf. Udy,
1965; Heydebrand, 1973). Third, previous schemes have focused on parameters
that vary across organizations at given points in time, but that have been assumed
to be constant over time. Use of these parameters reflects a strong structuralist
bias and has tended to preclude the possibility that a given whole organization
may belong to a particular class at one point in time but belong to another class
at some later point. Finally, estimates of central tendency, such as means or
modes, have been employed to characterize the organizational units to be sub-

grouped. Etzioni (1961) studied the modal form of authority; Woodward
(l965), Perrow (1967), and Thompson (1967) viewed the primary technology;
Becker and Gordon (l966) examined the degree to which resources are stored in
specific form, as opposed to general form, and the degree to which procedures are
self-determined, as opposed to organizationally determined. Van Riper (1966)
used the concept of the "relative strength of the control over individuals in the
name of the group."
In brief, although a few researchers, such as Haas et al. (1966), Pugh et al.
(1969a), and Pinto and Pinder (1972) have employed multiple parameters for
establishing empirical clusters of organizations, most previous attempts have
relied upon the use of modal estimates of one or two parameters that were
treated as characteristic of entire organizations, as well as somewhat constant
over time. Consequently, most organizational taxonomies have served little more
than simple descriptive purposes.
The success of future attempts to develop useful contingency and midrange
theories will depend heavily upon the techniques used for isolating subsets of
phenomena, contingencies, and circumstances within which limited descriptive
and normative statements will apply. Therefore, this article proposes some spe-
cific new guidelines for organizational taxonomizing that deliberately circum-
vent the problems of previous attempts, discusses the problem of selecting
subgrouping dimensions for the formation of a taxonomy of organizations, and
proposes one possible set of such dimensions.
We do not propose to subvert or replace the systems approach in organization
theory. In fact, many of the proposals that follow are consistent with the image
of organizations as open systems . .It is the futility of attempting to develop gen-
eral theory of any kind at the present state of maturity of organization theory
which provides the impetus for our remarks. Some organizations may be like
systems and therefore be viewed profitably from the systems perspective. But
we feel that the appropriateness of the systems metaphor is an empirical ques-
tion, and should not be accepted on the basis of faith as applicable to all organi-
zations. We favor middle range theorizing rather than general theorizing: If the
systems model assists that enterprise, it should be exploited accordingly.
GUIDELINES TOWARD A NEW TAXONOMy2
Employ Multiple Parameters
Parsimony is a desirable characteristic for any model or theory (Gibbs, 1972), as

long as it does not conceal more than it illuminates. Critical reviews, such as
those of Burns (1967), and Hall, Haas, and Johnson (1967) have suggested that
simple sub grouping schemes based on economical (rather than multidimensional)
models, tend to leave out a great deal, making it hard to sort organizations into
groups reliably, and consequently providing the researcher with little predictive
or explanatory assistance. Organizations are complex phenomena, and, as such,
it seems only reasonable that meaningful and useful subgrouping schemes incor-
porate commensurate levels of complexity (Weick, 1969). Therefore, the first
guideline we propose is that multiple parameters be employed to attempt to
capture the multidimensionality of organizational similarities and differences.
Study Organization Subunits
The second guideline we propose is that whole organizations be grouped, but

that the data used for determining category membership be collected from sub-
units of the organizations to be classified. For all but the smallest organizations,
there is considerable diversity between subsystems within organizations, and
these intraorganizational differences can be rich sources for understanding the
essential characteristics of organizations (Aldrich and Herker, 1977; Tushman,
1977).
Use Time-Sensitive Parameters
Many of the interesting phenomena that characterize organizational behavior

can be attributed to changes in important parameters and to the rate of change
of these parameters. The oft-noted distinction between structure and process
in organizations (cf. Weick, 1969; Hall, 1972; Gibson, Ivancevich, and Donnelly,
1976) provides justification for our suggestion that researchers focus on param-
eters that yield meaningful variances within organizations over time in a process-
oriented fashion. Here the suggestion is to focus attention on processes of
change, forming subgroups so that organizations grouped together are those that
have experienced common change or evolutionary processes. The study of
phyletics in biological systematics (Ross, 1974) represents the application of this
general approach. In organizational analysis, examples of such an approach in-
clude studying growth rates rather than size, as Kimberly (1976) suggests; study-
ing changes in leadership over time, as Gouldner (1954) and Grusky (1963, 1970)
have done; or studying changes in technology, as several researchers have done
in the past (Trist and Bamforth, 1951; Morse and Reimer, 1956; Susman, 1970;
Mann and Williams, 1972; Billings, Klimoski, and Breaugh, 1977). Bigelow
(1979) and McKelvey (1979) have discussed the notion of organizational evolu-
tion and phyletics.
Similarly, changes in organizational structure, changes in the demographic

mix of members, or changes in the distribution of power might also reveal inter-
esting internal dynamics that could lead to rich hypothesis generation. The liter-
ature on change processes (Alderfer, 1976) and the dynamics of resistance to
change (Lawrence, 1969), as well as the notions of unconfirmed expectancies,
violations of psychological contracts (Schein, 1970), and adaptation levels (Rel-
son, 1964), all suggest that commonalities in the experience of change might
provide useful bases for forming subgroups. In short, so many of the interesting
phenomena that characterize organizational behavior are attributable to change
and evolutionary processes that it would make more sense to seek middle range
theories pertaining to groups of organizations that share common change pro-
cesses than to study groups of organizations whose members merely share com-
mon static characteristics.
Subgroup on the Basis of Intraorganizational Variability
In addition to using multiple dimensions for subgrouping organizations and to

employing time-sensitive measures, we propose the use of measures of variance
or dispersion on those parameters chosen, rather than, or in addition to, statis-
tics reflecting averages, central tendencies, or modal characteristics. Strict reli-
ance on modes and means masks the richness attributable to intraorganizational
differences. In most organizations of any size, there is considerable internal vari-
ation in terms of most parameters of interest. For instance, most organizations
employ more than one type of technology, although one type often dominates.
Most organizations reveal internal differences in the responsiveness of their
reward structures, in the types of leadership styles used, in the types of climate
perceived by their members, in span of control and closeness of supervision, and
so on. Internal variation in terms of employee skill level and job complexity,
formalization of rules and procedures and general bureaucratization, turbulence
of subenvironments, and other important characteristics are also obvious in most
organizations. Thus, attempts to measure the average, the typical, and/or the
dominant level of any of these parameters for the sake of subgrouping overlook
these crucial internal differences and, in many cases, yield estimates that are, in
fact, unrepresentative of any of the parti~ular subunits of a given organization.
An example of this approach is to study the differences in leadership styles
among the subunits of an organization, since most large organizations tend to
have more than one formal leader. Variations between the styles leaders employ
can have important consequences for perceptions of distributive justice (Romans,
1961), equity (Adams, 1963) and other consequences of the social comparison
processes that occur in organizations (cf. Goodman, 1977). Another example is
to study the differences in the sizes or structures of subunits within an organiza-

tion, since differences in these parameters may be related to differences in
bureaucratization, opportunities for social rewards, power potential, promotion
opportunities, closeness of supervision, and other determinants of individual
behavior. A third example is a study of Weick's (1976) speculation that patterns
of degrees of coupling among subsystems may be more illuminating than modal,
overall estimates of the strength of average bonds inside a system. The lead times
required to coordinate subgroup A with subgroup B, for example, might be an
important aspect of the planning of both A and B for their respective relation-
ships with other subgroups inside or outside the same system. Things might roll
smoothly for A and B in such a case if the strength of ties they have with other
groups are similar. However, differences in the strength of ties among subgroups
could provide rich sources of insight into intraorganizational conflict, power
differences, promotion patterns, "communication problems," and other pro-
cesses. A fourth example is to investigate interunit differences in the demographic
makeup of members. Differences in age distribution can be relevant for differ-
ences in relative power, promotion opportunities, and overall prestige. Differ-
ences in skill level pose special problems of coordination, communication, and
power dispersion. A final example is represented by the work of Lawrence and
Lorsch (1969), who used differences between subunits in terms of the uncertainty
of their respective subenvironments as the major determinant of prescribed levels
of differentiation and integration of the structure of the organization. Also noted
in that study were interesting corresponding differences in time perspectives
and goal orientations.
The literature is rich with examples of the consequences of contrast and dif-
ference effects. For example, Dearborn and Simon (1958) have shown how
perceptions vary as a function of departmental affiliation, while l1eberman
(1956) has shown how changes in roles can affect changes in attitudes. More
recently, Johnston (1976) has shown how differences in seniority can account
for differences in climate perceptions, while both Johnston (1976) and Drexler
(1977) have discussed the possibility of multiple climates within a given orga-
nization. These studies suggest that a strategy for taxonomizing should employ,
as a primary basis for subgrouping, similarities among organizations in terms of
their internal variations along critical dimensions.
Operationalizing the Variance Concept
There are at least three ways in which internal differences can be operationalized
for use in a taxonomic effort, each of which could be useful for capturing the
sort of intracluster richness we expect from the subgrouping process:
1. grouping on the basis of similar variances among subunit averages on clus-

tering dimensions;
2. grouping on the basis of similar organization wide variances on clustering
dimensions; and
3. grouping on the basis of similar average variances within subunits on clus-
tering dimensions.
Depending upon the parameters in question and possibly upon the size of the
organizations to be grouped, anyone or any combination of these variance mea-
sures could serve in a profIle for clustering organizations.
The first tactic forces an aggregation or averaging of data at the subunit level
for subsequent interunit comparisons. An example of the first approach might
use technology as a clustering variable. For the sake of the illustration, we use
Thompson's (l967) typology of technologies. Organizations would be sorted
into clusters on the basis of similar intra organizational patterns of subsystem
technologies. Thus, one cluster might be constituted primarily of organiza-
tions featuring a particular blend of long-linked and mediating subunit tech-
nologies, while a second cluster features a particular blend of intensive and
mediating styles.
Notice that in some situations, this approach could result in the same sort of
richness-masking of which we have accused earlier taxonomic schemes. For ex-
ample, if it were applied to some potential clustering parameters, such as em-
ployee age, the "average" figures used to represent each subunit would mask the
richness that might be observed as a consequence of intrasubunit differences in
age. Consequently, this tactic for operationalizing the variance concept is prob-
ably most suited, as in the example above, for use with subunit characteristics
that apply almost universally to the members of a particular subunit but that
vary among the subunits of any particular organization. However, it would not
be necessary that only nominal level variables be operationalized in this fashion.
The second tactic for operationalizing intraorganizational differences avoids
the aggregation or combination of individual data into subunit scores, as in the
case described above. Instead, subunit boundaries would be disregarded, so that
variance measures on the parameter(s) used would be based upon total, system-
wide differences around the overall mean (or other measure of central tendency)
for the organization. This tactic would probably be most useful for operation-
alizing variance measures on individual-level dimensions that are equally relevant
to all parts of the organization, regardless of subunit. For instance, organization-
wide variation in seniority or salary might be operationalized in this fashion.
This second approach might be especially useful in subgrouping small, relatively
undifferentiated organizations.
The third tactic makes use of average variances. That is, organizations would
be grouped on the basis of their similarities with regard to the average amount
of variability they have within their respective subgroups on the parameter(s)
in question. This tactic might be employed to ope rationalize the variance con-
cept when intrasubunit variability in terms of a clustering parameter could be
expected to be a source of interesting outcomes. For example, the amount of
intragroup adherence to official rules or informal norms is a potentially inter-
esting parameter that, if used to form homogeneous clusters of organizations,
might facilitate some revealing analyses of intersubunit phenomena, such as
intergroup conflict. The richness would most easily follow if organizations were
grouped according to similarities of the average degree of variance in compliance
to these norms. For example, a cluster of organizations might be formed so
that they all featured a similar average level of subunit compliance to official
rules, such as a very high degree of compliance within most of their respective
subsystems.
In short, there at least three ways to operationalize a general strategy of
grouping organizations on the basis of similar internal differences. The particu-
lar tactic used would vary from one parameter to another, and all three tactics
might be employed on different parameters to yield the variability measures
that would serve as the basis of a particular organizational taxonomy.
An Example of Applied Taxonomy
In order to illustrate how subgroups based on the guidelines proposed in the

preceding sections might be formed and might assist theory development, a
hypothetical cluster of organizations that could emerge from this approach to
subgrouping is described in this section. The reader is cautioned that this exam-
ple is entirely fictitious and meant only to illustrate and assist the understanding
of the strategy outlined in the previous sections.
It is assumed in this example that a number of dimensions have been judged
as important for constructing the classification, and that the organizations to be
discussed have been sorted into one category on the basis of their similar profiles
on these dimensions. Other organizations not discussed in the example are as-
sumed to have been sorted into other categories. The parameters used to define
"similarity" are assumed to have been constant for all organizations in the
sample, and the assignment of organizations to classes is assumed to have been
conducted according to the guidelines presented above, and with a view to
achieving intraclass homogeneity.
For the example, the dimensions used to define similarity are rate of growth
in terms of number of people on the payroll, the frequency of change in execu-
tives at the institutional level over the past ten years, and the variance among
subunits in terms of the degree of uncertainty of their respective major sub-

environments, as perceived by subunit managers. It is assumed that the orga-
nizations in this particular hypothetical group have all recently undergone
considerable growth and, with it, high rates of leadership succession, as compared
to organizations sorted into other categories. (Note that we have used rates of
growth rather than size, and rates of succession rather than current leadership
styles.) It is also assumed that these organizations feature wide variation in the
average skill levels of their participants, as well as a broad age mix among them.
Further, it is assumed that organizations in this group feature relative hetero-
geneity among the dominant subenvironments they face. (Note the use of inter-
nal subunit differences and variations between subunits rather than central
tendency estimates based on the distributions of the whole organizations.) In
such a subgroup, we might find a small electronics firm, a conglomerate forest
products firm, a private postsecondary educational institution, and a volunteer
center. Analysis of the behavior of subunits of organizations within such a
cluster should reveal many interesting commonalities among them. These be-
haviors would tend to occur in patterns, because of the intraclass homogeneity
created by the clustering process. For example, the rapid growth and succession
they have all witnessed would lead to the predictable bureaucratization that
often accompanies these processes. But the imposition of widespread rules and
procedures might cause serious problems for subunits that require organic man-
agerial styles in order to cope with their rapidly changing subenvironments,
whereas the reliance on bureaucracy as a control device is well suited to the
remaining subunits. As a result, conflict over the observance of rules based on
the need for equal treatment would eventuate. The internal differences among
their employees in terms of age and skill mix, especially if differences on these
variables overlapped with differences in bureaucratic vs. organic managerial
needs, would result in difficulties in coordination and integration, tending to
exacerbate the subsequent conflict.
Presumably, many groups of organizations would emerge from this sort of
investigation, depending upon the number and types of parameters used, the
variety of organizations sampled, and the statistical criteria adopted for deter-
mining membership in a group. Once formed and found to be statistically
reliable, theory and research concerned with relationships among various in-
dependent and dependent variables might then proceed within groups.
We do not see the process of theorizing within clusters as being much differ-
ent from the usual strategies of the field, other than in terms of the generality of
the propositions developed and tested (pinder and Moore, 1979a). However,
the imagery appropriate for describing organizations in one class might be
entirely different from that used to represent other clusters. Moreover, the
problems addressed, the methodologies employed, and, subsequently, the de-
General Theory of Organizations
(pertaining to all types of
organizatIOns)
Midrange theory of higher order

for Organization type-alpha
Theory for Theory for Theory for Theory for Theory for Theory for Theory for
Organization 1 OrganizatIOn 2 Organization 3 Organization 4 Organization 5 Organization 6 Organization 7
Figure 1. Gradual Ascension to General Theory through the Integration of Progressively More Abstract Bodies of Midrange Theories
......
\0
-...J
scriptive and normative theories developed, would also be different from those
employed in the context of other clusters of organizations. Our expectation is
that models of organizational behavior might fare somewhat better within sub-
groups, simply because so much of the usual "error variance" will have been
controlled by the grouping process. Moreover, the rich insights for theory con-
struction that would ensue would be attributable to the fact that the groups were
formed on the basis of common internal variation and common processes, as
opposed to common internal similarities in terms of modal, static characteristics
Eventually, it might be possible to develop higher order theories based on a
more complete understanding of the organizational relationships within groups,
and the detection of similarities of phenomena and relationships across groups.
In the end, general theory might be possible, but, if so, its emergence would
be based on the valid synthesis of bodies of lower order, midrange theories,
which reveal meaningful similarities and which provide a good understanding of
the organizational phenomena within their respective domains. As such interclus-
ter similarities emerge and prove reliable, the total number of paradigms in the
field may also be reduced. 3 This long range process is illustrated in the Figure.
THE PROBLEM OF SELECTING SUBGROUPING PARAMETERS
Thus far, our discussion has begged the question of specifically which parameters
should be used for sorting organizations into groups, as well as the more general
question of what criteria might be employed for making this selection. Here lies
the major challenge for organization theorists who wish to pursue the taxonomic
approach to theory development.
Core Dimensions and Species Concepts
Examination of the efforts of scientists in many fields in which taxonomies are

employed reveals the universality of the problems of determining the basis for
classifying individuals into species and higher order groups. It seems desirable to
isolate at least one ultimate criterion that permits the unequivocal classification
of individuals. How this criterion is identified, however, is problematic. Some
taxonomists favor a strictly atheoretic approach wherein simple phenotypic
similarity guides the selection of a number of parameters. Individuals are then
assessed on these "obvious" dimensions and statistically sorted into categories.
It is assumed that such an empirical approach will ultimately provide insights
into the reasons for diversity among individuals, but that the initial selection of
sorting parameters must be as free as possible from the influence of current
theory. Consequently, many characteristics are usually assessed, and complex

multivariate cluster analysis techniques are employed on the data to identify
classes (Sneath and Sokal, 1973).
Other taxonomists favor the a priori adoption of some type of species con-
cept and the subsequent use of this concept in forming a classificatory scheme.
Research in biological systematics since the time of Darwin's seminal work, for
example, has seen a succession of criteria for the delineation of "species." First,
"character discontinuity" (Ross, 1974), or a criterion of apparent physical dif-
ferences, served as the major means of delineating one species from another.
Difficulties arose, however, when it was frequently found that individuals from
presumably different species shared more important characteristics than mem-
bers from what were believed to be a common species. Specifically, geneticists
continually encountered instances in which plants or animals that were thought
to be of a common species could not produce progeny together, whereas indi-
viduals from what were thought to be different species were able to procreate
successfully. Consequently, a criterion of "genetic compatibility" became more
widely accepted. This criterion seemed viable until botanists periodically en-
countered hybrids that emanated from "species" that seemed too dissimilar
for procreation to seem possible. Zoologists also found sufficient numbers of
cases of interspecific crossings to compromise the value of the criterion of
genetic compatibility. Recently, a third criterion, which classifies individuals
according to commonalities of their evolutionary lineages, has been proposed
(Ross, 1974).
In the social sciences, Leach (1961) has reported on a similar problem in
determining classification criteria in social anthropology, in which, for example,
societies have been classified according to whether they are patrilineal or matri-
lineal. He raises the possibility that these categories, while they provide an ob-
vious and handy basis for classification, may be essentially useless for illuminating
important or interesting underlying sociological differences between societies,
in the same way that "the creation of a class [of] blue butterflies is irrelevant
for the understanding of the anatomical structure of lepidoptera" (p. 4). Many
previous organizational taxonomists fell into a similar sort of trap by making use
of clustering dimensions that seemed obvious and handy, but that did not facili-
tate theory construction for the understanding of species.
Little has been done to forward concrete suggestions regarding specific core
criteria for organizations that are comparable to those used in biological system-
atics. On the other hand, a number of authors have addressed the general issues
related to the problem. Hempel (1965) has explored some issues in taxonomy
construction and discussed how they related to the classification of mental dis-
orders. Among the guidelines he proposed that seem germane to our purpose are
the following: First, the parameters used and the measures employed to assess
them must be universally understood, facilitating inter-rater agreement. With

regard to this suggestion, Sells (1964) outlined a design for a taxonomic approach
to the study of organizations that utilizes three broad categories of variables
that determine organizational behavior: the characteristics of the organization
itself, such as goals, tasks, group structure, etc.; the characteristics of the physi-
cal environment; and the characteristics of the social environment. Sells' strategy
seeks a taxonomy of organizations that uses the differences found in multivariate
models, employing composites of the above variables as predictors of dependent
variables. A major contribution of Sells' work is the trichotomy of parameters
he listed, and his suggestion, which we adopt, that parameters for subgrouping
be drawn from several different categories rather than from one category exclu-
sively, since no single level of analysis in present theory has proven to be the
exclusive locus of the most important independent or dependent variables. At
the very least, parameters must be chosen that are relevant to all organizations
in the population to be subgrouped.
Hempel's second suggestion was that, as much as pOSSible, valuational over-
tones should be excluded from the assessment of a unit for classification. That
is, parameters with poles that require no interpretation as to being either good
or bad should be used. This point has some interesting implications if the search
for relevant sub grouping parameters leads, by whatever criteria used, to the
adoption of measures of what we would otherwise call dimensions of organiza-
tional "effectiveness." It is argued below that effectiveness measures that reflect
what we call the "characteristic adaptation style" of particular organizations
may provide a very useful type of dimension for subgrouping, so we reject
Hempel's second point in the context of organizational taxonomizing.
Third, Hempel suggested that the parameters selected should facilitate worth-
while comparisons between groups and foster the real understanding of what the
units have in common, leading to plausible theory construction within each
group. In terms of the model shown in the Figure, this would be necessary be-
fore theories of higher levels of generality could emerge from syntheses of lower
level midrange theories.
Fourth, the sorting scheme should permit efficient sorting of all the units in
the universe in question into discrete categories, minimizing the number of cases
that might fit more than one category. This issue, parenthetically, was one of
the major bases for the criticisms raised to the schemes of Etzioni (1961) and
Blau and Scott (l962) by Hall et al. (l967). Hempel noted that many scientific
phenomena are so complex that neat sorting into discrete categories is often very
difficult. Instead, he proposed the development of schemes that permit degrees
of similarity with or membership in a given category, thereby obviating the prob-
lem of "borderline" cases. Sub grouping techniques employing multiple dimen-
sions help to circumvent this problem somewhat, because, although a particular
unit might appear to fit more than one category based on only a few dimensions,
the use of additional parameters assists in a decision as to which category is
most appropriate.
Further, Hempel believed that the parameters used should lend themselves
to valid and meaningful generalizations and thereby facilitate prediction for new
cases that can later be sorted into one class or another on the basis of their de-
grees of similarity with each of the previously formed classes.
Finally, Hempel noted that concepts such as "average" and "typical" ignore
problems of dispersion and interunit differences. Therefore, statistical methods
that reflect averages as well as the representativeness of the averages should be
used. Much more emphasis, however, needs to be placed on dispersion and pro-
cess than was proposed by Hempel. McKelvey (1975) has recently provided
some specific suggestions for the empirical development of taxonomies of orga-
nizations. His guidelines follow closely a general strategy for careful statistical
inference, employing random sampling procedures for the selection of organiza-
tions and parameters to be studied, as well as of observers to provide data on the
sub grouping dimensions.
While we agree with most of McKelvey's suggestions, we take issue with two
of his points. The first concerns the problem of the role played by prior sub-
grouping schemes and parameters. One of McKelvey's directives is to avoid as
much as possible being influenced by previous approaches. Herein lies a major
problem. Whether a taxonomist opts for an empirical approach or one explicitly
guided by a specie concept, it is difficult, if not impossible, to completely avoid
the influence of extant theory when selecting sub grouping parameters. The
empiricist tends to rely on obvious phenotypic characteristics, but exactly which
dimensions appear obvious to him will be heavily influenced by his theoretical
training and predispositions. Taxonomizing with the deliberate use of a specie
concept, of course, also entails a strong influence by extant theory and research.
The influence of previous thinking is virtually impossible to avoid.
Moreover, there tends to be a tautolOgical relationship between prevailing
theory and scientific classification systems. Theories attempt to explain how and
why given classes of phenomena occur; they also simultaneously serve to delin-
eate boundaries between those phenomena that belong to a class and those that
do not. Kuhn's (1970) recounting of how Dalton reconciled a number of anom-
alies in the behavior of what were previously thought to be chemical compounds
is an example. Essentially, the problem was that a number of presumed "com-
pounds" did not combine according to the prevailing law of fixed proportions.
Two French chemists, Proust and Berthollet, each marshalled considerable
evidence to debate the point of whether or not all chemical reactions occurred in
fixed proportions. Proust claimed they did; Berthollet argued that many chemi-
cal reactions did not result in fixed proportions. The two chemists talked past
each other until Dalton provided a taxonomic resolution. Basically, all that
Dalton did was to designate those reactions that occur in fixed proportion as
"chemical"; those that do not were arbitrarily considered to be something other
than chemical. The point is that prevailing theories and paradigms delimit the
variety of bases or parameters that might be used to form categories of phe-
nomena. Dalton managed to avoid the tyranny of the prevailing paradigm which
held that forces of mutual affinity were responsible for the formation of com-
pounds largely because he was not trained as a chemist.
In the context of organizational taxonomizing, the difficulty of avoiding
extant theory is apparent. Most of us who want to attempt to develop a taxon-
omy have been educated within the confines of one or two prevailing paradigms.
Consequently, considerable difficulty is encountered in preventing a self-
perpetuating cycle wherein, for example, systems concepts are operationalized,
organizations are measured and sorted, and pseudo-systems theories ultimately
emerge within categories. Systems theorists would be quite comfortable with
this outcome because of their belief in the Ubiquity and inevitablility of systems.
Others who are less committed to the systems model, however, while they may
feel that the familiar systems imagery will lead to the same sort of theorizing as
before, may fmd that middle range theorizing within organizational categories is
easier than before because within each class certain systems-oriented qualities
will be predominant, while others are less pronounced. The systems paradigm
(or any other paradigm) need not necessarily prevent useful midrange theory
construction within classes of organizations. To the extent that not all organiza-
tions feature fully all of the characteristics of general systems, differences among
groups and homogeneity within groups in terms of "systemness" of constituent
organizations should facilitate the emergence of midrange theories, each of
which is more useful than any single theory based on a general, overarching
image pertaining to all organizations in the universe.
The second point of McKelvey's with which we take issue is his view that the
parameters used in sub grouping should receive equal weights. We caution that
the real weight of a parameter in a subgrouping scheme will, in part, be deter-
mined by the degree of variability that measures of that parameter detect in the
sample to be subgrouped. A parameter with apparent double weighting, as deter-
mined by some coefficient, will, in effect, have no bearing on the final determi-
nation of categories (i.e. will have zero weight) if the "scores" of all of the units
on that measure are equal, or nearly equal. Further, to the extent that parameters
covary in the population, their use in the subgrouping analysis will, in a sense,
result in unequal weights for any underlying construct that causes the empirical
parameters to covary. This is not necessarily bad if the actual weighting arrange-
ment reflects the conceptual basis for subgrouping intended by the researcher.
An issue that was not explicitly raised by either Hempel or McKelvey pertains
to whether a multivariate classification scheme should be "compensatory," or

whether multiple criteria should be met at minimum levels to justify the inclu-
sion of a unit in a given classification (cf. Dvorak, 1956). Some statistical algo-
rithms permit the identification of subgroups using "distance" measures, such as
those employed in profile analysis (Cronbach and GIeser, 1953), which make
it possible for large amounts of one attribute to offset small amounts of another,
yielding an overall index of interunit similarity. With such an algorithm, for
example, extra units of the size dimension might offset a low profile score for
the formalization dimension, so that a unit of large size and low formalization
might be sorted into the same class as a small unit with a high degree of formal-
ization. Pinto and Pinder (1972) have demonstrated the use of such a technique.
The utility of this type of sorting strategy might be limited, especially if the
compensating variables tend to be empirically independent. Therefore, a classifi-
cation scheme may be of more value if all units in a given category meet certain
minimum degrees on the various subgrouping dimensions. However, if these cri-
teria are made too stringent, the risk is run of having too few cases qualify for a
given category.
Haas et al. (1966) reported one of the earliest multivariate empirical attempts
to form a taxonomy of organizations. They observed that it is desirable to de-
velop taxonomies in a hierarchical fashion, ranging from small groups that possess
a great deal in common, to groups that are much broader and in which the mem-
bers of a given class possess comparatively less in common.
Of special interest to our guidelines are comments made by Haas et al. (1966:
178) regarding the fact that organizations change:
The individual units (organisms) being classified in a zoological taxonomy
may undergo marked change in attributes during the life cycle. But the zool-
ogist has discovered that the changes follow a pattern and, therefore, he is not
tempted to shift a particular specimen from category to category as it changes
over time. Indeed the patterns of such changes in the life cycle are utilized as
attributes which define the various classes.
The organizational analyst, however, is apparently dealing with less stable
cases. Organizations also may change drastically over time. But if a class of
organizations has a patterned and, therefore, predictable sequence of changes,
it is yet to be discovered. Despite all that has been said about the tendency
for organizations to resist change, it is manifestly clear that some organiza-
tions change so markedly that it seems entirely appropriate to assert that they
are now qualitatively different social entities from what they were earlier. To
the extent that such changes include alteration of significant organizational
attributes (i.e., those which are definitive for taxonomic categories), the orga-
nization would need to be relocated in the taxonomic system.
In effect, Haas et al. seemed to view the possibility of change in organizations
as a problem for the effort of classification. On the contrary, we suggest that

commonalities of change be used as a major basis for the development of taxon-
omies of organizations and propose further that researchers and practitioners
who wish to use taxonomies of organizations expect the membership of given
organizations in classes to be more or less fluid and subject to change.
Pugh and his colleagues (1969a) took a unique approach to organizational
taxonomy. Instead of taxonomizing organizations, per se, they developed a
"taxonomy of organizational structures," using three basic dimensions of
structure uncovered in their previous work (Pugh et al., 1968, 1969b). These
three orthogonal dimensions, structuring of activities, concentration of author-
ity, and line control of work flow, were used to identify natural clusterings
among the 52 organizations included in their sample. Post hoc examination of
the organizations in their clusters revealed some interesting intracluster similar-
ities, which eventually led to a cubic model with seven pseudo-ideal forms of
organization. One aspect of this work with particular relevance to this paper is
the post hoc developmental model they formulated to explain differences
between various of the clusters they identified. Although their design was cross-
sectional, Pugh et al. (l969a) hypothesized a developmental sequence of bureau-
cracy, in which increases in the structuring of activities and decreases in line
control of authority are related to increases in size and the passage of time. The
relevance of their model lies in the fact that they developed a tridimensional
taxonomy that came close to revealing the sort of "underlying core dimensions"
Hempel (1965) said are needed. In this case, the core dimensions were simply
age and growth. Different types of structures were seen as being related to dif-
ferent stages in the developmental process. This observation is consistent with
our proposal that organizational taxonomists use similarities among organiza-
tions in terms of their internal change processes as a basis for clustering.
McKelvey (1978, 1979) has discussed the idea of species concepts in general,
and has examined the relevance of the notion for organizational systematics.
Drawing on the work of biological taxonomists, he has identified four major
approaches to classification: essentialism, nominalism, empiricism and numerical
taxonomy, and evolution and phyletics. McKelvey rejects essentialism (the view
that objects and organisms have hidden underlying realities that generate the
observed differences among them) on the basis of his view that organizations are
not totally analyzable entities. Likewise, he rejects nominalism (the view that all
entities are unique and cannot be sorted into homogeneous groupings). Instead,
McKelvey favors an approach to organizational classification that combines
evolutionism and empirical approaches. Evolutionism attempts to explain the
origins of the groupings to be classified and thereby provides some guidance in
the selection of subgrouping dimensions. But the inductive, a priori classification
schemes resulting from the evolutionist strategy require deductive testing through
the application of the more objective methods of numerical taxonomy (Sneath

and Sokal, 1973).
The specific specie concept advocated by McKelvey (1978, 1979) is what he
calls the "core technology" of an organization.This includes both the technology
used in the organization's primary work place, following Davis and Engelstad
(1966) and Miller and Rice (1967), as well as the managerial technologies asso-
ciated with managing the primary work place. McKelvey argues that his technol-
ogy specie concept satisfies three crucial requirements of such concepts (Mayr,
1969). First, it provides insight into the ecological origins of organizational
types. Second, it reveals the sorts of mechanisms that serve to isolate types of
organizations from one another, perpetuating interorganizational variability.
Finally, it can be seen that core technologies are passed from generation to suc-
ceeding generation, thereby perpetuating the species. McKelvey has made some
important contributions to organizational systematics, both in the guidelines he
has proposed (1975) for empirical approaches, as well as in his proposal for an
organizational specie concept.
As an alternative to, or modification of McKelvey's technology criterion, we
propose the use of a multidimensional criterion, reflecting what we call the
"Characteristic Adaptation Style" (CAS) of the organization, that is, the set of
managerial behaviors typically employed by the organization to sustain its exis-
tence and to be effective. Selection of this criterion is based on the tautological
observation that all organizations at a point in time have at least one thing in
common: they have adapted (more or less well) to their respective environments.
Our proposed criterion, therefore, has universal relevance to all existing organiza-
tions, thereby satisfying the requirement of universality. Moreover, the criterion
of adaptation style provides an important and very meaningful sorting mecha-
nism, since it deals with the very basis of existence of organizations. We recog-
nize that measuring this construct is likely to require the use of dimensions
closely related to what we would presently call "organizational effectiveness."
Thus, for example, the seven "performance objectives" discussed by Gross
(1965) might guide the development of a multifaceted criterion of adaptation
style to serve the purpose of subgrouping. Gross' criteria are universal require-
ments of all organizations that seek to survive. Each of the dimensions, securing
inputs, forestalling entropy, behaving rationally, reinvesting in the system, satis-
fying interests, producing outputs, and making efficient use of inputs, is also, in
tum, multidimensional, insofar as there are numerous ways in which organiza-
tions attempt to accomplish it (Gross, 1978).
The selection of conceptual dimensions such as Gross' does not avoid the
influence of previous knowledge about organizations. In fact, the use of these
particular dimensions involves reliance upon concepts that are very much systems-
oriented by nature. Our goal is not to usurp or supplant the systems approach
through taxonomy. We merely wish to reduce the universalistic descriptions and

prescriptions that emanate from any general framework. The explicit adoption
of a systems-rooted model such as Gross' may be beneficial for the very reason
that we are openly acknowledging the difficulty, or impossibility, of avoiding
extant knowledge; we are using it for what it is worth and are aware of its
potential influences. It is entirely possible, for instance, that clusters of orga-
nizations that feature low- or zero-scale scores on some or all of the systemic
dimensions generated from the Gross model might emerge from an analysis of
the sort we are proposing. In such a case, theory and research might lead in very
asystemic directions. On the other hand, some clusters of organizations might be
close to what could presently be called "ideal type" systems: organizations
featuring in full-blown fashion all the systemic characteristics commonly attri-
buted to organizations (cf. Katz and Kahn, 1966). One way to control the influ-
ence of the dominant paradigm is to use it, but to take "systemness" as a set of
variables rather than as an immutable constant.
Considerably more work is needed to isolate a core criterion (or core criteria)
that is essentially meaningful, that avoids problems of the "blue butterfly"
type, that is essential to all extant and possible organizations, that can be opera-
tionalized and reliably scaled, and that meets the other major guidelines dis-
cussed by Hempel (1965) and others. Such a criterion will stand a better chance
of meeting all of those guidelines and of guiding theory construction if it can
be dimensionalized to reflect the vastness of interorganizational differences.
Further, once it is identified, dimensionalized, and scaled, the criterion (or
criteria) can be best employed if indices of similarity of variability and process
provide the mathematical bases for grouping. It is hoped that this discussion
will provoke more thought on the matter and will generate refmements and/or
alternative strategies and dimensions for taxonomizing organizations. We detect
a movement toward middle range theorizing in organization theory (Weick,
1974; Behling, 1978; Pinder and Moore, 1979a) and feel that a resurrection of
the scientific strategies of taxonomy can help that cause.
NOTES
1. Following the distinction made by Pugh et al. (1969b), we use the term "typology" to
refer to a priori classification schemes, such as those developed by Etzioni (1961), Blau and
Scott (1962), and others (see Cap low , 1964 for a review of anum ber of much earlier typol-
ogies), and the term "taxonomy" to refer to empirically derived, multivariate classificatory
schemes. For descriptive purposes, many of the early subgrouping schemes served well. For
the sake of theory advancement, however, most have been of limited value, because of their
failure to provide explanatory or predictive power.
2. In the guidelines that follow, it is crucial to recognize two important points, First, our
proposed strategy requires the adoption of metaphysical and epistemological assumptions

that organizations have real, objective existences and identifillble boundaries independent of
the minds of observers, actors, and scientists. Explicit recognition of this assumption is
important in view of the current movement by some organization theorists, such as Pondy
and Boje (1976) and Weick (1977), in favor of a metaphysics that considers organizations as
having existence only in the minds of actors and observers. A second feature of our pro-
posed strategy is its reliance upon the formal structures of organizations as the basis for the
collection of data for the subsequent grouping process. We recognize that informal struc-
tures are major sources of organizational dynamics. But to be tractable, strategies of taxono-
mizing must, initially, rely on formal structures to guide the collection and aggregation of
data. Although the formal structure is seldom a perfect surrogate for the informal, it is
hoped that insights into the latter might be facilitated and improved by the conceptual and
empirical work that would follow the classification of organizations into clusters.
3. In the limit of course, it would follow that higher order reconciliations between clus-
ters of organizations might eventually lead to the emergence of a single paradigm in the field.
However, the present authors have argued elsewhere (Pinder and Moore, 1979b) that this is
not likely to occur in the foreseeable future because of a number of factors pertaining to the
study of organizations that will continue to cause different observers to hold different images
of organizations.
REFERENCES
Adams, J. Toward an understanding of inequity. Journal of Abnormal and Social

Psychology, 1963,67, 422-436.
Alderfer, C. Change processes in organization. In M. D. Dunnette (Ed.), Hand-
book of Industrial and Organizational Psychology, Chicago: Rand McNally,
1976.
Aldrich, H.E., and Herker, D. Boundary spanning roles and organization struc-
ture. Academy of Management Review, 1977,2, 217-230.
Becker, S. W., and Gordon, G. An entrepreneurial theory of formal organization,
part 1. Administrative Science Quarterly, 1966, 11, 315-344.
Behling, O. Some problems in the philosophy of science of organizations. Acad-
emy of Management Review, 1978,3, 193-201.
Bigelow, J. Evolution and middle range theories: Toward a matrix of organiza-
tional modes. In Larry F. Moore and Craig C. Pinder (Eds.), Middle Range
Theory and the Study of Organizations, Martinus Nijhoff, 1979.
Billings, R.S., K1imoski, R.J., and Breaugh, J.A. The impact of a change in
technology on job characteristics: A quasi-experiment. Administrative Science
Quarterly, 1977,22, 318-339.
B1au, P.M., and Scott, W.R. Formal Organizations. San Francisco: Chandler,
1962.
Burack, E.H. Aspects of the state of the art in systems theory: Accomplish-
ments and challenges. Organization and Administrative Sciences, 1975, 6,
101-104.
Burns, T. The comparative study of organizations. In Victor Vroom (Ed.),

Methods of Organizational Research. Pittsburgh: University of Pittsburgh
Press, 1967.
Caplow, T. Principles of Organization. N.Y.: Harcourt, Brace Jovanovich, 1964.
Cronbach, L.J., and GIeser, G.C. Assessing similarity between profiles. Psycho-
logical Buf/etin, 1953,50,456-473.
Davis, L.E., and Engelstad, P.H. Unit operations in socio-technical systems:
Analysis and design. Working Paper No. T-894, London: Tavistock, 1966.
Dearborn, D.C., and Simon, H.A. Selective perception: A note on the depart-
mental identification of executives. Sociometry, 1958,21, 140-144.
Drexler, J.A. Organizational climate: Its homogeneity within organization. Jour-
nal of Applied Psychology, 1977,62, 38-42.
Dvorak, B.J. Advantages of the multiple cut-off method. Personnel Psychology,
1956,9,45-47.
Etzioni, A. A Comparative Analysis of Complex Organizations. New York: Free
Press, 196 1.
Fiedler, F. A Theory of Leadership Effectiveness. New York: McGraw-Hill, 1967.
Friedman, M. Essays in Positive Economics. Chicago: University Press, 1953.
Gibbs, J. Sociological Theory Construction. Hinsdale, Ill.: Dryden Press, 1972.
Gibson, J. L., Ivancevich, J. M., and Donnelly, J. M. Organizations (rev. ed.).
Dallas: Business Publications, 1976.
Goodman, P. Social comparison processes in organizations. In Barry Staw and
Gerald R. Salancik (Eds.),New Direction in Organizational Behavior. Chicago:
St. Clair, 1977.
Gouldner, A. Patterns of Industrial Bureaucracy. New York: Free Press, 1954.
Gross, B. M. What are your organization's objectives? Human Relations, 1965,
18, 195-215.
- - . Personal correspondence with authors, 1978.
Grusky, O. Managerial succession and organizational effectiveness. American
Journal of Sociology, 1963,69,21-31.
- - . The effects of succession: A comparative study of military and business
organization. In Oscar Grusky and G. Miller (Eds.), The Sociology of Orga-
nization. New York: Free Press, 1970.
Haas, J.E., Hall, R.G., and Johnson, N.J. Toward an empirically derived taxon-
omy of organizations. In R. V. Bowers (Ed.), Studies on Behavior in Organiza-
tions. Athens, Ga.: University Press, 1966, 157-180.
Hall, R.H., Haas, J. E., and Johnson, N.J. An examination of the Blau-Scott and
Etzioni typologies. Administrative Science Quarterly, 1967,12, 118-139.
Hall, R.H. Organizations: Structure and Process. Englewood Cliffs, N.J., 1972.
Helson, H. Adaptation Level Theory. New York: Harper & Row, 1964.
Hempel, C.G. Aspects of Scientific Explanation. New York: Free Press, 1965.
Heydebrand, W. V. (Ed.). Comparative Organizations. Englewood Cliffs, N.J.:
Prentice-Hall, 1973.
Homans, G.C. Social Behavior: Its Elementary Forms. 1961.

Johnston, H.R. A new conceptualization of source of organizational climate.
Administrative Science Quarterly, 1976,21,95-103.
Kast, F., and Rosenzweig, J.B. General systems theory: Applications for orga-
nization and management. Academy of Management Journal, 1972,15,447-
467.
Katz, D., and Kahn, R. The Social Psychology of Organizations. New York:
1966.
Kimberly, J. R. Organizational size and the structuralist perspective: A review,
critique, and proposal. Administrative Science quarterly, 1976,21, 571-597.
Kuhn, T. The Structure of Scientific Revolutions. Chicago: University Press,
1970.
Lawrence, P. R. How to deal with resistance to change. Harvard Business Review,
1969,47,4-13.
Lawrence, P. R., and Lorsch, J. Organization and Environment. Homewood, Ill.:
Irwin, 1969.
Leach, E. R. Rethinking Anthropology. New York: Humanities Press, 1961.
Lieberman, S. The effects of changes in roles on the attitudes of role occupants.
Human Relations, 1956,9,385-402.
Mann, F.C., and Williams, L.K. Organizational impact of white collar automa-
tion. In L.E. Davis and J.e. Taylor (Eds.), Design of Jobs. Harmondsworth,
Middlesex, England: Penguin, 1972,83-90.
Mayr, E. Principles of Systematic Zoology. New York: McGraw-Hill, 1969.
McKelvey, B. Guidelines for the empirical classification of organizations. Admin-
istrative Science Quarterly, 1975,20,509-525.
- - . Organizational Systematics. Book in progress, 1978(a).
- - . Organizational systematics: Taxonomic lessons from biology. Management
Science, 1978,24, l428-1440(b).
and realizations. Organization and Administrative Sciences, 1975,6,3-10.
Merton, R. Social Theory and Social Structure. Glencoe, Ill.: Free Press, 1968.
Milburn, T.W., Negandhi, A.R., and Robey, D. Evaluating systems theory con-
cepts: Comments. Organization and Administrative Sciences, 1975,6, 11-13.
Miller, E.J., and Rice, A.K. Systems of Organization. London: Tavistock, 1967.
Morse, N.C., and Reimer, E. The experimental change of a major organizational
variable. Journal of Abnormal and Social Psychology, 1956,52, 120-129.
Perrow, C. A framework for the comparative analysis of organizations. American
Sociological Review, 1967, 32, 194-208.
Pinder, C. C. Facets of a movement toward middle range theory in organizational
behavior. Unpublished working paper, Faculty of Commerce and Business
Administration, University of British Columbia, 1978.
Pinder, C. C., and Moore, L.F. (Eds.). Middle Range Theory and the Study of
Organizations. Martinus Nijhoff, 1979(a).
- - . The inevitability of multiple paradigms and the resultant need for middle
range analysis in organization theory. In Larry F. Moore and Craig C. Pinder
(Eds.), Middle Range Theory and the Study of Organizations. 1979(b).
Pinto, P. R., and Pinder, C. C. A cluster-analytic approach to the study of orga-
nizations. Organizational Behavior and Human Performance, 1972, 8, 408-
422.
Pondy, L.R., and Boje, D.M. Bringing mind back in: Paradigm development as
a frontier problem in organization theory. Unpublished Manuscript, Univer-
sity of Illinois, Department of Business Administration, 1976.
Pondy, L. R., and Mitroff, I. Beyond open system models of organization. In
B.M. Staw and L. L. Cummings (Eds.), Research in Organizational Behavior.
Greenwich, Ct.: JAI. 1979.
Pugh, D. S., Hickson, D. J., and Hinings, C. R. An empirical taxonomy of work
organizations. Administrative Science Quarterly, 1969(a),14, 115-126.
Pugh, D.S., Hickson, D.J., Hinings, C.R., and Turner, C. Dimensions of organiza-
tional structure. Administrative Science Quarterly, 1968, 13, 65-105.
- - . The context of organization structures. Administrative Science Quarterly,
1969(b), 14,91-114.
Ritzer, G. Sociology: A multiple paradigm science. The American Sociologist,
1975,10, 156-167.
Ross, H.H.Biological Systematics. Reading, Ma.: Addison-Wesley, 1974.
Schein, E. Organizational Psychology (2nd ed.). Englewood Cliffs, N.J.: Prentice-
Hall, 1970.
Scott, W. G. Organization theory: An overview and an appraisal. Academy of
Management Journal, 1961,4, 7-26.
Sells, S.B. Toward a taxonomy of organizations. In W.W. Cooper, H.J. Leavitt,
and M. W. Shelley (Eds.), New Perspectives in Organizational Research. New
York: Wiley, 1964, 515-532.
Sneath, P.H.A., and Sokal, R.R. Numerical Taxonomy. San Francisco: W.G.
Freeman, 1973.
Susman, G. I. The impact of automation on work group autonomy and task
specialization. Human Relations, 1970,23, 567-577.
Thayer, F. General system(s) theory: The promise that could not be kept.
Academy of Management Journal, 1972,15,481-494.
Trist, E. L., and Bamforth, K. W. Some social and psychological consequences
of the longwall method of coal-getting. Human Relations, 1951,4, 3-38.
Tushman, M. L. A political approach to organizations: A review and rationale.
A cademy of Management Review, 1977,2, 206-216.
Udy, S.H. The comparative analysis of organizations. In James G. March (Ed.),
Handbook of Organizations. Chicago: Rand McNally, 1965,678-709.
Van Riper, P.P. Organizations: Basic issues and a proposed typology. In R. V.
Bowers, (Ed.), Studies on Behavior in Organizations. Athens, Ga.: Univer-
sity of Georgia Press, 1966, 1-12.
Vroom, V., and Yetton, P. Leadership and Decision Making. Pittsburgh: Uni-
versity of Pittsburgh Press, 1973.
Weick, K.E. The Social Psychology of Organizing. Reading, Ma.: Addison-
Wesley, 1969.
- - . Middle range theories of social systems. Behavioral Science, 1974,19,357-
367.
ence Quarterly, 1976,21, 1-19.
- - . Enactment Processes in organizations. In Barry M. Staw and Gerald R.
Salancik (Eds.), New Directions in Organizational Behavior. Chicago: St.
Clair Press, 1977, 267-300.
Woodward, J. Industrial Organization. Glasgow: Oxford University Press, 1965.
Zaleznik, A. Book review of The Social Psychology of Organizations. Adminis-
trative Science Quarterly, 1967,2, 682-691.
17 FUNCTIONALISM AS A BASE
FOR MIDRANGE THEORY
IN ORGANIZATIONAL
BEHAVIOR/THEORY
Understanding anything as complex as a large, formal organization is obviously

difficult. It is a task which, subjectively, appears to be progressing unevenly in
Organizational Behavior/Theory, the discipline charged with investigating organi-
zations and the groups and persons who make them up. In this paper we consider
functionally based analyses as supplements to and perhaps even replacements for
the invariant order midrange theories currently in use in Organizational Behavior/
Theory. Specifically, we: (1) define and describe functionalism and comment
briefly on its history in other disciplines; (2) indicate why it may be especially
suited to serve, if not as "an all-inclusive speculation comprising a master con-
ceptual scheme" (Merton, 1957: 6), at least as a solid basis for midrange theory
in Organizational Behavior/Theory; (3) identify some of the pitfalls that threaten
the unwary functional analyst and which in no small measure led to the decline
of the paradigm in sociology and anthropology; and (4) provide an example of
how a functional analysis might proceed in Organizational Behavior/Theory in
the form of a crude functional analysis of leadership.
THE NATURE OF FUNCTIONALISM
The term junction has different meanings in common conversation, the literature
of traditional management, and in sociology and anthropology. Thus, unfortu-
212
FUNCTIONALISM AS A BASE FOR MIDRANGE THEORY 213
nately, it is necessary to nail down what we do not mean by the term before
we can meaningfully defme it. Function as used in this paper refers to none of
the everyday uses of the term: (1) a ceremonial social gathering, (2) the activities
of a "political functionary," or (3) a mathematical relationship of one variable
to one or more others in the sense of depending upon. Nor do we use the term
to designate activities of managers, as did traditional management writers:
Fayol's five elements of administration were: (a) planning (prevoyance), (b)

organization, (c) command, (d) coordination, and (e) control. These elements
generally referred to what were later called duties or functions of manage-
ment. Other writers have added to the number of functions and renamed the
third and fourth; however (then) current classical theory still universally
recognizes planning, organization, and control as useful classifications for
studying management (Massie, 1965: 388).
Even within the mainstream of anthropology, where functionalism was a major

force structuring thought in the early years of the twentieth century and in
American sociology where it was equally important in the years following World
War II, "function" means different things to different writers. At one extreme
stands Levy, who holds:
Shorn of careless uses of definitions and of teleology, structural-functional

analysis is simply a synonym for explicit scientific analysis in general (1965:
22).
At the opposite extreme, though certainly closer to the modal use of the term,
stands Hempel:
The kind of phenomenon that functional analysis is invoked to explain is

typically some recurrent activity or some behavior pattern in an individual
or group; it may be a physiological mechanism, a neurotic trait, a culture
pattern, or a social institution, for example. And the principal objective of
the analysis is to exhibit the contribution which the behavior pattern makes
to the preservation or the development of the individual or the group in
which it occurs. Thus, functional analysis seeks to understand a behavior
pattern or a sociocultural institution in terms of the role it plays in keeping
the given system in proper working order and thus maintaining it as a going
concern (1959: 258).
Part of the definitional problem stems from the fact that functionalism is as
much a state of mind as a single unified technique, making it difficult to capture
its essence in a few phrases or sentences. Nevertheless, we believe that two key
characteristics define this approach, which was and is widely applied in the bio-
logical sciences and which also has been extensively used in anthropology, psy-
chology, and sociology to lead Merton (1949) and Martindale (1960) to label it
respectively as "most promising" and "dominant." To be classified as func-

tionalism so far as this paper is concerned, an approach must necessarily:
1. Seek understanding of events, artifacts, or processes (structures, in the

terminology of functional analysis) in terms of their consequences (eufunc-
tions and dysfunctions) for superordinate systems of which they are
parts, rather than by attempting to identify those things which "cause"
them. 1
2. Do so using concepts and constructs derived from study of the specific
type of situation under investigation, rather than by attempting to im-
pose single conceptual frameworks on all areas of study.
The subtleties involved in differentiating functionalism from the more familiar

invariant order causal explanations of the same phenomena become more ap-
parent in the discussions of the uses, advantages, and disadvantages of functional
analysis as a tool in Organizational Behavior/Theory which follows.
Uses of Functionalism
Functionalism can aid in understanding organizations and the behaviors of

groups and individuals within them in two ways.
First, functionalism can serve as an alternate base for attempts to describe
organizational phenomena. A simple illustration of functionally based descrip-
tion can be derived from consideration of a carburetor in an internal combustion
engine. The combination of a blueprint and a specification sheet provide an ex-
cellent static or Aristotelian (Argyris, 1975) description of it and its component
parts. Such a description of a carburetor has obvious value. For example, it
would permit a person with the proper skills, tools, and materials to duplicate
an existing carburetor with little difficulty. Similarly, one can easily develop a
dynamic description of a carburetor by outlining the motions of its various com-
ponents and the flows of gasses through it. Again, this is a valuable description
that would permit experts to predict the impacts of changes in its characteristics
upon its operation and outputs.
Many would argue, however, that neither of these descriptions captures all
there is to know about a carburetor and indeed that the essence of understanding
it lies in understanding thefunction it performs for the engine of which it is a part.
When one understands that the function of the carburetor is to provde the air/
gasoline vapor mixture necessary to support combustion in the cylinders, one
has some important insight into its operation. For example, it is now possible
to determine whether the particular structure is requisite, that is whether the
engine will continue to operate without it, or auxiliary, in the sense that its
operation would be unchanged were the structure removed or inoperable. Simi-
larly, one can now identify functional equivalents. A fuel injection system, for
example, neither looks like a carburetor-that is, static descriptions of the two
have little in common-nor does it operate in the same fashion as a carburetor;
that is, dynamic descriptions of the two also vary. Yet they are equivalent in
the sense that they perform the same function for the engine. Such knowledge
not only allows for the potential replacement of one with the other, but also
both demands and provides greater insights into the engine, the superordinate
system, of which the structure, the carburetor, is a part.
Second, functionalism can aid in explaining as well as describing organiza·
tional phenomena. A good place to begin an understanding of how functionalism
does this is with a framework developed by MacIver (1973) who identifies six
methods of explaining social phenomena. Three of these: the Why of Obliga-
tion, the Why of Inference, and the Why of Social Conjecture are not directly
relevant to the thrust of this paper. The remaining three: the Why of Invariant
Order, the Why of Organic Function, and the Why of Objective, Motivation, or
Design, however, are directly relevant. They are, in order: the dominant force in
the development of midrange theory in Organizational Behavior/Theory and
thus functionalism's most important competitor; functionalism itself; and, be·
cause it is often confused with functionalism, a major source of problems in the
project considered here.
The Why of Invariant Order involves what is normally considered "cause and
effect." Evered points out:
A "cause" is generally thought of as that factor or agent which produces
(and thus accounts for and may even be responsible for) some thing or other
occurrence. . . . With pure causal determinism, S 1 invariably and necessarily
follows from the action of the causative agent C, occurring in a situation of
the kind, So and the transformation (C + So) -+ SI is described by a law
(1976: 263).
Thus, the phenomenon to be understood is seen as stemming from some "cause"
in a lawful, objectively observable manner. The relationship can be specified at
a conceptual level and is invariant in the sense of always occurring, given iden-
tical cause and situation. Finally, the cause precedes the effect in time, except in
a few cases where cause and effect occur at the same instant. Evered points out
the importance of the Why of Invariant Order in science:
The traditional causal model of explanation is by far the most prominent of
all modes of scientific explanation and has historically speaking, formed the
core of the scientific method .... Since the beginning of modern science, the
term "cause" has been restricted to (this) notion. The pure causalist position
corresponds roughly to the physical or mechanical view of the world, and, in

the social sphere, to an automaton view of man (1976: 264).
The Why of Invariant Order is accepted as the mode of explanation in the main-
stream of Organizational Behavior/Theory to such an extent that many in the
field seem unaware that alternatives exist. Functionalism-Maciver's Why of
Organic Function-provides a clear and, we believe, useful alternative.
Advantages of Functionalism
Functionalism appears to have two major advantages as a basis for midrange

theory in Organizational Behavior/Theory. First, functionalism is highly appro-
priate for a field which, though it investigates a wide variety of other linkages
(Behling, 1978), has its roots in the need of administrators to understand the im-
plications of various structures for organizational survival and performance.
Because functionalism's main thrust is "expressed in the practice of interpreting
data by establishing their consequences for larger structures in which they are
implicated" (Merton, 1957: 46-47), it offers a near-ideal basis for bridging the
gap between the pure and applied ends of Organizational Behavior/Theory. The
outcome of a functional analysis is a description of eufunctional and dysfunc-
tional structures. The translation of such research fmdings into meaningful pre-
scriptions for practice is usually substantially simpler than where speculations
about relationships are framed in terms of the invariant cause model. It appears
likely, in fact, that a shift to functionally based midrange theory in Organiza-
tional Behavior/Theory would effectively erase the gap between the pure and
applied segments of the field. At the minimum, it would reduce the difficulties
involved in applying research findings to the solution of managerial problems.
Second, functionally based midrange theories can supplement systems anal-
ysis as means of understanding organizationally important phenomena. The
systems approach as we define it here has two key characteristics. First, it, like
functionalism, seeks to explain important phenomena in terms of their implica-
tions for higher order systems of which they are parts. Second, the systems ap-
proach is a grand conceptual scheme that attempts to explain diverse phenomena
through a single, parsimonious set of constructs. In what is admittedly one of
the most extreme statements of this idea, Miller points out:
General systems theory is a series of related definitions, assumptions, and
postulates about all levels of systems from atomic particles through atoms,
molecules, crystals, viruses, cells, organs, individuals, small groups, societies,
planets, solar systems and galaxies. General behavior systems theory is a
subcategory of such theory dealing with living systems, extending roughly
from viruses through societies (1955: 5 I 3).
It is our position that the major problems encountered in the application of

systems theory stem more from the latter characteristic than from the former.
Specifically, we hold, as is appropriate at a midrange theory conference, that the
attempt to explain all organizationally important phenomena in terms of a single
set of constructs may lead to some fruitful outcomes, but that it is also likely to
obscure important relationships. We are not alone in this thought: Gouldner, for
example, points out:
If social behavior is to be understood by application of systems models, a
generalized concept of a system alone is insufficient. For there is always a
question of the kind of system model that shall be employed in the under-
standing of social behavior (1959: 242).
Similarly, Weick (1974) in his classic critique of systems theory calls for its re-
duction to the status of a middle range theory because "the high abstractions
found in social systems theory often defy falsification" (1974: 357).
Functional analysis, then, shares the ability of systems theory to direct at-
tention to the eufunctional and dysfunctional consequences of structures for
superordinate systems without the disadvantage of attempting to impose a
single, limited conceptual framework on all phenomena of interest. It permits
the development of individual sets of constructs for the analysis of particular
phenomena. In this sense, functional analysis is to systems theory as conven-
tional invariant order midrange theory is to invariant order grand conceptual
schemes. Functional analysis as proposed here yields functionally based mid-
range theories explaining various aspects of organizations rather than by viewing
them all through a single conceptual lens. As such, it provides a richness of in-
formation sometimes lacking in conventional systems analyses.
Problems with Functionalism
In this section we consider five key problems that hamper the application of
functionalism to Organizational Behavior/Theory. They are: (1) the confusion
of function with purpose, (2) the teleological reversal, (3) the confusion of
sufficiency and necessity, (4) the assumption of universal functionality, and (5)
the assumption of closely linked systems.
MacIver's (1973) Why of Objective, Motivation and Design provides the basis
for a discussion of the first of these problems. He defines "objective" as " ... the
foreseen and intended end-result of any act or series of acts" (MacIver, 1973:
16), while "motivation" designates the emotions, needs or other subjective
conditions leading to behavior; for example, "I did it because I was angry," and
"design" refers to conformity to an accepted pattern as in the case of giving
gifts on birthdays-it is customary to do so. The difference between such ex-
planations and functional ones frequently goes unrecognized as writers assume

that functionality must be a foreseen and intended product of some individual
or individuals. Merton (1949), on the other hand differentiates between manifest
functions - those intended and recognized by participants in the system and
latent functions- those neither intended nor recognized.
Closely tied to the confusion of function and purpose is the problem of tele-
ological reversal. Specifically, functional elements are held to exist because they
are functional, implying the existence of some preordained reason or plan to the
system. Some of the systems of greatest interest to Organizational Behavior/
Theory are not planned or preordained but are unpredicted outgrowths of com-
plex interactions.
A third major problem dogging attempts to apply functionalism in the social
and behavioral sciences is the confusion of functions with structures that per-
form them. There is, by way of illustration, the apocryphal passenger on the
Titanic who was saved at the last moment by climbing aboard the floating grand
piano from one of the ship's salons. From that day forward he refused to sail
aboard any ship not equipped with a piano. Similar logic is applied by individ-
uals who confuse the abolition of specific structures with the elimination of the
function.
Fourth among the problems of functional analysis is the assumption of
universal functionality. While most writers in the field recognize the possibility
that structures having no functional value may persist, others seek hidden func-
tionality in every structure. For example, Malinowski writes:
The functional view of culture insists therefore upon the principle that in
every type of civilization, every custom, material object, idea and belief
fulfills some vital function (1926: 132).
However, particularly in a benign environment or one which brings its weight to
bear upon ill-fitted organizations relatively slowly, structures may persist for
long periods even though they may be nonfunctional, or dysfunctional.
Finally, among the problems encountered in attempts to apply functionalism,
is the assumption of closely linked systems. Functionalism is influential in biol-
ogy because the organic unity of the entities typically studied is unquestionable,
giving the idea that organisms survive through the contributions of their com-
ponents' meaning. Organic unity of business firms and other organizations was
implicitly assumed in previous eras of Organizational Behavior/Theory, but this
assumption is widely challenged by key writers in the field today. For example,
Cohen, March, and Olsen indicate:
... organized anarchies ... are ... characterized by three general properties
.... The organization operates on the basis of a variety of inconsistent and ill-
defined preferences ... its own processes are not understood by its mem-
bers . . . participants vary in the amount of time and effort they devote to
different domains; involvement varies from one time to another.... These
properties of organized anarchy have been identified often in studies of
organizations. They are characteristic of any organization in part-part of
the time (1972: I).
These authors are certainly not alone. Ashby (1960) in his concept of "poorly
joined" systems and Weick (1969, 1974) in his concept of "loosely coupled"
systems and Cyert and March (1959) with their idea of ambiguous goals also
question the assumption of organic unit for many organizations.
Thus far, we have demonstrated that functional analysis may yield new
insights into organizational phenomena, but that five problems may hamper
its use in Organizational Behavior/Theory disciplines. If we restate those prob-
lems as constraints, we defme the limits of a modest but useful functional
approach to understanding organizations:
1. Functional structures need not be traced to conscious, purposive action,

nor need their functionality be recognized by organizational participants.
2. Functionality per se cannot be used to justify the existence of structures.
3. A functional analysis that concludes that a function is performed by only
one structure is suspect; normally functions can be fIlled by several
structures.
4. Functional analyses must identify eufunctional, dysfunctional, and non-
functional structures to be considered successful; not every structure need
be eufunctional.
5. Functional analysis can be applied only to systems where functional unity
can be demonstrated.
These constraints reduce the scope (and fun) of functional analysis, but they
also minimize the chances of falling into pitfalls that have plagued its application
in other social and behavioral disciplines.
AN EXAMPLE OF FUNCTIONAL ANALYSIS
To illustrate how functional analysis might work in Organizational Behavior/

Theory, let us speculate about a limited functional analysis of leadership. This
topic is appropriate for two reasons. First, existing paradigms are under severe
attack (e.g., Miner, 1975) and the field appears ripe for a paradigm shift. Second,
a number of attempts to develop such an approach already exist. Some of them
(Bales, 1953, 1958; Thibaut and Kelley, 1959; Bavelas, 1950; Cartwright and
Zander, 1968; Hollander and Julian, 1969; Lord, 1977) are well grounded in
functional analysis. Other authors (Evan, 1970; House, 1971; Miner, 1975; Kerr,
1977) appear to be moving in the direction of functionally based theories of
leadership without clearly recognizing them for what they are. The following
paragraphs outline a crude functional analysis ofleadership built on these works
to illustrate the potential of functional approaches and the processes through
which they may be developed.
Functional explanations begin not with the structure of interest-in this case,
leadership - but rather with the superordinate system of which it is a part. Thus,
a functional analysis of leadership begins by looking at work groups and inquiring
about the functions requisite for their survival in current form or for task ac-
complishment. For this analysis we borrow a list that is both simple and widely
accepted. Building on Homans' (1950) discussion of the internal and external
systems of groups, Bales (1953, 1958); Thibaut and Kelley (1959); Katz and
Kahn (1966) and Lord (1977) identify task functions:
The group must operate effectively upon (its) environment in order to pro-
vide adequate rewards to its members. Thus a family must have adequate
financial income or sufficient provision of food and materials from its land;
a baseball team must have financial and moral support from its fans. Hence
one general problem that confronts this generalized type of group is that of
controlling its social and physical environments so that they yield high out-
comes for its members ...the adequacy of performance of task functions
determines how much usable power the group has (Thibaut and Kelley, 1959,
pp.274-275).
In work groups embedded in larger organizations the task function is half of an
exchange relationship between the work group and the parent organization. The
organization defmes the contribution the group is to make in terms of schedules,
quotas, objectives, and the like. Management information, output reporting,
performance evaluation, and accounting systems test the group's contribution
to the parent organization.
The same writers also identify maintenance functions concerned with reward-
ing work group members to assure their participation and performance:
Even if usable power is relatively high, various circumstances may intervene
to make group survival problematic in the absence of special measures to
insure it. First, the scheme for allocating rewards within the group may be
such as to place some members perilously close to (the standard they use in
deciding whether to remain in or leave the group). Second, in task activities
as in any activities, costs are involved, and these generally are incurred before
rewards are forthcoming. Moreover, certain members may incur unusually
heavy costs. And, finally, in many task activities the rewards are fed back to
the group only after considerable time delay .... We refer to these activities
by which each member maintains his position (above the standard used in
deciding to remain in or leave the group) as the maintenance functions

(Thibaut and Kelley, 1959: 275).
Maintenance functions (sometimes referred to as "socioemotional," e.g., Bales

and Slater, 1955) involve allocating and timing rewards for work group mem-
bers, creating new rewards, cutting negative payoffs, and managing information
to change the ways work group members value rewards of group participation
relative to those available elsewhere.
Functional analyses cannot be falsified in the conventional sense. It is pos-
sible, however, to test the degree to which postulated requisite functions are
indeed necessary, the second phase of functional analysis. If work groups exist in
which the task or maintenance functions are absent, for example, the model
must be reevaluated.
Tests of the requisiteness of the task function are relatively unambiguous.
Lord (1977) in a survey of six studies of leadership and in his own study found
activities of leaders contributing to the task function in all cases. The issue is less
clearly resolved relative to the maintenance function. It has long been an article
of faith in Organizational Behavior that while satisfaction might ambiguously
relate to employee performance or effort, it is closely related to propensity to
leave the work group and thus by defmition to the stability of the organization.
There is good support for this contention (e.g., Steers and Porter, 1973). Nichol-
son, Brown, and Chadwick-Jones, however, reviewed 29 studies and examined
the relationship of three measures of absence to five modified Job Description
Index satisfaction scales among 1,222 production workers in sixteen organiza-
tions and found that, "in most instances job dissatisfaction and absence from
work were unrelated, and the lack of relationship was not attributable to arti-
factual or extraneous influences" (1976: 728). This raises the possibility that the
maintenance function may not be requisite.
The third phase of the development of a functionally based midrange theory
is the identification of the structures through which the function can be accom-
plished. Miner (1975) concerns himself primarily with the task function of the
organization and identifies five structures through which it may be manifested:
1. Self-Control-Self-control reflects residues of early parental training in

the form of a conscience, superego or ego ideal.
2. Hierarchic Control- Individuals holding organizationally superordinate
positions direct the actions of their subordinates.
3. Professional or Jdealogical Control- The employee relies on values, norms
or codes of outside groups for task direction.
4. Group Control-Task direction is provided by peers in the work group
itself, operating through afflliative and conformity motives.
5. Task Control-Task direction is provided by the job itself. Miner cites the
pacing of an assembly line or the crush of customers around a sales clerk
as examples.
Kerr (1976) identifies eleven structures in the subordinates, the task, and the
organization which may substitute for the conventional task leadership that sup-
plements Miner's list. Further, Kerr lists four structures that may substitute for
maintenance leadership, fulfilling the maintenance functions:
1. Professional Orientation- The employee receives rewards from his/her

professional peers, which may supplement or supplant those received
from the individua1's organizational superior.
2. Unconcern with Available Rewards - The employee receives adequate
satisfaction from other spheres of his/her life, making their presence or
absence in the work situation less important.
3. Intrinsically Satisfying Tasks-The work itself may be intrinsically mo-
tivating, increasing performance, and decreasing turnover.
4. Reward Systems Beyond the Control of the Leader-Membership in
informal social groups in the work place as well as compensation pro-
grams controlled at higher organizational levels may serve to tie individual
employees to the work group.
Obviously, neither the list of task or maintenance structures is complete. Yet,

carefully developed and operationalized, they provide the basis for the fourth
step in functional analysis: empiric test. This takes two forms, both of which
yield substantial insights into the nature and operation of the system of concern.
The first is removal of existing structures. What happens to a work group for
example, as in the case of Roy's classic interference with "banana time" (1959)
when it loses a key maintenance structure? The second form of empiric test
deals with attempts to tease out differences in consequences of different struc-
tures fulfilling the same function. For example, Miner's "hierarchic control" and
the employee's "professional control" both serve task functions. Yet the con-
sequences of a shift from one to another as a basic means of fulft11ing this
function obviously are complex and demand empiric investigation. Similarly, as
discussed by House (1971) the introduction of additional superficially functional
structures to a system where a function is adequately realized appears to trigger
negative attitudes.
The final step in functional analysis, if any scientific endeavor can be thought
of as having an end, is application. As previously discussed this phase is easy, re-
lative to attempts to apply the knowledge generated by invariant order analyses,
since the information yielded by functional analyses parallels directly key issues
of concern to practitioners of management. Thus, functionally based midrange

theories may prove valuable supplements to invariant order analyses in Organi-
zational Behavior and Theory.
NOTE
1. Functional analysts unfortunately use the term "structure" not in reference to com-
munications channels, work flows, authority and responsibility relationships and the like, as
it is commonly used in Organizational Behavior/Theory, but to refer to any observable
aspect or pattern in a system under study.
REFERENCES
Argyris, C. Dangers in applying results from experimental social psychology.
American Psychologist, 1975,30,469-485.
Ashby, W.R. Design for a Brain (2nd ed.). New York: Wiley, 1960.
Bales, R. F. The equilibrium problem in small groups. In Working Papers in the
Theory of Action. T. Parson, R.F. Bales and E.A. Shils (Eds.), New York:
Free Press, 1953.
Bales, R.G. Roles in problem-solving groups. In Readings in Social Psychology.
E.E. Macoby, T.W. Newcomb and E.L. Hartley (Eds.), New York: Holt,
Rinehart & Winston, 1958.
Bales, R.F., and Slater, P.E. Role differentiation in small decision-making
groups. In Family, Socialization and Interaction Process. T. Parsons and
R.F. Bales (Eds.), Glencoe, Illinois: Free Press, 1955.
Bavelas, A. Communication patterns in task-oriented groups. Journal of the
Acoustical Society of America, 1950,22, 725-730.
Behling, O. Some problems in the philosphy of science organizations. Academy
of Management Review, 1978,3, in press.
Cartwright, D. and Zander, A. Group Dynamics: Research and Theory (4th ed.).
New York: Harper & Row, 1968.
Cohen, M.D., March, J.G., and Olsen, J.P. A garbage can model of organiza-
tional choice. Administrative Science Quarterly, 1972,17, 1-26.
Cyert, R.M., and March, loG. A behavioral theory of organizational objectives.
In Modern Organizational Theory. M. Haire (Ed.), New York: Wiley, 1959.
Evan, M.G. The effects of supervisory behavior on the path-goal relationship.
Organizational Behavior and Human Performance, 1970,5, 277-298.
Evered, R.D. A typology of explicative models. Technological Forecasting and
Social Change, 1976,9,259-277.
Gouldner, A.W. Reciprocity and autonomy in functional theory. In Symposium
on Sociological Thought. L. Gross (Ed.), Evanston, Illinois: Row, Peterson,
1959.
Hempel, C.G. The logic of functional analysis. In Symposium on Sociological

Thought. L. Gross (Ed.), Evanston, Illinois: Row, Peterson, 1959.
Hollander, E.P. and Julian, J. W. Contemporary trends in the analysis of leader-
ship processes. Psychological Bulletin, 1969, 71, 387-397.
Homans, G.C. The Human Group. New York: Harcourt, Brace and World, 1950.
House, R.J. A path-goal theory of leadership effectiveness. Administrative
Wiley, 1966.
Kerr, S. Substitutes for leadership (WPS 76-11). Columbus, Ohio State Univer-
sity College of Administrative Science, 1976.
Koontz, H. The management theory jungle. Academy of Management Journal,
1961,4,174-188.
Lord, R.G. Functional leadership behavior: Measurement and relation to social
power and leadership perceptions. Administrative Science Quarterly, 1977,
22.
MacIver, R. M. Social Causation. New York: Harper, 1973.
Malinowski, B. Crime and Custom. London: Paul, Trench, Trubner, 1926.
Martindale, D. The Nature and Types of Sociological Theory. Boston: Houghton-
Mifflin, 1960.
Massie, J.L. Management theory. In Handbook of Organizations. J.G. March
(Ed.), Chicago: Rand McNally, 1965.
Merton, R.K. Social Theory and Social Structure. Glencoe, Illinois: Free Press,
1949.
- - . Social Theory and Social Structure (2nd ed.). Glencoe, Illinois: Free Press,
1957.
Miller, J.G. Toward a general theory for the behavioral sciences. American Psy-
chologist, 1955,10, 5l3-53l.
Miner, J.B. The Uncertain Future of the Leadership Construct: An Overview.
Paper presented at the 1975 Southern Illinois University Leadership Confer-
ence, Carbondale, Illinois.
Nicholson, N., Brown, C.A. and Chadwick-Jones, J.K. Absence from work and
job satisfaction. Journal of Applied Psychology, 1976,61,728-737.
Porter, L. W. and Steers, R.M. Organizational, work, and personal factors in em-
ployee turnover and absenteeism. Psychological Bulletin, 1973,80, 151-176.
Roy, D.F. Banana time. Human Organization, 1959-60,18, 158-168.
Thibaut, LW. and Kelley, H.H. The Social Psychology of Groups. New York:
Wiley, 1959.
Wesley, 1969.
- - . Middle range theories of social systems. Behavioral Science, 1974, 19, 357-
367.
Weiss, R.S. A structure-function approach to organization. Journal of Social
Issues, 1960,12,61-67.
18 ORGANIZATIONAL IMPLICATIONS
OF EXCHANGE THEORY
Is It Time for a General Middle
Range Theory?
Claremont Graduate School
The lesson of early organizational theory is now painfully clear to both re-
searchers and practitioners. Bluntly, our general theories collapsed as we dis-
covered that they applied only in limited circumstances. Since the theories were
intended to be broad in scope, they were necessarily abandoned, with interest
turning toward more promising new possibilities. The wreckage of the grand
motivational, leadership, and job design theories from the first half of this cen-
tury provides obvious examples of this process. The overall shift from engineer-
ing, through human relations, to the present organic (organizational behavior)
approach to our field is evidence of the same phenomenon.
One result of the failure of general theories has been a reaction against
the use of any theory. Many researchers are now compiling atheoretic studies,
which merely compare the results obtained under different sets of conditions.
Thus, for instance, we have learned that personnel managers discriminate in
hiring certain groups under some conditions-but until recently there has been
little effort to develop a theoretical explanation for this observation (Larwood
and Wood, 1977).
The atheoretic approach is most useful to researchers making preliminary
surveys or gathering data for use in solving immediate and limited problems.
Although it is initially productive, atheoretic research is an inefficient long-term
225
practice because it produces results of undetermined generalizability or meaning.

Lacking a theoretical rationale, each study is technically independent of all
others. Implications for later research or conclusions from past work, are often
lost because no theoretical mechanism is available by which such implications
might be drawn.
Most researchers recognize the drawbacks of both too general a theory and
the lack of theory, and have turned to the formulation and testing of middle-
level rationales. Middle-level theories strike a compromise in which they attempt
to explain only a limited range of data and modestly offer no formal predictions
concerning other phenomena. At this point, for example, the vertical dyad
linkage model of leadership (Dansereau, Graen, and Haga, 1975) is concerned
only with the interactions between leaders and followers. Factors such as the
nature of the work environment may also be important to leadership, but are
specifically excluded from consideration except insofar as they alter the leader-
follower relationships. Since middle range theories such as this are designed to
fit a limited set of events, they can readily be altered or abandoned as new infor-
mation is obtained. Thus, a high degree of predictivity is possible within a tightly
focused category of observation.
While undoubtedly advantageous, the limited scope of effective middle range
theories results in the need for a much larger number of theories, posing other
problems for the researcher. Presumably, new theories are required as research
interest turns to new phenomena. Nonetheless, by their limited nature, middle
range theories provide no guidelines for the development of new concepts or for
the appropriate selection between competing hypotheses in new areas. In time,
of course, we might expect to fmd middle range theories filling in these gaps,
but their doing so poses a new problem: Which theory should be applied at the
interface between two or more notions?
The difficulties inherent to application, extension, and development of mid-
dle range theories might be overcome if the theories followed some set of
recognized superordinate decision rules or contained commonalities. The rules
or commonalities could be applied either in building new theory (as a first step),
deciding between competing predictions (accept those based on the rules first),
or extending existing theories (extend the rules with some certainty). The
present paper asserts that a developing theory of social exchange provides a
possible common link between many of the successful middle range theories
currently employed in organizational studies. As outlined in the next sections,
the elements of exchange theory are consistent with a number of middle range
approaches, allowing those approaches to be interpreted as parts of a broader
over arching network. Although integrated in the sense of being placed in a larger
perspective, the middle range theories are not absorbed; they can continue to be
applied in most of the circumstances for which they were designed. However, to
ORGANIZATIONAL IMPLICATIONS OF EXCHANGE THEORY 227
the extent that the exchange concepts are accepted, these concepts suggest limits
for some middle range theories, referee conflicting predictions between them,
and suggest new lines of future research.
EXCHANGE THEORY
Stripped to its most basic form, exchange theory suggests that the behavior of
an individual in a social context is interdetermined with the perceived behavior
of at least one other participant (Chadwick-Jones, 1976; Ekeh, 1975). That is,
any individual's behavior is "exchanged" with that of others, with the nature of
the exchange following predictable rules based on the perceptions of the partici-
pants. Although the exchange rules are assumed to follow consensual norms,
until recently no detailed testable rationale explained either why this process
should occur or how such a concept of exchange might be applied. Perhaps in
consequence of this frustrating lack of defmition, the more general theory has
been cannibalized by less ambitious middle range theories. For example, limited
theories of power (Blau, 1964) and distributive justice (Homans, 1961, 1974)
have employed restricted notions of the exchange process, which proved useful
in examining certain political and economic behaviors. The absence of explicit
exchange guidelines has unfortunately allowed these limited theories to be
applied well beyond the phenomena for which they were designed (Chadwick-
Jones, 1976). The resulting failure to confirm their generalizability has been
incorrectly interpreted as a failure of both the middle range theories and the
underlying exchange notion (cf. Goodman and Friedman, 1971; Lawler, 1971).
It is easy to understand why a mathematically explicit derivative, such as
equity theory (in which individuals are assumed motivated to ensure payoffs
proportional to their relative contributions-see Adams, 1965; Walster, Ber-
scheid, and Walster, 1973; Walster, 1975), would be accepted and tested in
place of its more general and vague parent. Do we have an alternative? It seems
likely that recent developmental research has finally provided an acceptable
rationale as to why exchanges develop and how they work (cf. Larwood, 1979;
Lerner, 1977). Rather than replacing viable middle range theories with an un-
tested theory of exchange, the latter is now coming to be seen as a mechanism
that might be used to sort between other limited theories (cf. Chadwick-Jones,
1976; Cook, 1975; Deutsch, 1975). The dimensions of this emerging theory are
briefly sketched below.
1. In agreement with the notions of sociobiology and learning theory, indi-

viduals are assumed to be "wired" to ensure their survival (or, at least,
their genetic survival-Wilson, 1975). Thus, people react behavioristica1ly
to reinforcements and learn from their experiences. Interpreted another

way, people exercise control over their behavior, preferentially selecting
those actions which they consider most advantageous (or "self-efficacious"
in the terms of Bandura, 1977). Of course, the particular behaviors
thought to be most efficacious depend on the learning, ability, alterna-
tives, and time frame of the individual.
2. Although the assumptions above suggest that people behave in their im-
mediate self-interest, the concept of self-efficaciousness cannot be so nar-
rowly interpreted. In fact, as they grow older most children develop a
"personal contract" (Lerner, Miller, and Holmes, 1976) in which they
learn to defer present gratification for more highly valued later rewards.
Observation of a contingent relationship between present behavior and
future reinforcements would seem to be a precondition to the develop-
ment of the willingness to delay gratification (cf. Mischel, 1973). Persons
with lower intelligence, with noncontingent rearing, or who otherwise
value immediate gratification may be less willing than others to accept
any delay.
3. In a process similar to that with the personal contract, people learn to
develop a "social contract" in which they exchange socially valued be-
havior for more highly valued reinforcements from others (cf. Blau,
1964). Again, the extent and nature of the social contract vary with the
individual. It might be noted at this point that item 3 is the common
element of middle level theories using the exchange notion. Nonetheless,
as the other items attest, this is only a portion of the present theory.
4. The individual develops a sense of "justice" requiring delivery on the
social and personal contracts of the self and others with whom he/she
is in close contact or identifies (Adams, 1965; Lerner, 1977). Although
often treated as an inscrutable force by researchers, the need for justice
is both comprehensible and predictable. Capricious rewards or punish-
ments serve to undermine the predictivity essential to maintenance of the
personal and social contracts. Thus to the extent that the individual has
made psychological or behavioral commitments to those exhanges, she or
he predictably defmes the exchanges as fair and defends them. A just sys-
tem of rewards is then one which the individual feels some degree of com-
mitment towards; other systems may be equally just to other persons. As
a number of researchers have pointed out, the defense of justice can take
a variety of forms including: adjustment of one's own performance to
correspond with the salient exchange norms (Garland, 1973), influencing
the performance of others toward the fair norm (Roy, 1952), disassociat-
ing one's self from the unfortunate (Lerner, 1971), and adjusting one's
perception of the situation (Gergen, Morse, and Bode, 1974).
The four components above should be considered concurrent and consistent

with one another. For example, the question of whether an individual behaves
according to self-interest (item 1) or justice (item 4) does not arise: Justice is
defmed as being in the individual's subjective self-interest. The exchange theory
described above is deliberately general in the sense that it can be applied across a
wide variety of situations. The theory makes allowances for differences in out-
comes by predicting differences in individual perception, response capability,
and reaction (Larwood, 1979). Most important for the present purpose, it inter-
faces with a number of middle range theories; some examples appear in the
next sections.
OPERATIONAL OVERVIEW
What is the major thrust of exchange theory from the point of view of organiza-
tional behavior? As may be seen from the items listed above, the theory is
strangely undistinguished-theorists have examined these elements in one form
or another for over half a century. The manner in which the parts are brought
together is new, however. It seems reasonable to expect that the new combina-
tion will provide different and perhaps more useful insights (cf. Weick, 1977).
Taken as a whole, the theory suggests that prospective and new members of
an organization are highly motivated to learn the prevailing system of organiza-
tional norms (concerning pay, roles, growth, output, ethics, etc.) because each
norm implies an exchange in which something is given up while something else
is gained. The particular set of organizational norms is not specified by exchange
theory; nonetheless, it must be such (or appear to be such) as to attract the type
of personnel who will sustain the system. Individuals discovering a set of norms
too distant from those they had previously seen as justified would predictably
refuse the enterprise's employment offer, or would quickly leave for other
opportunities. They would of course feel that the system was subjectively unfair
or incomprehensible, that is, threatening to their sense of justice.
Individuals who stay are more often those who feel comfortable with the sys-
tem, those who can accommodate or can alter minor points of unfairness, those
who change their perceptions toward agreement with the system of norms, or
those who have no choice. Those voluntarily staying with the organization do so
presumably because they see this as their best (most efficacious) option-and
because they see the option itself as relatively acceptable. Thus, the theory pre-
dicts that they behave according to the prevailing norms and begin to build a
stake in the organization's continuance. The norms, of course, need not neces-
sarily work toward management's advantage; they definitely will not work to the
individual's disadvantage (as perceived by the individual) for any length of time.
After a period, the individuals in any given position may actively resist change
because of its potential for disrupting the system that they have come to accept
as fair. Personnel who have experienced considerable mobility are obviously
exceptions, since they may see mobility itself as the norm.
Employees who are staying with the organization involuntarily (perhaps
locked in by a poor job market or the possibility of blacklisting) are unpredict-
able from a management viewpoint. Seeing themselves as temporary and the sys-
tem as unfair to them, they have little reason to invest themselves in the position
to any greater extent than necessary to keep their job. Thus, these individuals
would predictably slacken their performance below the norm unless monitored
in some manner. Some might engage in unethical practices in efforts to restore
justice as they perceive it (sabotage is justified, for example, if one is forced to
work rapidly for low pay).
To some extent, an organization can alter the norms by replacing the existing
personnel with others who are unaware of the previous norms. Nonetheless,
social expectations also influence the norms anticipated by incoming personnel.
As well, the individual differences greatly influence the manner in which an
employee perceives or operates in the enterprise. Consequently, some individuals
will need more rapid reinforcement because they come from social strata in
which little long-term planning was possible; other persons may see themselves as
managers and resist norms they feel do not make management possible for them.
In short, an effective fit between individual and job requires that the individual
view the exchange norms as just from his or her point of view; the question of
what is just must be resolved on an individual basis.
APPLICATIONS
In the preceding section, the nature of the specific exchange norms that might
be considered by exchange theory was left open. Several different middle-level
theories already examine normative types of organizational relationships. To the
extent that exchange theory is useful in orchestrating those theories, we would
expect to find that the exchange concepts are consistent with the major predic-
tions the middle-level theories make ... but that some of those theories will be
extended somewhat, while others may be limited.
Fair Economic Exchange
The exchange formulation above can readily retain equity theory as the norm in
many economic relationships: People undeniably do exchange pay in proportion
to performance, and do defend that relationship as just across a wide variety of

circumstances (cf. Adams and Freedman, 1976). However, exchange theory sug-
gests that the relationship perceived as fair will be defended. There seems no
reason to suggest that equity, as perceived by the researcher, will be thought to
be the only fair relationship.
While apparently minor, the change is an important one. Under the analysis
of exchange theory, equity becomes one of a variety of objective economic
relationships (Cook, 1975; Deutsch, 1975). If the change is accepted, employers
who are attempting to fix a fair rate of remuneration must first determine the
nature of the exchange that employees are referencing. In fact, at least two other
important exchanges occur besides equity: equality, in which the parties divide
rewards equally, and winner-take-all, in which one individual is rewarded while
others are not.
Studies examining the exchange selection process suggest that cooperative
situations are more likely than competitive circumstances to elicit the equality
exchange (Larwood, Kavanagh, and Levine, 1978). Those who are not members
of a system paying them on a direct incentive basis often prefer the equality
exchange across a number of situations (Larwood and Blackmore, 1977; Lar-
wood and Moely, 1979). Other studies have found that the exchange selected is
considered fair and is defended-irrespective of which exchange it is (Austin and
McGinn, 1977; Kahn, Lamm, and Nelson, 1977; Larwood and Blackmore, 1977).
Finally, individuals seem to take the exchange preferences of others into account
before making their own selection, thereby being more certain of selecting a
consensually fair exchange (Kidder, Bellettrie, and Cohn, 1977; Larwood and
Blackmore, 1977; Leventhal, 1976; Reis and Gruzen, 1976). Together these
studies provide strong evidence that the predictivity of equity is enhanced by
determining first that the individual is apt to be using it. Equity is useful, but
equity in the context of exchange theory is more useful.
Role Theory
Role theorists have tended to view roles as flXed consensual behavior patterns
that individuals carry across situations, or more recently, which individuals
exhibit in particular situations, such as in the work environment (Katz and
Kahn, 1978; Sarbin and Allen, 1968). Exchange theory does not disturb the
bulk of role theory research, but it does offer a reinterpretation that may be of
particular use in understanding role change and role conflict phenomena. The
exchange perspective suggests that role behaviors are manifestations of the social
contract-behavior modalities, which the individual exchanges in return for the
rewards expected from others. This is in agreement with research indicating that
the individual's expectations determine the role to be played, and that the role
will be performed only so long as the conditions making it appear subjectively
efficacious continue (cf. Graen, 1976; Larwood and Wood, 1977).
Role conflict can readily be interpreted as a situation potentially upsetting
the individual's return on the social contract; such situations can understandably
arouse anxieties and will be preferentially avoided by most persons (Kahn, Wolfe,
Quinn, Snoek, and Rosenthal, 1964). In dealing with unavoidable role conflict
(situations suggesting the use of two mutually exclusive roles), the individual
may be predicted to behave in the manner considered most efficacious. While
either role may (depending on the situation) seem justifiable, some alternatives
may seem threatening and unpredictable when viewed by other parties to the
exchange. Thus, the individual chooses a path that seems both fair and reason-
ably free of the negative sanctions of others, and which potentially leads to the
outcomes for which the role relationship was originally entered. Role change is
similarly accomplished when another role appears more efficacious in the time
frame adopted by the role player-or when no role is sufficiently worthwhile
to warrant continuing social investment and delay of gratification.
From the organizational viewpoint, this theoretical change is substantial. Role
theory (as modified here) now says that the presence of different personnel in
the same room with an employee may result in the employee's use of quite dif-
ferent objective roles (Larwood, O'Carroll, and Logan, 1977; Larwood, Zalkind,
and Legault, 1975). The role differences come about because of the differences
in expectations the audience is believed to have; a different set of role behaviors
may be produced to meet the demands of each new exchange situation. While
the pattern of changes is predictable, the best predictions will require that we
examine the perceptions of the role player.
Leadership
Although fascinating to organizational theorists, the construct of leadership

has proven particularly hostile to theory development (Stogdill, 1974). Re-
cently, middle-level theory concerning leadership seems to have entered a new
stage, however. By examining the interaction between the leader and the sub-
ordinate, some theorists have developed forms of an exchange model in which
leaders are able to lead by demonstrating to subordinates that it is to their bene-
fit (or efficacious in the terms used here) to perform as requested (cf. Rouse,
1971). Presumably the leader similarly gains from appropriate behavior on the
part of the subordinates.
The vertical dyad linkage model (Cashman, Dansereau, Graen, and Raga,
1976; Dansereau, Graen, and Raga, 1975; Larwood and Blackmore, 1978) ap-
pears to take the exchange model an additional step: Role relationships are
defined between leaders and subordinates. While some subordinates are members
of the leader's ingroup, others might more appropriately be considered members
of the outgroup. Ingroup members perform different types of behaviors for the
leader and are accorded different advantages from outgroup members. Thus, for
instance, in the Dansereau et al. study, the ingroup members performed more
challenging assignments and indicated a higher satisfaction with the relationship,
while the outgroup members withdrew from the leader and developed a high
turnover rate.
Although the exchange theory developed in this paper agrees with the vertical
dyad linkage model, the theory also has several ideas to add to the model. For
instance, the concepts of efficacy and fairness might be applied directly to pre-
dict who is most satisfied with or likely to drop from the subordinate group. The
nature of the work environment might be expected to affect the group's perfor-
mance in a complex manner by providing cues to some group members that their
personal or social contracts are at risk, causing them to react so as to restore the
contracts. In this manner, the vertical dyad linkage model might be usefully ex-
tended to include a wider cross-section of leadership phenomena.
Ethics
Despite the embarrassing number of public investigations into organizational

ethics (Watergate, ITT/Chile, fluorocarbons, etc.), little experimental research
has focused on the conditions under which management does, or does not,
behave in a socially responsible manner. Although useful, survey studies such as
that of Brenner and Molander (1977) are normally subject to impression man-
agement and may reflect neither actual opinions nor associated behavior. Per-
haps as a consequence of the difficulties innate to the research (How does one
ethically or legally manipulate and observe organizational ethics?), the area of
business ethics has little distinct theory. Instead, some observers seem to be
borrowing the expectancy theory model from motivational and decision theo-
rists. Under this rubric, managers are predicted to examine the attractiveness of
responsible and irresponsible alternatives, and to weigh the risks associated with
them (Forrest and Johnson, 1977; Keirn, 1977).
Exchange theory readily accepts the notion of a manager acting in a rational
manner to maximize her or his subjective outcome; expectancy theory can be
seen as a way in which the efficacy concept can be operationalized. A single
revision is suggested, however. In deciding the value of a particular alternative,
exchange theory predicts that fairness is taken into account. In the more usual
terms, people can be predicted to choose the behavior or pattern of behaviors
that maxImize the value of their subjective expected outcome (calculated

without considering fairness) provided that the outcome is as subjectively
fair as its alternatives; if the apparently optimal outcome is considered unfair,
reversals may occur. The normative "rules of the game" obtained from experi-
ence, competitive standards, and the law can all be used to derme fairness in
this formulation.
The stipulation above suggests a number of testable hypotheses. The predic-
tion may be made that managers perceive established government regulations as
reasonable, and that they try to consider them in their final decisions. Increasing
govemment regulation aimed at forcing social responsibility would be seen as
an unpredictable change in the rules, threatening the social contract, and there-
fore unfair. Since injustice may be threatening to both the organization and its
management, it may be justifiably countered in any "reasonable" manner. On
the other hand, a system of bribery is not unfair (as seen from management's
point of view) unless it is used by one firm alone or is demanded unexpectedly;
instead, government intervention to end such a system is unfair. In short, ex-
change theory asserts that much of the behavior seen as objectively socially
irresponsible may be ethically justified by those engaged in it.
CONCLUSIONS
In any overview of the theoretical progress of organizational research, the ob-

server cannot avoid the conclusion that middle range theories have "come of
age." The middle range theories often bring a considerable degree of predictivity
to a limited area of observation; they can be proposed in sufficient detail to be
useful in field settings, and can be flexibly altered when necessary. The limita-
tions that make middle range theory useful are also problematical, however. Re-
searchers are tempted to overextend limited theories, are sometimes confused by
overlapping theories, and in some cases have no theory. Some of the problems
with middle-level rationales might be avoided if the theories were integrated into
some overall network. This paper has suggested that exchange theory provides
such a network.
Four concepts have been identified as basic to exchange theory: people be-
have according to modified self-interest, trade their present efforts for future
gratification, trade with others according to normative exchange patterns, and
defend their idea of justice. Because exchange theory is conceived as a unifying
set of principles, it should be consistent with many of the predictions made by
other middle range theories. Where inconsistencies are found, exchange theory
can be used to modify (restrict, expand, reinterpret) the middle range theories.
Brief examples indicated the marmer in which exchange theory can be applied
to middle-level theories concerning economic exchanges, roles, leadership, and

social responsibility. In part, the success of exchange theory depends on the
success of the modifications proposed.
Other questions also need to be resolved. To what additional middle-level
rationales can exchange theory be applied? Presumably, it is useful only in situa-
tions allowing the four core concepts to be applied. More specifically, the sit-
uations of concern to the exchange theory described here involve continuing or
repeated voluntary social interactions that appear to have some potential value
to the participants. Unique or valueless encounters are beyond the scope of this
proposal because they present no incentive for the learning or performance of
exchange norms.
Were the middle range theories considered here particularly well suited for
the application of exchange theory? Other middle range theories might as readily
have been examined. For example, the exchange notions can be used to modify
two earlier limited views of power - those of Etzioni (power as control of re-
sources, 1961) and of French and Raven (power as perceived by others, 1959).
The resulting theory of control retains both middle range theories but suggests
that neither is by itself sufficient to describe the exchange: Both participants
have apparent control over their own resources, and control one another to the
extent that each desires the other's resources. In short, exchange theory suggests
that power is an exchange exercised by all parties simultaneously; one party can-
not unilaterally control another in a voluntary interaction. It is of some interest
that this result is compatible with the exchange elements of other power models
(Blau, 1964; Katz and Kahn, 1978; Pfeffer, 1977). Additional applications of
exchange theory might be suggested for reexamination of the organizational
socialization process, communications theory, and the creation of normative
management policies.
Since middle-level theory was developed partly in reaction to more general
concepts, it seems fair to ask whether the imposition of exchange theory does
not return researchers to their starting point. Is exchange theory a general or a
middle range theory? By stating how individuals will behave across the range of
exchange situations, this theory is painting with a very broad brush. However, as
seen above, the theory is limited to certain social activities. Further, the theory
lacks either the detail or pretense of detail necessary for it to fully describe or
predict behavior in the situations in which it is used. Put another way, the
theory is able to deal with a wide range of phenomena because it relies on
middle range theories for its specific application. While exchange theory can be
used to modify and guide middle range theory, it is itself a middle range theory
comprised of some principles about which we have reasonable trans-situational
agreement.
Is it time for a general middle range theory? Although there is a need for
theories linking middle range concepts to one another, the success of such a
theory remains to be determined. The present theory asserts that its four com-
ponent concepts underlie many social interactions. As more information is
discovered concerning those concepts, it seems reasonable to expect that ex-
change theory itself will change shape somewhat. While the changes may bring
the theory more closely into agreement with some middle range theories, the
possibility also exists that a corrected exchange theory will become too spe-
cialized to serve as an integrative network for others. In that event other such
middle-level "theories about theories" will be needed. In the meantime, ex-
change theory does seem to provide a fertile perspective for the examination of
both organizations and theories about organizations.
REFERENCES
Adams, J. S. Inequity in social exchange. In L. Berkowitz (Ed.), Advances in

Experimental Social Psychology, Vol. 2. NY: Academic Press, 1965.
Adams, J.S., and Freedman, S. Equity theory revisited: Comments and an anno-
tated bibliography. In L. Berkowitz and E. Walster (Eds.), Advances in Exper-
imental Social Psychology, Vol. 9, NY: Academ •.: Press, 1976.
Austin, W., and McGinn, N.C. Sex differences in choice of distribution rules.
Journal of Personality, 1977,45, 379-394.
Bandura, A. Self-efficacy: Toward a unifying theory of behavioral change. Psy-
chological Review, 1977,84, 191-215.
B1au, P.M. Exchange and Power in Social Life. NY: John Wiley & Sons, 1964.
Brenner, S.N., and Molander, E.A. Is the ethics of business changing? Harvard
BusinessReview, 1977,55 (1), 57-71.
Cashman, J., Dansereau, F., Jr., Graen, G., and Haga, W.J. Organizational under-
structure and leadership: A longitudinal investigation of the managerial ro1e-
making process. Organizational Behavior and Human Performance, 1976,15,
278-296.
Chadwick-Jones, J.K. Social Exchange Theory: Its Structure and Influence in
Social Psychology. London: Academic Press, 1976.
Cook, K. S. Expectations, evaluations, and equity. American Sociological Re-
view, 1975,40,372-388.
Dansereau, F., Jr., Graen, G., and Haga, W.J. A vertical dyad linkage approach to
leadership within formal organizations: A longitudinal investigation of the
role-making process. Organizational Behavior and Human Performance, 1975,
13,46-78.
Deutsch, M. Equity, equality, and need: What determines which value will be
used as the basis of distributive justice? Journal of Social Issues, 1975, 31 (3),
137-149.
Ekeh, P.P. Social Exchange Theory. Cambridge, Mass.: Harvard University Press,
1974.
Etzioni, A. A Comparative Analysis of Complex Organizations. Glencoe, Ill.:

Free Press, 1961.
Forrest, C.R. and Johnson, A.C. The employment of minorities: A conceptual
framework. Academy of Management Proceedings, 1977,266-270.
French, J.R.P., and Raven, B. The bases of social power. In D. Cartwright (Ed.),
Studies in Social Power. Ann Arbor, MI: University of Michigan Press, 1959.
Garland, H. The effects of piece rate underpayment and overpayment on job
performance: A test of equity theory with a new induction procedure. Jour-
nal of Applied Social Psychology, 1973,3, 325-334.
Gergen, K.J., Morse, S.J., and Bode, K.A. Overpaid or overworked? Cognitive
and behavioral reactions to inequitable rewards. Journal of Applied Social
Psychology, 1974,4, 259-274.
Goodman, P.S., and Friedman, A. An examination of Adams' theory of ineq-
uity. Administrative Science Quarterly, 1971,16,271-288.
Graen, G. Role-making processes within complex organizations. In M. D. Dun-
nette (Ed.), Handbook of Industrial and Organizational Psychology. Chicago:
Rand McNally, 1976.
Homans, G.C. Social Behavior: Its Elementary Forms. NY: Harcourt Brace
Jovanovich, 1974; London: Routledge and Kegan Paul, 1961.
House, R.J. A path goal theory of leadership effectiveness. Administrative Sci-
ence Quarterly, 1971,16,321-338.
Kahn, A., Lamm, H., and Nelson, R.E. Preferences for an equal or equitable
allocator. Journal of Personality and Social Psychology, 1977,35, 837-844.
Kahn, R.L., Wolfe, D.M., Quinn, R.P., Snoek, J.D., and Rosenthal, R.A. Orga-
nizational Stress: Studies in Role Conflict and Ambiguity. NY: John Wiley &
Sons, 1966.
Katz, D., and Kahn, D.L. Social Psychology of Organizations, 2nd ed. NY: John
Wiley, 1978.
Keirn, G. D. Corporate social responsibility: An assessment of the enlightened
self-interest model. Academy of Management Proceedings, 1977, 285-289.
Kidder, C.H., Bellettrie, G., and Cohn, E.S. Secret ambitions and public perfor-
mance. The effects of anonymity on the reward allocations made by men and
women. Journal of Experimental Social Psychology, 1977, 13, 70-80.
Larwood, L. Educating the law-abiding citizen-A psychological view. In G.C.S.
Benson (Ed.), Essays on Ethics and Law Enforcement. Lexington, Mass.:
Lexington Books, 1979.
Larwood, L., and Blackmore, J. Fair pay: Field investigations of the fair eco-
nomic exchange. A cademy of Management Proceedings, 1977,81-85.
- - . Sex discrimination in managerial selection: Testing predictions of the verti-
cal dyad linkage model. Sex Roles, 1978,4,359-367.
Larwood, L., Kavanagh, M., and Levine, R. Perceptions of fairness with three
alternative economic exchanges. Academy of Management Journal, 1978,21,
69-83.
Larwood, L., and Moely, B. Sex role and developmental evaluations in the just
world. Sex Roles, 1979,5, 19-28.
Larwood, L., O'Carroll, M., and Logan, J. Sex role as a mediator of achievement
in task performance. Sex Roles, 1977,3, 109-114.
Larwood, L., and Wood, M.M. Women in Management. Lexington, Mass.: Lex-
ington Books, 1977.
Larwood, L., Zalkind, D., and Legault, J. The bank job: A field study of sexually
discriminatory performance on a neutral role task. Journal of Applied Social
Psychology, 1975,5,68-74.
Lawler, E.E. Pay and Organizational Effectiveness: A Psychological View. NY:
McGraw-Hill, 1971.
Lerner, M.J. Observer's evaluation of a victim: Justice, guilt, and veridical per-
ception. Journal of Personality and Social Psychology, 1971,20, 127-135.
- - . The justice motive: Some hypotheses as to its origins and forms. Journal of
Personality, 1977,45, 1-52.
Lerner, M.J., Miller, D.T., and Holmes, J.G. Deserving and the emergence of
forms of justice. In L. Berkowitz and E. Walster (Eds.), Advances in Experi-
mental Social Psychology, Vol. 9. NY: Academic Press, 1976.
Leventhal, G. S. The distribution of rewards and resources in groups and organi-
zations. In L. Berkowitz and E. Walster (Eds.), Advances in Experimental
Social Psychology, Vol. 9, NY: Academic Press, 1976.
Mischel, W. Processes in delay of gratification. In L. Berkowitz (Ed.), Advances
in Experimental Social Psychology, Vol. 7, NY: Academic Press, 1973.
Pfeffer, J. Power and resource allocation in organizations. In B.M. Staw and
G.R. Salancik (Eds.), New Directions in Organizational Behavior. Chicago: St.
Clair Press, 1977.
Reis, H. T., and Gruzen, J. On mediating equity, equality, and self-interest: The
role of self-presentation in social exchange. Journal of Experimental Social
Psychology, 1976,12,487-503.
Roy, D. Quota restriction and goldbricking in a machine shop. American Journal
of Sociology, 1952,57,427-442.
Sarbin, T.R., and Allen, V.L. Role theory. In G. Lindzey and E. Aronson (Eds.)
Handbook of Social Psychology, Vol. 1, Second edition. Reading, Mass.:
Addison-Wesley, 1968.
Stogdill, R.M. Handbook of Leadership. NY: Free Press, 1974.
Walster, E., Berscheid, E., and Walster, G. W. New directions in equity research.
Journal of Personality and Social Psychology, 1973,25, 151-176.
Walster, G.W. The Walster et al. (1973) equity formula: A correction. Represen-
tative Research in Social Psychology, 1975,6,63-64.
Weick, K.E. Enactment processes in organizations. In B.M. Staw and G.R. Salan-
cik (Eds.), New Directions in Organizational Behavior. Chicago: St. Clair
Press, 1977.
Wilson, E.O. Sociobiology. Cambridge, Mass.: Harvard University Press, 1975.
19 A STEP OUT FROM THE MIDDLE
Thoughts Stimulated by Papers
of McKelvey, Bigelow, Behling,
and Larwood
Walter Nord,
Washington University
I will make two types of comments. First, I will summarize my thoughts as they
relate to the four individual papers. Then, I want to focus on some ideas that
these papers stimulated about middle range theory and organizational behavior
more generally.
Both sets of comments rest on several assumptions or biases. First, I share
Moscovici's (1972) preference for " ... any theory to absence of theory. (47)."
Second, I am persuaded by Washburn's (1978) view that " ... a theory is only
a license for research (407)." Third, I believe that theories, almost independent
of their ultimate success in coming to terms with reality, motivate research as
well as guide research and observation. Thus, I will take an eclectic position and
maintain that all of the orientations presented can have considerable benefit for
advancing organizational behavior. However, as will become apparent, I do tend
to see exchange theory as the more useful of the basic approaches advocated,
although this preference is undoubtedly a function of the fact that I am more
familiar with this approach than I am with the others. Having confessed my
initial biases, I will examine the papers.
239
MCKELVEY'S PAPER
The organizational speciation and the general ecological approach were relatively
new territory to me. Consequently, I found McKelvey's paper extremely instruc-
tive in laying out some of the basic concepts and terminology. While I was
bothered about the general applicability of this approach, some of this concern
was eased by Bigelow's view that we treat evolutionary theory itself as a middle
range theory that is differentially applicable; (i.e., it will be more useful in some
situations than in others). Still, it remains to be seen how widely the species
notion is useful as an analogue for organizational processes.
Focusing specifically on McKelvey's paper, I would like him to elaborate on
two interrelated points, both of which have to do with the species analogue for
organizational types. The first point is methodological. The development of
good taxonomies is most feasible where: (a) it is possible to make repeated ob-
servations on the relevant objects, (b) the observations can be made in ways that
appear to have little effect on the basic characteristics being measured, and (c)
the observations can be repeated relatively independently of environmental con·
ditions. Consequently, it is possible to pick these objects up, to look at them
from all angles, to create them, to take them apart, to subject them to all types
of tests, and to measure them in relatively precise physical terms. To the degree
that concepts and theories are a function of the measurement and observation
techniques possible, it would seem that the evolution of taxonomies is much
better suited for inquiry in research niches such as those which exist in biology
than those in which the study of organizations must occur.
Undoubtedly, there are at least a few aspects of organization that are sus-
ceptible to such analysis. McKelvey makes a good case for its value for dealing
with technology. I suspect that structure is another leading candidate. How
many others are possible? How many are necessary for this perspective to be
worthwhile? Adequate answers to these question will be easier to arrive at when
the paradigm has been applied more widely. Consequently, I think McKelvey'S
position that his efforts should be viewed as beginning discussion on this topic is
the appropriate one. For the time being, the burden of proof is on the advocates
of the position and I suspect this burden will motivate some valuable research.
The second set of points I'd like McKelvey to address is more conceptual.
First, are social organizations as closely coupled with their environments as
biological ones? If not, what are the implications for the organization speciation
concept? Second, there are a number of characteristics of organizations and their
environments that are apt to be quite different from the comparable set of factors
for biological organisms. For example, how useful is the organization speciation
concept in very abundant environments? What changes in the model must be
made to deal with systems having the ability to create their own niches? Also, how
A STEP OUT FROM THE MIDDLE 241
constant must organizations be in order to properly classify them as species?

Moreover, in biology, I suspect the ratio of the total number of organisms
that exist in the world to the number of categories is a very large number (Le.,
total number of organisms . ...
number 0 f c1aSSl·filcaf lOns =large number). Thus, preClse speClatlon IS not only
possible but pragmatic. By contrast the comparable fraction for organizations
would seem to be much smaller. What must the fraction be for the concepts of
species to be worth the effort of developing the taxonomy?
I would like McKelvey to elaborate on one other issue. Early in the paper he
indicates five principles of inquiry. It would seem that dialectical analysis
(Benson, 1977) is a significant omission from this list. Was the omission delib-
erate? Does dialectical analysis qualify as another category?
BIGELOW'S PAPER
As I indicated earlier, I think Bigelow's paper is a valuable complement to

McKelvey's in that it suggests some useful limits on the evolutionary model as
a tool. Since I stated some of my concerns about the biological model in general
in my discussion of McKelvey's paper, my comments on Bigelow's paper will be
briefer.
Early in the paper Bigelow indicates a strength of the evolutionary perspective
is its utility as a model having strong potential for practical application. However,
he did not demonstrate this point throughout the paper. Instead, most of his
attention was devoted to developing a four-dimensional classification scheme.
While this classification scheme may be very useful for academics, I fear that the
movement from stage to stage may be very difficult for a practitioner to observe
and that the definition of niche may vary very rapidly. Consequently, I am skep-
tical about how useful most managers would find this particular model and pes-
simistic about our ability to develop tools in the foreseeable future that will
permit ready applicability.
Secondly, I think both Bigelow and McKelvey make an important point in
stressing the importance of dealing with historical processes in the analysis of
organizations. I would urge, somewhat idealistically, that such analysis be as
broadly based as possible to avoid the tendency to take niches as relatively fixed
and hence to take the current forms of organization as inevitable. Instead, the
concept of niches and their creation can be used productively-to create the
type of organizations we encourage and permit to survive. Historical analysis
may help us to perceive alternative possibilities and to guide interventions at
the level of the superstructure.
Finally, the borrowing of biological concepts by McKelvey and Bigelow
stimulated me to think about what other concepts we might borrow from the
more "hard" sciences. Although I am generally bothered by the widespread
practice of mindless borrowing of models from other diSCiplines, I will offer one
other candidate-subatomic physics (See Capra, 1975). In many ways the images
that subatomic physicists hold of type of processes, changes, indeterminancies,
and relationships of subatomic "particles" to the field in which they exist, seem
to be very analogous to the ways organizations behave-perhaps more so than
the more static concept of species. This type of thinking seems to be consistent
with some of Bobko's points.
BEHLING'S PAPER
The pros and cons of functional analysis have been widely discussed in sociology,
and I have little to add to the controversy except to stress that many of the
criticisms of functionalism attack aspects that are not necessarily a part of func-
tionalism but rather the way functional analysis is used.
Behling's perspective, which views functionalism as a state of mind as op-
posed to a unified technique, seems to be the most fruitful use of the model.
Throughout the paper he uses the model in this way and avoids the fruitless de-
bates triggered by more absolutistic views. Moreover, in much the same way as
Bigelow observed about evolutionary theory, Behling suggests that functionalism
may be more useful in some situations and for some purposes than for others.
However, this still leaves some important questions unanswered: In how many
situations is functional analysis useful? For what purposes is it useful?
With respect to the breadth of use for organizational analysis, Behling writes
"functional analysis can be applied only to systems where functional unity can
be demonstrated." He acknowledges that this, along with the other constraints,
restricts the scope of functional analysis. Given the work of the scholars such as
March, Weick, Cyert, and others, that Behling cites, the functional unity require-
ment seems to be a serious restriction. If organizations are in fact viewed as
political coalitions, loosely coupled systems, and so on, does functional analysis
help us very much?
Second, to the degree that organizations are composed of a variety of at least
partially conflicting interests, the notion of arguing for functional analysis based
on the merits to the managers who control the system seems to me to be prob-
lematic both practically and ideologically. On a pragmatic level, every particular
structure or action may be functional for a number of different interests and
may be functional for some and dysfunctional for the others. While for some
purposes it is useful to look at the functions and dysfunctions from only one set
of interests, such analysis may underemphasize important organizational pro-
cesses. For example, when organizations are observed through a functional lens
that emphasizes the interests of those in charge, expressions of conflicting inter-
ests, and "deviant" acts (e.g., theft and industrial strife) are likely to be viewed
as dysfunctional and "bad." As a result, analysis that leads to an understanding
of their functions for a number of other interests is apt to be rare.
Of course, functional analysis can be used to examine all the different inter-
ests that are served or restrained by particular actions. However, usually such
analysis is called conflict theory rather than functional theory. Moreover, it is
difficult to apply when the number of competing interests is large.
Two final points about Behling's paper. First, I think it is important to recog-
nize that in sociology, functionalism has frequently been attacked for its con-
servative bias. While I want to stress the fact that a conservative bias is not
inevitably a part of functionalism, it tends to operate that way. I think Behling's
emphasis on the value of this approach, because we are interested in the concerns
of those who are in charge of organizations, demonstrates this tendency and runs
the risk of viewing actions that challenge the existing order as illegitimate. For
example, consider Behling's statement that "Functionality per se cannot be used
to justify the existence of structures." Logically this is correct, but de facto it is
incorrect-the functional point of view is used in this very way. Finally, I hope
Behling will elaborate more fully on how functional theory will aid in the appli-
cation of behavioral science analysis. I agree that the concepts of manifest and
latent consequences and of functions and dysfunctions can be extremely prac-
tical. These, however, have been in our textbooks and training programs for
years. What other insights can we expect practitioners to get from functional
analysis?
LARWOOO'S PAPER
As I noted earlier, I personally tend to see exchange theory as a very useful

framework for integrating diverse sets of materials concerning organizations. The
exchange notion can be oriented to describing the origins, maintenance, and
breakdown of social order and social norms.
Larwood's use of the approach seems to be limited to cases where stabilizing
norms exist and is oriented to describing behavior given these norms. This em-
phasis stems from her third assumption that individuals trade with others accord-
ing to normative exchange patterns. I think this is an unnecessarily restrictive
assumption. While normative exchange patterns do characterize many social
relationships, many other social relationships are characterized by nonnormative
aspects as well. As Ellis (1971) has observed, the normative aspects themselves
require explanation, and exchange theory can be useful in doing so. Moreover, in
a number of situations (such as under conditions of scarcity) it is prudent to

violate these norms, and restricting exchange theory to behaviors governed by
existing norms reduces the likelihood that certain types of exchanges (e.g.,
coercive exchanges) will be analyzed. I think the tendency is evident in some of
Larwood's analysis, where often she recognizes that coercive and involuntary
elements occur but then tends not to analyze these processes further.
A second issue I would like Larwood to comment on in more detail has to
do with the particular notion of justice and fairness used in the paper. As Heath
(1976) observed there are a number of criteria for justice. For example, justice
may be based on needs or on input. While Larwood stresses the fact that justice
depends upon the perception of the individual in a situation, she emphasizes
justice and fairness as a function of rewards to inputs as opposed to other pos-
sible criteria. Some of the most interesting events occur in organizations when
different people apply different criteria. Although Larwood does mention sab-
otage and other reactions, her emphasis is on the fact that these are temporary
conditions because these people will leave the system as soon as possible-that
is, conflicting norms tend to be dealt with by exit or restricted entrance in
Larwood's analysis. If my interpretation is accurate, I hope Larwood will com-
ment on the ability of exchange theory to deal with on-going conflicts involving
different people using conflicting norms.
I want to raise three other issues. First, to what degree is Larwood's notion
of exchange theory different from expectancy theory? Is the only difference
that fairness is included in one's calculations? Second, to what degree is the
utility of exchange theory as an analytical tool affected when a variety of inter-
ests and objectives are assumed? While exchange theory can account for many of
the formal and informal exchanges that are occurring simultaneously, this can
become a very laborious task. Also such an extension can make it impossible to
logically arrive at any conditions that might refute the theory. Third, there seem
to be two general sources of injustice. One occurs when Person gets less than hel
she feels is appropriate with respect to some given norm. The other occurs when
Person and Other are using different norms. Does exchange theory deal equally
well with both cases?
Despite these concerns, I think Larwood has done much to show the value of
exchange theory. I suggest that exchange theory, stripped of at least some of its
normative notions, provides an extremely valuable tool for analyzing organiza-
tions at all levels of analysis. I assume that the major processes that occur in
organizations involve the flow of resources to different groups and individuals,
for different purposes, under different conditions. Exchange theory provides a
set of concepts that appears to be extremely useful for focusing on these pro-
cesses. While the current uses of exchange theory appear to stop short of realizing
the true potential of this approach, exchange analysis appears to lead directly to
a political economy of organizations, which I believe is the lowest level that

various middle range theories can be integrated and still account for a large
number of important organizational processes. Exchange theory seems to pro-
vide an orientation and a partial set of concepts to bridge the micro-macro
interface, to account for events at both micro and macro levels, and to deal with
both stability and conflict and with the origins, maintenance, and decay of
social order.
GENERAL CONCLUSIONS
Having dealt with the papers one by one, I would like to mention several general
feelings I had about middle range theory and organizational behavior as I was
reading these papers.
First, there seemed to be a tendency for this particular set of papers, and I
expect other papers born in the traditions these papers stem from, to be written
from the perspective of the managerial elite. As such, they have many manifest
and latent eufunctions. However, they may sacrifice our understanding of pro-
cesses that conflict-oriented approaches or approaches written from different
ideological or interest perspectives would reveal.
Secondly, although most of the writers were interested in bridging the gap
between theory and practice, I got the clear impression that theory was more
important than practice. It seemed as if all authors were assuming that a theory
integrating the various middle range theories in an intellectually satisfying way
would be a practical theory. The assumption that there is a close coupling
between the intellectual merits of a theory and its utilization in managerial
practice seems to me to be dubious. I suggest that we need to study the utiliza-
tion of knowledge if we are really serious about bridging the theory-practice gap.
None of the papers really dealt with this issue.
Further, there seemed to be an implicit assumption that the models derived
from other social or biological sciences are the most useful ways to integrate
various middle range theories. However, I think it is important to stress that
the theories of scientists are only one way of "punctuating reality" (Weick,
1977). The problems experienced by various interest groups in a social system
are another way of punctuating reality. Particularly if utilization is the major
point of interest, I would offer that the study of a variety of problems as they are
experienced by members of social systems may be more valuable for integrating
various middle range theories than are additional layers of abstractions laid on
by intellectuals. In this regard, I find C. Wright Mills' (1959) attack on both
grand theory and abstract empiricism very relevant. One strategy for being
"relevant" as well as developing meaningful social science, is to study problems

as seen by participants in various social settings.
Fourth, the role of analogies borrowed from other disciplines seems to war-
rant some detailed analysis. Are there some criteria that might be developed to
distinguish "good" borrowing from "bad" borrowing? Oftentimes organizational
behaviorists and other social scientists seem to borrow concepts in a rather
thoughtless manner. The differences in the subject matter between the "lending"
and the "borrowing" discipline are seldom considered. The borrowed concepts
become distorted and reified in the new discipline in ways that are quickly dis-
owned by students in the lending discipline. Thus, the borrowed models are easy
prey for critics and are rejected-perhaps prematurely. As a result, their true
value is never seriously considered and much time and energy is spent for little
gain.
Finally, these papers, like much current work in our field, gave little attention
to some basic value issues. As Tom Mahoney suggested, we give little attention
to what or who organization theory is for. Similarly, we need to ask what or
who are organizations for? These questions show the need for what Ken Benson
called for-a more detailed look at the creation process of organization theo-
rizing and research. The quest for a middle range theory of middle range theory
may be a very productive one.
REFERENCES
Benson, J.K. Organizations: A dialetical view. Administrative Science Quarterly,

1977,22, 1-21.
Capra, F. The Tao of Physics. Boulder, Colo.: Shambhala Publications, Inc.,
1975.
Ellis, D.P. The Hobbesian problem of order. American Sociological Review,
1971, 36, 692-703.
Heath, A. Rational Choice and Social Exchange. A Critique of Exchange Theory.
London: Cambridge University Press, 1976.
Mills, C. W. The Sociological Imagination. New York: Oxford University Press,
1959.
Moscovici, S. Society and theory in social psychology. In J. Israel and H. Tajfel
(Eds.) The Context of Social Psychology. A Critical Assessment. New York:
Academic Press, 1972, 17-68.
Washburn, S.L. Human behavior and the behavior of other animals. American
Psychologist, 1978,33,405-418.
Weick, K.E. Blindspots in organizational theorizing. Paper presented at 37th
Annual meeting of the Academy of Management, August 16, 1977.
20 A REJOINDER TO NORD
Orlando Behling,
Nord raises several important points regarding "Functionalism as a Basis for Mid-
range Theory in Organizational Behavior/Theory," four of which clearly deserve
some sort of response. He holds that:
1. The applicability of functional analysis may be severely limited if we

require that functional unity be demonstrated before it is applied to a
particular organization or phenomenon.
2. Functionalism emphasizes the merits of particular activities for managers
who control the organization, rather than leading to an understanding of
their merits for a variety of additional interests.
3. Functionalism has a conservative bias; that is, it tends to be used to justify
the continuation of present practices in the face of challenges to the exist-
ing order.
4. The benefits of functional analysis as an alternative to more conventional
frameworks for the study of organizations needs to be demonstrated more
clearly.
Let me comment on each of these, one at a time.
247
THE REQUIREMENT OF FUNCTIONAL UNITY
Nord and 1 agree on the necessity of functional unity as a precondition for func-
tional analysis, as anyone who fully understands functionalism must, since it is
a requirement practically by defmition. What we disagree on is the likelihood
that this requirement reduces the class of organizations and phenomena subject
to functional analysis to a null set. Nord cites the works of Cyert and March
(1959) and Weick (1969), critical of the traditional view of organizations as ex-
tensions of the wills of single owner-managers in support of his position, but
ignores the increasingly sophisticated work on "technology" (e.g., Aldrich 1972)
which argues that organizations vary markedly in the degree of interdependence
of their components and this interdependence is crucial in understanding their
operation. The functional unity of a particular organization must remain an em-
piric question and with it the usefulness of functional analysis as a tool for
exploring it.
FUNCTIONALISM AS DISCUSSED EMPHASIZES

MANAGERIAL CONCERNS
Theoretically, Nord's criticism is incorrect. Functionalism as described here em-

phasizes organizational survival and goal attainment, which are positive outcomes
for many of those who participate in its activities besides managers. At a prac-
tical level however, his criticism is correct; often there is a split between the
perceived short-run interests of management and those of other participant
groups. When this is the situation, functionalism as described here views organi-
zations from a perspective that is more acceptable to management than to others.
My response can only be in the words of the punch line from an old, mildly
obscene, blatantly sexist joke which 1 won't quote in full: "I understand the
situation, but 1 don't see 'the problem." A quick check of the roster of partici-
pants in this conference reveals at least 38 of the 42 are teachers of "manage-
ment," "business," "business administration," or "commerce." The teaching of
administrators is an honorable profession. 1 believe that functional analysis can
make that process more effective by yielding important insights into organiza-
tions and thus I'm in favor of it.
FUNCTIONALISM HAS A CONSERVATIVE BIAS
Functionalism can and has been misued to justify the status quo in the face of
demands for change, as Nord indicates. Clearly, this is a risk that must be kept in
A REJOINDER TO NORD 249
mind in the application of functional analysis to the concerns of specific organi-

zations, but it is not a basis for a blanket prohibition of its use in applied settings
nor certainly as an analytical tool in scholarly research.
CLEARER DEMONSTRATIONS OF THE BENEFITS OF

FUNCTIONAL ANALYSIS ARE NEEDED
The potential benefits of functional analysis are obvious. If we accept the as-
sumption that it is useful to think of some organizations, some times as entities
established to accomplish particular ends, functional analysis has the major
advantage of yielding results in which, unlike conventional invariant order
explanations, prescription is readily apparent; few intellectual leaps are required
to move from research to practice. It also has the potential advantage of focusing
scholarly attention on issues of practical as well as academic significance.
Unquestionably, however, the crude demonstration of the usefulness of
functional analysis that concluded my paper is adequate to demonstrate its
potential only in a very limited fashion. It can be hoped, however, that this
paper will stimulate additional attempts to test its usefulness.
REFERENCES
Aldrich, H. Technology and organizational structure: A reexamination of the

findings of the Aston Group. Administrative Science Quarterly, 1972, 17,
26-43.
Cyert, R.M. and March, J.G. A behavioral theory of organizational objectives. In
M. Haire (Ed.), Modern Organization Theory, New York: Wiley, 1959.
Nord, W. A step out from the middle - Thoughts stimulated by the papers of
McKelvey, Bigelow, Behling and Larwood. In C. Pinder and L. Moore (Eds.),
Middle Range Theory and the Study of Organizations. 1979.
Wesley, 1969.
21 REJOINDER TO NORD
Some Other Parts
of Exchange Theory
Laurie Larwood,
Clarement Men's College and
Clarement Graduate School
It is always pleasant to find one's self in agreement with one's critic. Professor
Nord's very perceptive examination of my paper has pointed to certain short-
comings-in the paper, but not in the theory of exchange developed there.
Limitations of space confined my consideration to only a few of the many
applications of exchange theory. As Professor Nord suggests, these were primarily
cases of the "fair exchange" -situations in which the parties to the exchange ob-
tain what they expect and agree is fair. Fair exchanges characterize ongoing
voluntary relationships; in contrast, any failure to obtain what one expects
might force a reconsideration by at least one of the participants.
Situations involving unjust, unfair, unpredictable, or unevenly perceived ex-
changes have been extensively elaborated in Larwood (1979) and will not be
detailed here. As suggested, however, the uneven or unjust exchange is of con-
siderable social importance. For example, an employee is generally engaged in a
fair exchange with the employer. However, if the individual's pay is unex-
pectedly cut or if the employee does not obtain the recognition expected, he or
she is plunged into the uneven exchange. The desire for justice may motivate a
number of activities, probably beginning with employer-employee negotiations
or resignation if alternative employment is readily available.
If negotiations fail and the employee is locked into the organization, other
250
SOME OTHER PARTS OF EXCHANGE THEORY 251
alternatives come into play. These have the effects of returning the employee to
the original position and/or denying the employer the fruits of the uneven ex-
change. In this manner, honest, hard-working, and well-meaning employees can
readily justify strikes, slowdowns, sabotage, theft, or unearned overtime. After
all, they deserve increased benefits, and the employer is not paying for what she
or he is getting. Moreover, if such safety valves are unavailable, one could predict
the growth of a dissatisfied group of workers who might be willing to engage in
major social change.
In this sense, an unjust or uneven exchange is unstable-one or more of the
parties will seek adjustment until all participants agree that the relationship
apparently achieved has become fair. Of course, exchange partners may perceive
the same relatonship differently, may attach different values to their resources,
or may covertly obtain a fair (to themselves) exchange in a manner unrecognized
by others. Thus, the determination of what is uneven rests on the participants;
outside observers can ask, but they cannot-at least at this point in development
of the theory -assert accurately what the terms are of any exchange and whether
that exchange is fair or unfair (Larwood, Kavanagh, and Levine, 1978; Larwood,
Levine, Shaw, and Hurwitz, 1979).
Professor Nord suggests that the reliance on exchange norms is unnecessarily
restrictive. Although not made explicit in the earlier paper, the term "norm" is
intended to be used in a broad sense. While society-wide norms exist for many
types of exchanges, adjustment in those norms is readily made. Since, as above,
individuals differ from one another in the particular resources they have to offer,
in those they value, or in their experiences and perceptions, one person may be
more willing than another to engage in an exchange based on consensual norms.
Any exchange that is intended to be fair with the less willing individual must
then be based on modified standards acceptable to both parties. The modifica-
tions become a part of the new norm for this exchange-they must now be met
for an ongoing relationship to take place.
Given the obvious potential complexity of multiparty exchanges involving
qualitatively different resources (communication, roles, pay, etc.), does ex-
change theory offer much utility of analysis? A fair exchange under such con-
ditions is an interlocked matrix, which cannot be completely comprehended by
reference only to a dyad or to one or two varieties of resource. Nonetheless, it
is easily possible for certain elements to be eliminated from usual consideration.
For example, certain resource inputs are more important to most persons in
specific sets of circumstances (Larwood, Levine, Shaw, and Hurwitz, 1979).
Similarly, people tend to interact with certain individuals more often and more
intensely than with others (see for instance, social comparison theory). Finally,
the particular nature of the exchange often depends on the situation (Larwood
and Blackmore, 1977; Larwood, Kavanagh, and Levine, 1978). As a consequence,
the matrix problem can be decomposed fairly readily into a set of approxima-
tions; presumably, as we learn more about the mechanism of exchange theory,
we will become more adept at selecting the most valuable approximations.
How does all of this differ from expectancy theory? The exchange conceptu-
alization emphasizes the links between self-interest and justice which must be
obtained before the individual can successfully operate in society (Larwood,
1979). When, for instance, the personal contract fails to develop adequately, the
individual is incapable of entering exchanges requiring long-term planning.
Further, exchange theory emphasizes an interpersonal normative process as
opposed to the individual self-interest model of expectancy theory. Of course,
expectancy theory can be modified to take into account many of the same
aspects as exchange theory; when modified, expectancy theory becomes a con-
venient vehicle by which we may operationalize Bandura's (1977) concept of
self-efficacy in the exchange process.
Nonetheless, expectancy theory is merely one potential middle range com-
ponent of exchange theory. Researchers who prefer Maslow's hierarchy, for
example, could as readily define self-efficacious behavior as that which fulfills
successively higher motivational levels; they can interpret injustice as any threat
to that fulfillment. The selection of expectancy theory as a middle range theory
encompassed by exchange theory was based on the evident empirical power of
expectancy theory rather than on any unique fit. Exchange theory is compatible
with a number of other motivational concepts, but does not stand or fall with
any of them.
Finally, Professor Nord's comments that we must bridge the gaps between
theory and use, and between the managerial elite and the managed are vitally
important. Some exchange research is already available to indicate how em-
ployees, as opposed to employers, feel about the exchanges in which they are
engaged (Larwood and Blackmore, 1977), or to determine at a micro level what
the important inputs and outputs seem to be (Larwood, Kavanagh, and Levine,
1978; Larwood, Levine, Shaw, and Hurwitz, 1979). So far, the principles do
seem to hold up.
While laboratory and organizational survey studies have in the past contrib-
uted most to the advancement of psychological and organizational theory, case
studies have been largely ignored. Exchange theory would seem to have wide
social implications, however, and case study may be the most rapid way to test
and expand this form of analysis. As examples, the California property tax re-
volt can be studied as the social result of a perceived failure in the expected
exchange; European worker occupations of plant sites may be interpreted as
the workers' failure to authorize the use of managerial power in an exchange;
the bloodless invasion of the Spanish Sahara by Morocco may have succeeded
because of a breakdown in the military role exchange for which the Spanish
SOME OTHER PARTS OF EXCHANGE THEORY 253
troops were prepared. In the absence of closer scrutiny, these suggestions are
rather wild speculations. The conclusion, still, is that a great deal of research of
every kind is needed before exchange theory can be accepted as a reliable and
predictive tool in broad organizational and social analysis.
REFERENCES
Bandura, A. Self-efficacy: Toward a unifying theory of behavioral change.

Psychological Review, 1977,84, 191-215.
Larwood, L. Educating the law-abiding citizen-A psychological view. In G.C.S.
Benson (Ed.), Essays on Ethics and Law Enforcement. Lexington, Mass.:
Lexington Books, 1979.
Larwood, L., and Blackmore, J. Fair pay: Field investigations of the fair eco-
nomic exchange. Academy of Management Proceedings, 1977,81-85.
Larwood, L., Kavanagh, M., and Levine, R. Perceptions of fairness with three
alternative economic exchanges. Academy of Management Journal, 1978,
21,69-83.
Larwood, L., Levine, R., Shaw, R., and Hurwitz, S. Equity or exchange theory?
Effects of varying inputs on selection of the equity, equality, and winner-
take-all exchanges. Organizational Behavior and Human Performance, 1979,
23, 60-72.
IV EXAMPLES OF
MIDDLE RANGE THEORY
22
MIDDLE RANGE
ORGANIZATION THEORIZING
Role Theory as an Example
Mary Elizabeth Beres and Karl F. Price,
Temple University
The purpose of this paper is to examine the nature of middle range organization
theory, to identify some of its major strengths and limitations, and to suggest
some guidelines for middle range theorizing. Role theory will be used as an illus·
tration throughout the discussion. This area of theory has been identified as
middle range by Merton (1968) and is likely to receive increased attention
through the new edition (1978)" of Katz and Kahn's The Social Psychology of
Organizations.
The concept of middle range theory was introduced by Merton (1949) as a
way of developing a better understanding of social processes. He suggested that
grand theories, which attempt to explain all phenomena, are often so abstract
as to offer little insight into specific situations. On the other hand, ungeneralized
knowledge of specific phenomena is not transferable. Hence, Merton suggested
that middle range theory be constructed from empirical data to develop gen-
eralizations that would be applicable to specific aspects or categories of sit-
uations. Thus, middle range theories would offer explanations of limited types
of social phenomena.
Since organizations are a specialized form of social unit, a general theory of
organizations would be, from Merton's perspective, a middle range social theory.
It appears then, that in proposing the development of middle range organization
257
theory, yet another step is being added in a sequence of abstraction from specific
data to grand social theory. From this perspective, a middle range theory of
organizations would be a set of related generalizations that: (a) link a limited set
of organizational variables, or (b) describe a limited set of organizational pro-
cesses, or (c) explain characteristics of special types of organizations.
Adding a step to a sequence of abstractions describing social processes is not
as straightforward as it may at first seem. Because organizations are one of a
variety of social aggregations (Blau and Scott, 1962; Weber, 1968), some of the
processes that occur within them also occur in other contexts. If a set of general-
izations addresses phenomena shared by organizations and other aggregates, then
the generalizations are both middle range theories of organizations and middle
range theories of more general social processes. Only theories that address pro-
cesses shaped by the differentiating characteristics of organizations can be exclu-
sively middle range organization theories. Since, at the present time, there is
disagreement about the distinguishing properties of organizations (Georgiou,
1973) and there is no generally accepted way to identify organizational bound-
aries, any sets of generalizations that refer to limited elements of organizations
or limited types of organizations will be considered here as middle range orga-
nization theories, whether or not the generalizations apply elsewhere.
An absence of a commonly agreed upon concept of organizations creates a
second problem in the identification of middle range organization theories. A
subset of phenomena can be identified only in relation to the whole set. Because
of disagreement on what constitutes "the organization," one theorist's middle
range theory may be another's organization theory. Resolution of this issue is
outside the scope of this discussion. For the purposes of proceeding, however,
the organization is considered here as consisting of Leavitt's (1965) set of com-
ponents, that is, people, structure, technology and task, augmented by the open
systems element of relations with the task environment (Thompson, 1967). In
this context, a general theory of organization would have to account for all the
relations in all types of organizations among all the organizationally significant
properties of people, structure, technology, task, and environmental relations.
Any theory that only accounts for some of these relations, or the relations in
special types of organizations, is a middle range theory of organization.
Two effects follow from the nature of middle range theory. First, identifying
the range of a theory focuses attention on the scope and limits of a set of gen-
eralizations. Second, advocating middle range theorizing gives legitimacy to the
study of partial aspects of the complex organization phenomenon. The remainder
of this paper addresses some of the implications of middle range theorizing in
terms of these effects.
MIDDLE RANGE ORGANIZATION THEORIZING 259
CONTRIBUTIONS OF MIDDLE RANGE THEORIZING
Hall (1972) has emphasized the need for a typology to give conceptual order to
the vast amount of organizationally relevant research already in existence. By
requiring the establishment of the scope of general theory in order to identify
less than general theory, middle range theorizing contributes to the development
of such a classification system. The range set for general theory provides the
dimensions for characterizing theories. Comparing a theory with these dimen-
sions indicates its range and, at the same time, its limits, making it possible to
identify a theory as middle range.
The process of classifying theories can be illustrated by using the general
frame of reference proposed earlier to describe the scope and limits of role
theory. First, a summary of role theory is in order.
Although identified as an area of theory by Merton (1968), Biddle and
Thomas (1966) in their book entitled Role Theory indicated that there is no
one grand "role theory." They state further that, "A close examination of that
which is regarded as 'role theory' indicates that its statements appear in essen-
tially three forms: (a) as single hypotheses, (b) as sets of logically unrelated
hypotheses on the same topic, and (c) as sets of logically as well as topically
related hypotheses" (p. 14). Only the last of these would qualify as a middle
range theory.
Katz and Kahn (1966) offer one such set of generalizations concerning roles
in their theory of organizations. As they view it, the role concept is a set of
logically and topically related generalizations attempting to explain human inter-
action in an organizational setting. The Katz and Kahn model of role is related
to that of Gross, Mason, and McEachern (1958) in that it focuses on the expecta-
tions held for the behavior of the focal person. It is largely a social-psychological
concept even though the impacts of the organization, as well as individual dif-
ferences, are included as influences in the basic role episode, which is largely an
interpersonal phenomenon. The Katz and Kahn role episode was derived from the
earlier work of Kahn, et aI., (1964) entitled Organizational Stress (See Figure 1).
The role episode begins with the role senders holding expectations for the
focal person. These expectations are influenced by the focal person's previous
performance and by organizational factors such as structure, technology, rewards
and punishments, and politics. The role senders convert their expectations into
influence attempts, such as shouted demands or a pat on the back, and com-
municate them to the focal person. These influence attempts are perceived by
the focal persons, and, distorted as they may be, are interpreted in view of his/her
perception of all the other expectations held by his/her role set. The focal per-
tv
N
0\
0\
o
Attributes of
the person
......
......
,
//
/ " "-"-\
/ \
I \
\l.,- I \
Role Senders I Focal Person
Expectations Sent Role

I Received Role Role Behavior
Organi- I
zational '1/,
't/,
Factors ,, Perception of Information Perception
Perce?tion of Compliance
Com?liance I
I---
-
.;
...,"" /f\ ... role, and resistance
II focal person's attempts /1'''-
behavior; at influence perception of "side effects"
evaluation I role sending
I
,~ I
'I' ,
I
!! I I
I
I 1I
\\ I
I/
\\ I/
\ //
\.
\. /
/
"-
"- ,
//
""
Figure 1. The Role Episode.
son then responds or copes in accordance with his/her own expectations and
perception of the role senders' expectations.
The model shows that the focal person's performance affects the role senders'
expectations for the focal person, renewing the episode. Furthermore, certain of
the focal person's reactions to, or feelings about the role sendings such as ten-
sion, may affect his/her personality. Certain other reactions, such as loss of
respect for role senders, may affect the interpersonal relations between the focal
person and his/her role senders. The focal person's personality and interpersonal
relations between role senders and focal person may, in turn, affect the expecta-
tions and influence attempts of the role senders and may also affect the percep-
tions and reactions of the focal person. This model of the role episode will be
used for the remainder of the discussion to illustrate the process of middle
range theorizing.
As indicated earlier, identifying the range of an organization theory involves
comparing its scope with the range of general organization theory. The latter,
according to the enlarged Leavitt framework, would include all organizationally
relevant relations within and among organization members, structure, technol-
ogy, task, and environmental relations for all types of organizations. In relation
to this general range, we can see that the role episode is concerned with a limited
number of variables. It is largely social-psychological in nature focusing on the
interaction between role senders and focal persons. The fact that environmental
factors may impact on the relationship is recognized, but the nature of the influ-
ence is not specified by the model. Socio-technical aspects of organizations are
largely ignored, making the model more applicable for the study of organizations
characterized by significant amounts of human interaction. By the same token,
the influence of individual differences is recognized, but not explained in the
model. Thus, the effects of personality and social environment [e.g., urban ver-
sus rural environment (Blood and Hulin, 1967)] are not addressed. Finally, the
potential effects of task properties, such as degree of ambiguity, on role sending
and receiving are never specifically treated. In summary, the role model can be
seen mainly to address relationships between organization members, a scope that
makes it middle range in comparison with the previously described general range.
In order to make such a comparison, however, the general range must be speci-
fied. Thus, attention to midrange theorizing increases the need to agree upon a
typology identifying the scope of general organization theory.
A second contribution of middle range theorizing is in the area of theory
development. By focusing attention on the scope and limits of theory, this ap-
proach provides for a clearer targeting of research. The scope of the theory
translates into criteria for research design decisions in the areas of sample selec-
tion and data collection. Identified limits indicate areas where controls need to
be considered. By clearly restricting the focus of research, middle range theo-

rizing also encourages more rigorous operationalizing of the generalizations
involved. The researcher is able to concentrate on a limited number of variables,
rather than the total complexity of the organization. In the case where theory
is derived from data, the rigor introduced by the emphasis on theory results in
a more precise identification of the limits of generalizability. The impact of
middle range theorizing on research design and theory development can be seen
in several implications the approach has for research on role theory.
Translating the limits of the role model presented earlier, the areas that would
have to be controlled in order to gain research comparability would be individual
differences, interpersonal relationships, and organizational factors. Differences
in terms of group characteristics (e.g., cohesiveness, norms, leadership), would
also have to be considered and probably controlled in any role theory research.
In fact, group processes such as sanctions, rewards, and norms may well be an
alternative explanation for outcomes attributed to interactions in a role set. In
addition, the emphasis of the theory on SOcial-psychological processes would
make it less appropriate for studying organizational components in which activ-
ities are machine intensive.
Middle range theorizing makes a third contribution by providing a context
for the integration of theories. At the present time, theories concerning organiza-
tions can be drawn from a great variety of disciplines, induding all of the be-
havioral and social sciences, industrial engineering, and management disciplines,
among others. By identifying the limits of anyone theory, the middle range
framework indicates the potential intersections between a given theory and
other theories. In the case of role theory, for example, we see that the role
episode focuses largely on the interpersonal aspects of the transmission of ex-
pectations. While perceptual processes are involved, the role literature tends to
ignore the impact of perception on the accurate reception of sent expectations.
If considered, the process of perception is often treated as a source of role
conflict in that it causes problems in reception. There is, however, a significant
body of psychological literature concerning perception, and the intersection of
these two areas of theory helps to build a more general model of organiza-
tional processes.
The impact of technology on organizational structures is also largely ignored
in the role model. We know that changes in technology may bring about changes
in structure, jobs, relationships, and so forth, but except to show that organiza-
tional factors impact on the role episode, the nature of technology's effect on
communication and perception is unspecified. Again, if the socio-technical
models could be integrated with the social-psychological ones, a more complete
theory of organizations would emerge. In like manner, each of the limits of a
midrange theory indicates a relevant variable or process that is not explained
within the theory. Other theories, that focus on the particular variable or pro-
cess, can be linked to the midrange theory according to the relationship specified
at the limits, thus, extending the range of explanation. In this sense, the mid-
dle range approach contributes to the development of a general theory of
organizations.
Through a recognition of the overlap between social processes within and
outside organizations, middle range theorizing can contribute to an under-
standing of the social nature of organizations. Theories that apply across the
boundaries indicate the extent to which organizations reflect general social
properties. Whereas, theories applying only within organizations indicate the
distinctive properties of this specialized form of collective activity. By focusing
on the basic interpersonal relationship between two individuals, the role model
can be used in studying any form of social collectivity. For example, relation-
ships between family members are as amenable to study using the role episode
as is the relationship between a superior and his/her subordinate. As such, "role
theory" is not only a middle range theory of organizations but also a middle
range theory of general social structures (Merton, 1968).
Middle range theorizing makes a fifth contribution in the area of contin-
gency theory development. The typological framework used to identify a theory
as middle range can also be used as a frame of reference for cataloging research.
Examination of such a research survey would show the areas in which research is
lacking. In helping to identify these areas, middle range theorizing would focus
attention on the gaps where research is needed and, thereby, provide a guide
for continuing research efforts.
As indicated earlier, significant research in the role literature is related to
single hypotheses or logically unconnected sets of hypotheses. If this research
were organized in terms of a common frame of reference, potential linkages
could be identified-expanding the scope of a logically coherent role theory. At
the same time, gaps needing further study would be highlighted.
A final contribution of middle range theorizing is in the area of theory appli-
cation. One of the major uses of organization theory is in the design and manage-
ment of organizations. From a contingency perspective, however, all theory is
not applicable to all situations. Relevance depends on a fit between the subject
of the theory and conditions in the organization. By focusing on the scope and
limits of theory, middle range theorizing helps to define the range of applicability
of a theory. Attention to this range can help to counteract the bandwagon
tendency to adopt practices simply because they have become popularly known
as successful. In the case of role theory, the discussion so far indicates that role
theory research would be most applicable in organizations characterized by high
levels of person to person interaction and where behavior is minimally con-
strained by technology and structure.
In summary, middle range theorizing can contribute to the study of organiza-

tions by:
1. promoting the development of a typological framework describing the

range of general organization theory;
2. encouraging more clearly targeted and operationally rigorous research;
3. providing a context for linking theories in the development of a general
theory of organization;
4. helping to distinguish between processes specific to organizations in con·
trast to those shared with other social phenomena;
5. identifying research and theory gaps in contingency frameworks;
6. identifying more clearly the applicability of research fmdings for the de-
sign and management of organizations.
LIMITATIONS OF MIDDLE RANGE THEORY
One of the most serious limitations of middle range theorizing follows from
legitimizing studies of partial sets of variables or a subset of organizational
processes. If, as systems theorists suggest (Katz and Kahn, 1966; Thompson,
1967), the processes in organizations are interdependent, then results from
studying any partial set of relationships are of questionable meaning. There is
always the possibility that fmdings are actually the consequence of variables or
processes outside the scope of the middle range theory. Where excluded factors
have a major effect, a middle range set of generalizations would give a distorted
interpretation of the organization phenomena.
As indicated earlier, group level processes could provide alternative explana-
tions to the outcomes ascribed to role theory. likewise, the technological aspects
of the "Organizational Factors" may directly explain behavior without the
necessity of examining the interpersonal relations. If these and similar possibil.
ities are never examined, we cannot be sure that the explanation offered by role
theory accurately represents organizational realities.
Limiting the scope of research also creates a problem when middle range
theory focuses on special types of organizations, such as manufacturing firms or
volunteer organizations. In such a case, independently pursued studies are likely
to focus on different generalizations and to ope rationalize variables in incom·
patible ways. As a result, findings relating to different types of organizations will
not be comparable, making it difficult to develop a general theory of organiza·
tions. In the case of role theory, most of the differences between organizations
have been masked by the fact that specific roles, either in a single organization or
in several comparable organizations were studied. In many instances the roles
studied were unique to a particular type of organization (e.g., nurses, professors,
schooi superintendents) making any generalizations risky at best.

A thIrd limitation of middle range theorizing concerns the significance of
theory. Generalizations that are easiest to ope rationalize are also likely to gener-
ate the greatest reward. There's no way of knowing, however, whether these
generalizations address the significant properties of organizations. Hence, a
strong possibility exists that middle range theorizing will encourage major com-
mitments of resources to marginally significant research. The relevance of role
theory as an explanation of organizational behavior has not been adequately
studied. In spite of this, most organizational research in this area has assumed
the validity of the role model and has focused on problems of role conflict. If, in
fact, technology, structure, and personality are the significant determinants
of organizational behavior, research on role conflict may be spurious.
A proliferation of overlapping defmitions is a fourth negative consequence
that can result from promoting middle range theorizing. As limited aspects of
organizations are studied independently, different terms may be developed for
the same or similar constructs and the same term may be used for different
operational measures, as has happened in studies of job satisfaction. This con-
fusion of terminology is especially probable given the lack of agreement on the
range of general organization theory. Thus, the independent study of subsets of
organizational phenomena will produce a plethora of research that is difficult
to link. Studies seemingly related to the same theory will have inconsistent re-
sults, and related research will be attributed to different theories.
This problem has already occurred in the area of role theory where the role
concept has been defined in a number of ways by people from different disci-
plines. Anthropology, sociology, and psychology had long recognized the value
of the role concept and in fact, three major defmitions of role evolved. Neiman
and Hughes (1951) identified the three defmitional threads as: " ... (a) defmi-
tions which use role to describe the dynamic process of personality develop-
ment; (b) defmitions in terms of society as a whole; (c) defmitions in terms of
specific groups within society" (p. 142).
The major limitations of middle range theorizing are, in summary:
1. producing spurious generalizations due to the exclusion of critical factors;

2. slowing the development of a general theory of organizations by generat-
ing noncomparable studies of different types of organizations;
3. focusing time and energy on the study of marginally significant relation-
ships;
4. generating theories that cannot be coherently linked because of over-
lapping and inconsistent terminology.
If these negative consequences prevail, middle range theorizing is more likely to

produce confusion than clarity in the study of organizations.
GUIDELINES FOR MIDDLE RANGE THEORIZING
Middle range organization theorizing focuses attention on the scope of theory

and research and encourages the study of limited components of the organiza-
tion phenomena. Since many limited range theories already exist and since re-
sources required for comprehensive organization research are difficult to amass,
there is little value in debating the merits of pursuing middle range theory. In-
stead, this discussion concludes with some suggestions about how to capitalize
on the contributions of this approach and counteract its limitations.
There are several facets to the middle range theorizing process, including: (a)
relating theory to a general frame of reference, (b) selecting areas of study , (c)
designing research, (d) constructing theory, (e) developing generalizations from
empirical data, and (t) applying research. The guidelines for theorizing are pre·
sented here as a checklist grouped according to the process involved.
The first process, relating theory to a general frame of reference, requires, of
course, the development of a typology defining the scope of organization theory.
This typology should be used:
1. to reach agreement on a vocabulary to facilitate the understanding of

research that already exists;
2. to defme variables and processes in any new studies using a consistent
set of terms.
3. to catalogue existing theory in order to identify gaps in research and
theory;
4. to design research that cumulatively contributes toward a general theory
of organizations.
Following the first two of these guidelines would counteract the tendency to
develop inconsistent terminology and would increase the comparability of sepa-
rate studies. The latter two guidelines promote the identification of significant
research issues and encourage studies in those areas, thereby emphasizing the
theory building contributions of the midrange approach.
In selecting areas of study, researchers should:
1. assess the significance of a project in relation to what is already known

and avoid marginally useful research;
2. determine whether critical variables are exluded from the scope of the
project, and, if so, either control for them or expand the study to include
them.
In addition to encouraging significant research, adherence to these selection

guidelines would increase the likelihood that meaningful, rather than spurious,
generalizations would be developed.
During the research design phase:
1. Samples should be selected to maximize variability within the scope of

the theory and control for potential extraneous influences.
2. Data collection should include adequate measures of all variables within
the scope of the theory and information on extraneous influences where
these cannot be controlled.
3. Research objectives should be described using commonly accepted
terms for variables, processes and relationships, rather than invented
terminology .
The first two design guidelines parallel the selection guidelines in promoting
studies that generate meaningful results within the domain of Significant mid-
range theory. They also reflect the contribution which this approach can make
to the operational rigor of research. Encouraging researchers to use current
terminology, rather than invent labels, supports typological development and
clarifies the relationships between a new research effort and other middle range
theoretical or empirical studies.
When proposing new conceptualizations, theorists should:
1. identify clearly (in relation to the typology) the scope and limits of the
generalization;
2. defme the scope of the theory so that it includes a reasonably self-
contained set of relationships;
3. indicate whether the phenomenon is limited to organizations or is shared
with other forms of social activity.
Identification of the scope and contents of a conceptualization would make it

possible to test the usefulness of its contribution and make certain that all
critical variables have been included, thereby preventing concentration on a
potentially spurious explanation. If the phenomenon is shared with other
forms of social activity, then the new conceptualization can utilize terminology
consistent with other disciplines, generating research comparable with that in
other fields.
Before attempting to develop generalizations from empirical data, researchers
should:
1. determine whether the data from different units represent a consistent

set of phenomena;
2. determine whether or not adequate measures are available for all the vari-
ables relevant to the phenomena;
3. determine whether potentially significant extraneous influences are
controlled;
4. describe the data using commonly accepted terminology that is relevant
to the phenomena.
If the above conditions cannot be met, any generalizations are very likely to be
distorted explanations, leading to spurious midrange theories and distracting
attention from significant areas of study. Failure to use relevant existing vocabu·
lary may generate seemingly unique theories that differ only semantically. Where
these guidelines are met, on the other hand, meaningful generalizations will have
the support of rigorous empirical descriptions.
Research fmdings should be applied only when the following are true:
1. Situations correspond with the situations addressed by the middle range

theory.
2. Factors extraneous to the theory are of minimal significance in the ap-
plication context.
These fmal guidelines should alert research users to the possibility of spurious
findings and the need to test comparability between the research setting and
their situation. In this latter respect, these guidelines institutionalize the contri-
butions of midrange theorizing to the applied use of research.
The preceding guidelines and their relationships to the contributions and
limitations of middle range organization theorizing are summarized in Table 1.
The guidelines are grouped according to the phases of the research activity to
which they are most relevant, and the contributions operationalized and limita-
tions addressed by following each of the guidelines are indicated by X's in the
table.
The establishment of a clearinghouse for organization theory and research
is the fmal recommendation that follows from this analysis of middle range
theorizing. The most severe limitations of the midrange approach include: dupli-
cation of effort, proliferation of terminology, and the pursuit of spurious or
insignificant research. On the other hand, identification of theories as middle
range places an emphasis on the total range of organization theory and legiti-
mizes the development of a general theory through the accumulation of studies
of partial aspects of organizations. A clearing house would provide a mechanism
for coordinating independent effort so that common terms could be used, sig-
nificant areas of research could be identified and pursued, and studies could be
designed to build on one another.
The development of a clearinghouse for research concerning middle range
organization theorizing is not without its problems. The first, and major obstacle
to overcome is the determination of an appropriate taxonomy of relevant orga-
nizational research issues into which research of a middle range nature can be
classified. Without such a taxonomy, the research results that already exist and
the results of new research cannot be analyzed to determine what has been done
and what needs to be done. The massive job of scanning organizationally rele-
vant research will be facilitated by a good taxonomy. This scanning must be
done if there is going to be any organized effort directed toward the develop-
ment of organization theory.
Once the initial tasks of developing a taxonomy and categorizing research
fmdings using the taxonomy are completed, or at least under way, the next
major task will be to gain the acceptance of researchers in the many disciplines
of the taxonomy and what it implies. If researchers are not willing to begin to
use common terminology; share data, ideas, and research methods; to, in fact,
realize that they are using middle range theories, then the task of the clearing-
house could prove impossible to accomplish.
To handle the problems and move the cause of middle range organization
theorizing forward will require a Significant effort. This conference is a very
useful first step in that it is focusing attention on middle range theorizing. Hope-
fully, some consensus about appropriate typologies for categorizing research
will evolve from the conference, but whether this occurs or not, a significant
input of time and effort will be required to develop a successful clearinghouse.
Some of the time and effort might be directed toward the following projects:
1. Producing an annual review of the state-of-the-arts in middle range orga-

nization theory. A monograph of this type would initially require the
analysis of research by the compiler, but in time researchers would, hope-
fully, submit material directly to the compilers.
2. Publish bibliographies of middle range organization theory and research.
These bibliographies would be categorized according to the typology de-
veloped and accepted by researchers in the field.
3. Commission specialized review articles to help eliminate ambiguities and
conflict in the relationships between existing research studies. A journal
devoted to middle range organization theory might be an outgrowth of
this type of endeavor.
4. Present symposia on a continuing basis to help develop consensus about
what middle range organization theorizing is and how it can aid the indi-
vidual engaged in organizational research.
tV
N
-...J
o
Cl Table 1. Contributions Operationalized and Limitations Addressed by Proposed Guidelines for Middle Range Organiza-
tion Theorizing
Contributions Operationalized Limitations Addressed
.....
I::
~
E ~ I::
I:: I:: C
.9 ~ ..... .::;
~
~
..... bE ~ I:: <.>
.....,;:.. ~ I:: ..t:> ~ ~
.~
.- c I:: I:l...-'
o ~ ~ li:: .~ ~
~ I:: .., ..t:> ~ ·9 1::'<::: ~
~ ~
I::~ ~.<:::
..... C <.l ~
I:l..
..,>l.., i.::~ ~
c
.., I:: .g .~ .~ >l.., .....
.- ... ~ ~ .....
.~ ~ ;:! .9 .~ <.l ~
0..9 ~
...
0
~ SI:l..
.., ~
<.l
~
..,
~l1 !:: ..,
c ..... ...
;:! . -
I:: .~ ~
I:: .-
-
~ ..... ~ ~ ·9
~ ....!>O .- ~ <.>~
!:: 0 ... ~
...oo ........~ I::\.:) I:: ..,
o I:l.. ~~ 1::0 ~.....,
~c:: ;:! I:: ~ ~ c<.> !::
...
Guidelines .~ ~ ~....., I:l.. ~
~h' c::0 ..!:;'C' \.:) c ~ c ~'C' ~\.:) ~~ ~~ ..!:;~
Using a Typology
1. Define with Common Terms xx X
X xx
2. Catalogue Theory X X X
X X
3. Design Cumulative Studies X X X X
X
Selecting Research Topics
1. Assess Significance X X X
X
2. Identify Excluded Variables X X
Designing Research
1. Select Samples to Maxi-
mize Internal Variability
and Minimize External
Variability X
X X
2. Gather Data on All Rele-
X
X X
X
vant Variables
3. Commo nly Defined Con- X
X
X
X X
X X
X
cepts X
X
Developing Theory X
X
X
X X
X X
X X
X
1. Identify Limits of Model
2. Include Relatively Self-
Contained Set of Relation-
X
X X
X X
X
ships
3. Identify Relevant Social X
X X
X
Phenomena X
X
Generalizing from Data

1. Test Consistency of Data X
X
X
X
Sources
2. Test Availability of Rele-
X
X X
X
vant Measures
3. Test for Controls of Ex-
X
X X
X
ogenous Variables
4. Utilize Common Termi- X
X
nology X
X
Applyin g Research Results

1. Test for Correspondence
between Research and
X
X X
X X
X
Application Sites
2. Test for Significance of
X
X X
X
Exogenous Variables
N
tv
-...)
--J
REFERENCES
Biddle, B.J., and Thomas, E.J. (Eds.). Role Theory: Concepts and Research.
New York: Wiley, 1966.
Blau, P.M., and Scott, W.R. Formal Organizations. San Francisco: Chandler,
1962.
Blood, M.R., and Hulin, C.L. Alienation, environmental characteristics, and
worker responses. Journal of Applied Psychology, 1967,51, 284-90.
Georgiou, P. The goal paradigm and notes toward a counter paradigm. Adminis-
trative Science Quarterly, 1973,18,291-310.
Gross, N., Mason, W. S., and McEachern, A. W. Explorations in Role Analysis.
Hall, R.H. Organizations: Structure and Process. Englewood Cliffs, N.J.:
Kahn, R.L., et al. Organizational Stress: Studies in Role Conflict and Ambiguity.
Wiley, 1966.
Leavitt, H. Applied organizational change in industry: structural, technological
and humanistic approaches. In J. G. March (Ed.), Handbook of Organizations.
Chicago: Rand McNally, 1965.
--. Social Theory and Social Structure (enlarged edition). New York: Free
Press, 1968.
Neiman, L.S., and Hughes, J.W. The problem of the concept of role-A resurvey
of the literature. Social Forces, 1951,30, 141-49.
Weber, M. Economy and Society (Vol. 1). New York: Bedminster Press, 1968.
23OF LIMITED DOMAIN THEORIES
ORGANIZATIONAL ENERGY
The title of this paper contains two basic elements-the concepts of limited
domain theory and of organizational energy. The first of these draws upon
Dubin's (1969) terminology and refers to theories that have a number of
boundary-determining criteria. Whether the theories to be presented are truly
theories of the middle range in the sense Merton (1949) intended is an open
question, but they are not grand theories; the boundaries within which they are
expected to prove valid are sharply delimited. It is because the defmition of
middle range theory is somewhat imprecise that the limited domain concept is
applied here. When exactly does a theory move from a narrow range into the
middle of the spectrum? If one is not sure about this matter of classification, the
limited domain terminology with its requirement of precise boundary limits, has
considerable appeal.
The second concept in the title, that of organizational energy, assumes that
organization structures are essentially static and that some source of energy is
required to make them dynamic, in pursuit of goals. This energy derives from
the human input to the organization and takes the form of motivated effort or
force. The energy is aroused and charmeled toward goal accomplishment by
certain kinds of role prescriptions, which may be either formally stated or infor-
mally operative, depending on the predilections of the particular organization.
273
There appear to be at least four major ways in which organizations induce

human effort or create energy. Each of the four draws upon a different set of
role prescriptions to engage member motivation. These four inducement systems
and their respective theoretical domains are considered in the remainder of this
paper-each in terms of a separate limited domain theory. The initial theory,
dealing as it does with organizational management, focuses on a relatively
familiar area. The remaining theories deal with what Kerr (1977) calls "substi-
tutes for leadership" and to varying degrees break new theoretical ground.
THE THEORY OF HIERARCHIC INDUCEMENT SYSTEMS
The hierarchic theory was first formally set forth in Miner (1965), although a
certain amount of research on the theory was conducted prior to that time. It
has generally been labelled managerial role-motivation theory. The domain of
the theory follows the boundaries of the ideal bureaucracy as described by
Weber (1947); it is assumed that the organization is large enough so that for-
malized, written communication is necessary, not merely face-to-face interaction
(Stinchcombe, 1974). In such an organization the key group is management,
especially top management. Management induces contributions from lower
members through the manipulation of positive and negative sanctions, and the
energy thus generated is channeled toward goal attainment through the establish-
ment of some type of role prescriptions, as specified in written job descriptions
to take one example.
But where does management obtain its energy? To some extent, of course,
through the manipulation of sanctions at yet higher levels of management. How-
ever, there is a limit to this process. The theory of hierarchic inducement systems
states that there are at least six general types of managerial role prescriptions
operating in the hierarchic context, each of which in tum must engage a particu-
lar intrinsic motivational constellation if managerial behavior appropriate to the
effective operation of this particular inducement approach is to occur. These
role prescriptions and the patterns of motivation (not necessarily a unitary
motive) they engage are as follows:
1. Role. The positive relations with authority figure role. Motivational Base.
A favorable attitude toward people in authority positions generally. Ex-
planation. In a hierarchic system there must be communication and inter-
action upward with superiors and favorable attitudes facilitate this.
2. Role. The competitive role. Motivational Base. A desire to compete. Ex-
planation. In a hierarchic system, rewards such as promotion to the highest
levels and the largest salaries are differentially distributed and relatively
LIMITED DOMAIN THEORIES OF ORGANIZATIONAL ENERGY 275
scarce and it therefore becomes necessary to compete with peers to attain

them.
3. Role. The imposing wishes role. Motivational Base. A desire to exercise
power over others. Explanation. In a hierarchic system manipulation of
sanctions and downward supervision are required. To the extent such
activities yield satisfactions, they are more likely to be performed.
4. Role. The assertive, father role. Motivational Base. A desire to assert
oneself. Explanation. In a hierarchic system the managerial role in many
respects is modeled on the traditional father role and accordingly a degree
of assertiveness (a "take charge" attitude) congruent with the father role
is required.
5. Role. The standing out from group role. Motivational Base. A desire to
assume a distinctive, differentiated role. Explanation. In a hierarchic sys-
tem, managers must assume a highly visible position clearly differentiated
from the relative homogeneity of their subordinates. Consequently, it
becomes necessary to want to assume such a distinctive position.
6. Role. The routine administrative role. Motivational Base. A desire to per-
form routine administrative duties in a responsible manner. Explanation.
In a hierarchic system various routine decision-making and communica-
tion tasks must be carried out, and therefore appropriate motivation and
a sense of responsibility for such matters are required.
THE THEORY OF PROFESSIONAL INDUCEMENT SYSTEMS
An alternative to the managerial role prescriptions of the hierarchic system is the

flow of role requirements from a profession and its norm transmitting associa-
tion(s). For this type of inducement system to operate, an individual must be a
member of a full-fledged profession. The domain of the theory is thus congruent
with the limits of professional membership; the individual may be self-employed,
a member of a professional organization, or operating in a professional capacity
in some other type of organizational context. In any event, we are concerned
with an inducement system that utilizes cosmopolitan rather than local methods
of influence (Gouldner, 1957). In some cases various pervasive ideological
systems, such as religions, may exert influence in much the same manner as
professions and achieve a very similar type of inducement effect.
The professional theory states that there are at least five types of role pre-
scriptions that combine with their appropriate motivational patterns to induce
organizational energy. These role prescriptions have been developed in part from
existing statements defming professions and professional organizations (Etzioni,
1964; Satow, 1975; Vollmer and Mills, 1966) and in part from research on pro-
fessionals (Hall, 1967; Harrison, 1974; Sorenson and Sorenson, 1974). The role
prescriptions and their motivational bases are as follows:
L Role. The knowledge acquisition role. Motivational Base. A desire to learn

and acquire knowledge. Explanation. In a professional system it is a sine
qua non that technical expertise and knowledge must be developed, trans-
mitted, and used in the service of clients. Accordingly, a professional
must want to acquire the knowledge that permits providing an expert
service.
2. Role. The professional commitment role. Motivational Base. A value-based
identification with the profession. Explanation. In a professional system
there must be a strong tie to a profession that keeps members responsive
to its ethical nor:ms, and this tie is achieved through a strong sense of per-
sonal identification or commitment.
3. Role. The status acceptance role. Motivational Base. A desire to acquire
status in the eyes of others. Explanation. In a professional system the
provision of services to clients is predicated on client recognition of the
professional's expert status. This means that a desire to acquire status and
the symbols of status must exist or the available services will go unutilized.
4. Role. The providing help role. Motivational Base. A desire to help others
to achieve their best interests. Explanation. In a professional system the
client-professional relationship is central and in that relationship the pro-
fessional is expected to assist the client in achieving desired goals, or in
some instances that which is specified by the profession as being in the
client's best interest, even if not consciously desired. As a consequence, a
professional must want to help others. Although this type of motivation
is normally interpreted in terms of service to others, it seems more likely
that it commonly involves the exercise of "helping power" as defined and
discussed by McClelland (1975).
5. Role. The independent action role. Motivational Base. A desire to act
independently. Explanation. In a professional system the individual has a
private and personally responsible relationship with the client that often
requires independent action based on one's own best professional judg-
ment. To meet this requirement it is necessary that the person be inde-
pendent and desire to act independently of others.
THE THEORY OF GROUP INDUCEMENT SYSTEMS
A second alternative to hierarchic inducement of effort is inducement generated

from group pressures. This would appear to be the focal type of inducement
system in most forms of participative management. It is evident for example

in System 4 as recently described in Likert and Likert (1976) and even more
clearly in the socio-technical approach (Herbst, 1976).
In the pure case of group inducement decisions are made by consensus or
majority vote and leadership is emergent, occurs at the will of the majority,
and is often rotating. Communication tends to be predominantly face-to-face,
and the basic unit is the workgroup. Coordination among groups within the more
comprehensive organizational context is achieved by interlocking and overlapping
group memberships. Groups achieve goal-directed energy inducement by exert-
ing concerted pressure on members to behave in certain ways, as originally
discovered in the Hawthorne research (Cass and Zimmer, 1975).
For such a system to operate there must be a cohesive group with a reason-
ably stable membership. The domain of the theory is delimited by the existence
of group decision-making and deciSion-implementation processes. Although to
some extent the energy arising from group inducement systems is intrinsic and
inherent in the fact of actually participating in the decision process (and thus
owning the decision), the theory of group inducement does not preclude the use
of negative sanctions in support of group norms, or even the use of force to gain
compliance or to maintain continued group membership.
The group theory specifies five role prescriptions and their motivational bases,
which serve to induce organizational energy. It considers these to be widely
operative within the domain without precluding the existence of other signifi-
cant role prescriptions. The five have been developed from a reading of the exist-
ing literature regarding the operation of participative industrial work groups,
voluntary organizations, legislative bodies and the like. In addition, statements
regarding the motivational and value foundations of organization development,
the modern manifestation of the insights derived at Hawthorne, were utilized
(French and Bell, 1973; Golembiewski, 1972; Schmidt, 1970). The five role
prescriptions and their correlative motives are as follows:
1. Role. The social interaction role. Motivational Base. A desire to interact

with others and to do so effectively. Explanation. In a group system the
members are required to spend considerable time interacting with others
in order to make and obtain implementation of decisions. It thus becomes
necessary that they enjoy affIliative relationships and the use of social
skills to exert influence and minimize nonproductive conflict.
2. Role. The group acceptance role. Motivational Base. A desire for con-
tinuing group membership. Explanation. In a group system it is necessary
that members maintain identification with the group both to facilitate
participation in group processes and to make them responsive to group
administered sanctions, including the threat of expulsion. Accordingly,
a desire not merely for social interaction, but for acceptance and a sense
of belongingness in the group, becomes necessary.
3. Role. The positive relations with peers role. Motivational Base. A favor-
able attitude toward peers and coworkers generally. Explanation. Just as
in hierarchic systems favorable attitudes upward are required to facilitate
hierarchic communication, in a group system favorable attitudes toward
group members, or potential group members, are also required. These
positive attitudes should take such forms as belief in others, trust, con-
sideration, and mutual respect.
4. Role. The cooperative role. Motivational Base. A desire to work with
others in a cooperative or collaborative fashion. Explanation. In a group
system, competition among members can disrupt communications and
lead to withholding of needed information -even to group disintegration.
As a result, members must eschew competition in favor of cooperative,
collaborative relations and gain satisfaction from doing so.
S. Role. The democratic process role. Motivational Base. A desire to partici-
pate in democratic group processes. Explanation. In a group system the
major method of getting things done is some variant of the power equal-
ized, democratic process including open expression of ideas and views,
and the use of democratic procedures such as the vote to reach decisions.
For these processes to function effectively members must be favorably dis-
posed toward them and desire to contribute to their effective operation.
THE THEORY OF TASK INDUCEMENT SYSTEMS
The third alternative to hierarchic leadership involves a situation where the

pushes and pulls of sanctions are built into the task to be performed, and thus
into the work itself. In such instances operative pressures and incentives derive
neither from superiors, nor professional systems, nor peer group members, but
rather from the way in which the work is structured. As such, they take on a
highly impersonal nature. The prime model for this kind of inducement system
is found in the entrepreneurial organization. Pulls or positive sanctions are in-
herent in the prospects for financial reward, community status, and personal
satisfaction, but at the same time there are pushes to stay in the situation and
continue efforts emanating from the threats of business failure and bankruptcy.
These forces operate primarily on the individual entrepreneur (Collins, Moore,
and Unwalla, 1964), but in the small entrepreneurial organization they permeate
other members as well to varying degrees; all are aware that if the firm succeeds
they may very well prosper, and if it does not, they certainly will be out of
ajob.
A task inducement approach is built into many business jobs, in each instance
modeled after the entrepreneurial case, which in fact specifies the boundary-
determining criteria of the theory. Examples are the profit center manager, the
corporate venture manager, straight commission sales representatives, and manu-
facturers' representatives. Much job enrichment utilizes a similar pattern (Ford,
1969; Herzberg, 1976). In all of these instances, achievement motivation theory
appears to provide valuable insights into the dynamics of work behavior (Mc-
Clelland and Winter, 1969).
McClelland (1961, 1962) specifies certain characteristics of situations that are
preferred by and tend to elicit achievement striving from individuals with a
strong achievement need. These characteristics may be viewed as providing a
minimum set of role prescriptions necessary for the operation of a task induce-
ment system. These role prescriptions and the motivations engaged (presumably
components of the overall need achievement variable) are as follows:
1. Role. The individual achievement role. Motivational Base. A desire to

achieve through one's own efforts and to be able to clearly attribute any
success to personal causation. Explanation. In a task system it is essential
that individuals continually be pulled into the task situation so that they
do not simply avoid any pressures involved by escaping from the work
context. The major source of this pull is an intrinsic desire to achieve
through one's own efforts and ability, and to experience the enhanced
self-esteem that such achievement permits-to be sure that one did it
oneself.
2. Role. The seeking results of behavior role. Motivational Base. A desire
for some clear index of the level of performance. Explanation. In a task
system feedback on the level and results of one's performance is necessary
in order to attribute any degree of success (or failure) to one's efforts; it
is crucial to know whether one has succeeded or failed. Consequently, the
individual must be motivated to actively seek out results-oriented feed-
back in terms of measures such as profitability, productive output, wast-
age, course grades, and so on.
3. Role. The personal innovation role. Motivational Base. A desire to
introduce novel, or innovative, or creative solutions. Explanation. In a
task system the pull of individual achievement works only to the extent
the individual can attribute personal causation. Original or creative ap-
proaches have a distinctive quality that makes it easier to identify them as
one's own and to take personal credit for them. As a result, a desire to
introduce such approaches is more likely to make task inducement func-
tion properly.
4. Role. The risk taking role. Motivational Base. A desire to take moderate
risks that can be handled through one's own efforts. Explanation. In a

task system the individual must face considerable challenge and the pros-
pect of being overextended, however briefly. Such a situation requires a
desire to take risks; tasks which the individual considers easy and already
knows how to deal with have little pull because there is no sense of
achievement to be had in accomplishing them. However, desires to take
high risks or to take risks related entirely to luck or fate (where one
cannot ultimately reduce the risk through one's own effort) are not
functional. In neither case can a sense of individual achievement be antic-
ipated with any reasonable probability, and thus energy is unlikely to be
engaged.
5. Role. The planning and goal-setting role. Motivational Base. A desire to
think about the future and anticipate future possibilities. Explanation. In
a task system the individual must be pulled by the prospect of anticipated
future rewards, and therefore must approach life with a strong future
orientation. There must be a desire to plan, to set personal goals that will
signify achievement, and to plot paths to goal achievement; this in turn
implies a minimal expectation, or fear, of future failure.
THEORETICAL QUESTIONS
There are eight basic questions that limited domain theories of the type set forth
in this paper should be expected to answer (Miner, 1977). In a number of in-
stances these questions may be formulated as theoretical hypotheses, and thus
they may be used in testing the underlying theories.
1. The Measurement Question
It is assumed that measurement procedures can be developed to ope rationalize

the variables of the theories. This involves obtaining scores for each of the sepa-
rate motivational patterns that a theory indicates must be engaged if behavior
matching the role prescriptions is to emerge. An overall measure obtained by
summing these component scores may be used to test the theory as a whole.
To date, this kind of operationalization of variables has been achieved most
successfully using sentence completion methods (Miner, 1965, 1977, 1978b).
Such measures have been developed for the hierarchic theory and on an experi-
mental basis for the variables of the professional theory. In due course, similar
measures for the group and task theories will be attempted.
2. The Validity Question
The theory-based measures should predict job performance levels within the
domain of their appropriate theories, that is, in the theoretically specified
organizational or work context. Relationships would be expected to be con·
siderably stronger if purified measures of motivational force were used as criteria,
but the theories do anticipate that the motivational patterns will explain sizable
amounts of performance variance. Furthermore, extrapolating from current per·
formance, career success should also be predictable.
In terms of the four theories the research hypotheses may be stated as follows:
1. Hierarchic (managerial) role motivation theory will predict managerial

success as defined by generally accepted measures.
2. Professional role motivation theory will predict professional success as
defined by generally accepted measures.
3. Group role motivation theory will predict status and influence within the
group as defined by generally accepted measures.
4. Task role motivation theory will predict success in jobs structured in ac-
cordance with the entrepreneurial case when success is defined by gener-
ally accepted measures.
3. The Training and Development Question
It should be possible to increase the levels of motivation, or the extent to which

motives dominate behavior, for each of the theoretically specified motive sets
through the use of appropriate training or development strategies. This training
should be designed to cast the individual in the roles specified by the theory at
least cognitively, should permit an exploration both cognitively and emotionally
of expected sources of anxiety and satisfaction if one should actually perform
in these roles, and should be carried out only with individuals who have some
reasonable prospect of performing the roles. To date, such approaches have been
found successful with the motives of the managerial role (Miner, 1965, 1977)
and the task role (McClelland and Winter, 1969).
4. The Vocational Guidance Question
Among individuals with sufficient vocational knowledge, the motivational pat·

terns of the theories should operate to guide occupational and organizational
choice. Thus, managers or prospective managers should score relatively high on

the managerial motive constellation and those in positions with more managerial
discretion (top versus bottom, line versus staff, business versus government)
should also score higher. Those with high original managerial motives would be
expected to be more likely to enter managerial occupations.
Similarly, high levels of professional motivation should lead to professional
careers, typically lifetime careers. Group motivation presents certain difficulties
in this regard, since most existing employing organizations are either hierarchic
or professional in basic form. The theory would predict that given a choice, high
group motivation people would prefer group inducement systems in which to
work. However, many may be forced to seek satisfaction of those motives out-
side of the employment sphere, in voluntary and social organizations. Task moti-
vation, when pronounced, should be associated with founding one's own firm
or with seeking out highly discretionary, impersonal feedback positions in other
organizational contexts. In general, individuals with the requisite knowledge will
make choices based on their dominant motivational pattern. However, where
more than one pattern dominates they will seek mixed work contexts that per-
mit optimization of motive satisfaction.
5. The Job Satisfaction Question
The theories would hypothesize that an appropriate mesh of motivational pat-

terns and role prescriptions should lead to greater job satisfaction. However,
there is no necessary reason to believe that this satisfaction would generalize
beyond the specific intrinsic motives and activities involved to the broad range
of factors usually measured in job satisfaction questionnaires (compensation
benefits, supervision, and the like). Furthermore, there are cases, such as much
entrepreneurial endeavor, where only those with relatively high motivational
levels would be expected to survive in the work. In such instances, restriction of
range on both motivational and satisfaction measures would be expected to
sharply limit correlations.
6. The Talent Supply Question
The problem of the availability of high levels of the different types of motiva-
tion in a population at a given point in time is an important one, and one which
the theories should help to solve. However, this question adds an historical
dimension to the theories, which has not really been considered to this point,
except as related to the training and development question. It seems logical that
families should develop strong motives in children just as training does in adults
(Turner, 1970). The more difficult problem is to determine how parents come to
change or permit a change in motivational patterns so that major shifts in the
concentration of motives in a society of the kind we know do occur (McClelland,
1961 ; Miner, 1977) come about.
On this point the theories must remain silent at the present time. It seems
highly likely that those societies which have perpetuated certain patterns-hier-
archic, professional, group, and/or task-over considerable periods may in a
relatively short time go neutral, and fail to enforce existing values. But it is by
no means clear why this occurs, and why a particular other motive pattern then
takes on a predominant role. It is simply premature to extend the theories in
these directions at present.
7. The Domain Definition Question
To the extent pure tests within a domain are possible, it should follow that the
hypotheses set forth under the preceding questions will hold within the appro-
priate domain, but not outside it. The four inducement-system-based motiva-
tional patterns are not, however, assumed to show any necessary relationships
with each other; they are not pure factors. Thus, both positive and negative
correlations are expected depending on circumstances, although large correla-
tions would not be anticipated. Since many organizational forms are based on
more than one type of inducement system, positive correlations between the
appropriate motive sets can be anticipated as a result of selective recruitment in
such instances.
8. The Construct Validity Question
Answers to the preceding questions will inevitably bear on matters of construct

validity. The most important construct validity issues relate to the total scores
for the variables of each theory. These scores should correlate with other in-
dexes, which may justifiably be considered to provide legitimate measures of
the same variables. Unfortunately, at this stage in the development of knowledge
there are few such measures. The most obvious exceptions to this statement are
certain indexes of managerial interest and dominance that do appear to operate
in a manner comparable to hierarchic motivation (Miner, 1977). In addition,
operationalizations of group motivation should be reasonably comparable to
existing indexes of social or affiliative motives, and operationalizations of task
motivation to achievement motivation scores. There appears to be relatively
little by way of comparable measurement within the professional domain.
RELATIONSHIP AMONG THEORIES
As this theoretical rubric has expanded from one domain to another-from the
hierarchic to the professional, and now to the group and task-it would appear
to take on many of the characteristics of a contingency theory and in fact it has
been proposed as an alternative to existing contingency theories, which utilize
environmental characteristics as the key contingency variables (Miner, 1978a,
1978c). However, the theoretical variables themselves have been generated
separately and are posited as being theoretically distinct and independent. There
is no statement as to the normative superiority of one theoretical framework
over another. As far as the theories are concerned, and existing knowledge as
well, one theory can produce as effective an organization as another provided
the motivational base is congruent with the role prescriptions. However, to ac-
complish this equality of effectiveness large organizations utilizing professional,
group, or task inducement systems will have to separate into small, reasonably
autonomous units at whatever point under variable structuring a change in in-
ducement approach among these three alternatives occurs.
In essence, then, we are dealing to this point with four distinct limited domain
theories of organizational energy inducement. The "our may also be considered
to be theories of organizational control, using that term in the broad sense it has
recently assumed (Lawler and Rhode, 1976). In fact, the first statements of the
theories utilized the term control (Miner, 1975). However, the ambiguities of
that term (Newman, 1975) make it unattractive for theoretical purposes, and its
use has been abandoned in subsequent writing.
Although it is appropriate and useful to state the four theories as independent
entities, it is also realistic to recognize that these domains may on occasion over-
lap within the confines of a specific organization. In particular, professional,
group, or task elements may appear in primarily hierarchic organizations. How
does one deal with such overlapping domain contexts?
To the extent two or more inducement systems are superimposed in an orga·
nizational unit or an organization as a whole, there is a high probability of role
conflict. Given that the role sets are independent of each other, multiple oppor-
tunities will exist for individuals to be pulled two ways at once when more than
one set is operating. For most people this means reduced levels of effectiveness
and intrinsic job satisfaction (Szilagyi, Sims and Keller, 1976). It seems entirely
possible, for example, that the negative effects of extrinsic rewards on intrinsic
motivation noted by Deci (1975) are a consequence of conflicts between hier-
archic and task inducement systems.
However, the introduction of variable structuring in an organization will
inevitably create certain positions in which inducement systems interface (Miner,
1974). This is true, for instance, of profit center general managers in a corpora-
tion, deans in a university, the first level of hierarchic supervision where socio-
technical work groups have been organized, and so on_ In these specific positions
role conflict is inevitable. It need not be debilitating for performance, however,
if the individual staffing the position is one of those relatively rare people who
possesses strong motivational patterns along both dimensions represented by the
inducement systems interfacing in the position. Because such motive combina-
tions are not expected to occur frequently, given the assumed low level of corre-
lation among the underlying variables, staffing interface positions is a difficult
matter and failures occur frequently. Yet, if the appropriate people can be
found, a high degree of resistance to the particular kind of role conflict present
is hypothesized.
REFERENCES
Cass, E.L., and Zimmer, F.G. Man and Work in Society. New York: Van Nos-
trand Reinhold, 1975.
Collings, O.F., Moore, D.G., and Unwalla, D.B. The Enterprising Man. East
Lansing, Mich.: Bureau of Business and Economic Research, Michigan State
University, 1964.
Deci, E. L. Intrinsic Motivation. New York: Plenum, 1975.
Dubin, R. Theory Building. New York: Free Press, 1969.
Etzioni, A. Modern Organizations. Englewood Cliffs, N.J.: Prentice-Hall, 1964.
Ford, R.N. Motivation Through the Work Itself. New York: American Manage-
ment Association, 1969.
French, W.L., and Bell, C.H. Organization Development. Englewood Cliffs, N.J.:
Golembiewski, R. T. Renewing Organizations: The Laboratory Approach to
Planned Change. Itasca, 111.: F. E. Peacock, 1972.
Gou1dner, A. W. Cosmos and locals: Toward an analysis of latent social roles. Ad-
ministrative Science Quarterly, 1957,2,281-306,444-480.
Hall, R.H. Some organizational considerations in the professional-organizational
relationship. Administrative Science Quarterly, 1967, 12, 461-478.
Harrison, F. The management of scientists: Determinants of perceived role per-
formance. Academy of Management Journal, 1974, 17, 234-241.
Herbst, P.G. Alternatives to Hierarchies. Leiden, Netherlands: Martinus Nijhoff,
1976.
Herzberg, F. The Managerial Choice: To be Efficient and to be Human. Home-
wood, Ill.: Dow-Jones-Irwin, 1976.
Kerr, S. Substitutes for leadership: Some implications for organizational design.
Organization and Administrative Sciences, 1977,8; 1,135-146.
Lawler, E.E., and Rhode, LG. Information and Control in Organizations. Pacific
Palisades, Calif.: Goodyear, 1976.
Likert, R., and Likert, J.G. New ways ofManaging Conflict. New York: McGraw-
Hill, 1976.
McClelland, D.C. The Achieving Society. Princeton, N.J.: Van Nostrand, 1961.
- - . Business drive and national achievement. Harvard Business Review, 1962,
40: 4,99-112.
- - . Power: The Inner Experience. New York: Irvington, 1975.
McClelland, D.C., and Winter, D.G. Motivating Economic Achievement. New
York: Free Press, 1969.
Miner, J.B. Studies in Management Education. Atlanta, Ga.: Organizational Mea-
surement Systems Press, 1965.
- - . The Human Constraint. Washington, D.C.: BNA Books, 1974.
- - . The uncertain future of the leadership concept: An overview. In J.G. Hunt
and L. L. Larson (Eds.), Leadership Frontiers. Kent, Ohio: Comparative Ad-
ministration Research Institute, Kent State University, 1975, 197-208.
--.Motivation to Manage: A Ten Year Update on the "Studies in Management
Education" Research. Atlanta, Ga.: Organizational Measurement Systems
Press, 1977.
- - . The Management Process: Theory, Research, and Practice. New York: Mac-
millan, 1978a.
- - . The Miner Sentence Completion Scale: A reappraisal. Academy of Manage-
mentJournal, 1978b, 21,283-294.
- - . The theory of strategy-structure-people fit. In D.E. Schendel and C.W.
Hofer (Eds.), Business Policy and Planning Research: The State of the Art.
Boston, Mass.: Little, Brown, 1978c.
Newman, W.H. Constructive Control. Englewood Cliffs, N.J.: Prentice-Hall,
1975.
Satow, R.L. Value rational authority and professional organizations: Weber's
missing type. Administrative Sciencey Quarterly, 1975,20, 526-531.
Schmidt, W.H. Organizational Frontiers and Human Values. Belmont, Calif.:
Wadsworth, 1970.
Sorensen, J.E., and Sorensen, T.L. The conflict of professionals in bureaucratic
organizations. Administrative Science Quarterly, 1974, 19, 98-105.
Stinchcombe, A.L. Creating Efficient Industrial Administrations. New York:
Academic Press, 1974.
Szilagyi, A.D., Sims, H.P., and Keller, R.T. Role dynamics, locus of control,
and employee attitudes and behavior. Academy of Management Journal,
1976,19, 259-276.
Turner, J.H. Entrepreneurial environments and the emergence of achievement
motivation in adolescent males. Sociometry, 1970,33, 147-165.
Vollmer, H.M., and Mills, D.L. Professionalization. Englewood Cliffs, N.J.:
Weber, M. The Theory of Social and Economic Organization. Trans. by A.M.
Henderson and T. Parsons. Oxford: Oxford Universtiy Press, 1947.
24 RELATIONSHIPS BETWEEN
MODES OF SOCIAL INTERACTION
E. Allen Slusher and Kenneth J. Roering,
University of Missouri-Columbia
The increased emphasis on improved middle range theories in social science

represents a response to the theoretical and practical limitations of broad,
general theories (see for example, Weick, 1974). In the spirit of this movement
toward middle range theories, we argue that three distinctive modes of social
interaction can be identified. Moreover, it is further argued that behavior within
a given mode is best explained by application of a unique middle range theory.
However, the central focus is on a different, but closely connected issue. This
issue concerns how movements occur between interaction modes. In effect, we
are asking how linkages can be built between middle range theories of social
interaction. For convenience, the analysis is limited to the simplest form of
interaction - the dyad. However, either one or both parties to the dyad can be
conceptualized as an individual, group, or organizational entity. Since our
current concerns are speculative and exploratory, social interaction is examined
abstractly without specifying a certain type. Later, the adequacy of this abstract
approach can be tested by reference to more specific cases.
287
MODES OF SOCIAL INTERACTION
Boulding (1970, p. 9), applying a general systems perspective, notes that there
are three basic genotypes in social systems that define the role relationships be-
tween two parties. The first genotype is a threat relationship, which " ... origi-
nates when one person says to another, 'You do something that I want ... or I
will do something that you do not want.' If the threatened person submits, we
get a threat-submission system ...." Second is an exchange relationship, which
" ... develops when somebody says to somebody else, 'You do something that I
want and I'll do something that you want.'" Third is an integrative relationship,
which " ... takes form when somebody says to somebody else, 'You do some-
thing because of what you are and what I am.'" Walton and McKersie (1965),
focusing on the narrower issue of social negotiations, offer an analytical frame-
work complementary to Boulding's social system genotypes. They identify three
interunit subprocesses:
The first subprocess is distributive bargaining; its function is to resolve pure

conflicts of interest. The second, integrative bargaining, functions to find
common or complementary interests and solve problems confronting both
parties. The third subprocess is attitudinal structuring, and its functions are
to influence the attitudes of the participants toward each other and to affect
the basic bonds which relate the two parties they represent.
Distributive bargaining refers to a set of activities " ... instrumental to the

attainment of one party's goals when they are in basic conflict with those of the
other party (p. 4)." In these situations, the structure of outcome interdepen-
dence between the two parties approaches what game theorists term a constant-
sum game, what one party gets the other loses. Such situations often involve the
threat relationship described by Boulding. When one party is more powerful
than the other, we can expect threats from the strong to produce submission or
compliance by the weak. In contrast to distributive bargaining, integrative bar-
gaining activities occur in situations where, despite some differences in
preferences, there are outcomes that can benefit both parties. Game theorists
view these varying-sum outcomes as representing mixed-motive situations. The
parties seek their own gain but are constrained by their desire to maintain a
mutually beneficial relationship. Such mixed-motive relationships seem to be
characterized by the exchange pattern Boulding notes. Additional structuring
aimed at solidifying the basic bonds between two parties helps produce the
integrative relationship Boulding has defined. In summary, both Boulding and
Walton-McKersie offer similar threefold taxonomies for social interactions.
Bonoma (1976), in a provocative article, addresses more explicitly and in
RELATIONSHIPS BETWEEN MODES OF SOCIAL INTERACTION 289
greater detail than do either Boulding or Walton and McKersie the issue of inter-
action modes. Bonoma sees social episodes involving power, conflict, coopera-
tion, and trust as falling into three categories: unilateral, mixed, and bilateral
power systems. The unilateral and mixed power systems are differentiated by
relative power. In the unilateral system there is a strong party and a weak party.
Mixed power systems imply a greater balance (although not necessarily equality)
of power between the parties. Although unilaterial and mixed systems are con-
ceived as being qualitatively different, the greatest contrast occurs when the
unilateral and mixed systems are compared with the bilateral power system. As
Bonoma notes "There is a third class of social episode implicating conflict
constructs, one in which the interactants share a unit bonding or belongingness;
it is different in kind and degree from the other two." Bonoma strongly
advocates the position that the relationship between the same theoretical
constructs will differ across the three types of social episodes. This suggests the
need for three middle range theories each aimed at explaining the relationship
among constructs for a given category. If a general theory were to be developed,
it must include modifiers for application to a given social context.
There is great similarity in the theoretical views of Bonoma, Boulding, and
Walton and McKersie. We wish to synthesize their view in a framework applicable
to the dyadic interactions common to much of organizational life. We defme
three modes of social interaction: dominance, exchange, and commitment. The
differences between these modes are summarized in Table 1.
When considering the modes of social interaction, one should not assume that
the characteristic theoretical constructs apply exclusively to a single mode. For
example, conflict might develop in a commitment interaction over scarce re-
sources. However, we do believe that the most probable source of conflict in
the committed mode will be differences in perspectives on individual and unit
welfare or the nature of joint policies. The interaction modes should be viewed
as describing general tendencies or highly probable behaviors rather than discrete
categories.
The question of exclusivity-nonexclusivity in assigning characteristic
theoretical constructs to the interaction modes is related to another fundamental
question. Does an interaction mode describe all interactions between two
parties, a subset of all interactions, or only a single interaction? We are not
referring here to interactions that represent mere repetition over time, but to
interactions that differ in some characteristic attribute (e.g., issue, importance,
duration, etc.). If a mode applies to only one interaction, then our typology can
only be applied in an issue or transaction specific sense. It might thus be
necessary to categorize three different interactions between the same two parties
as instances of dominance, exchange, and commitment. This approach not
only presents practical research difficulties, but ignores the strong influence of
N
\0
o
Table 1. Three Modes of Social
Socia I Interaction.
Modes
Theoretical Constructs Dominance Exchange Commitment
Power Unilateral Relative Extradyad

(A is strong, B is weak) (Neither A nor B clearly (Total unit power in rela-
strong or weak) tion to environment)
Trust Credibility of threats and Mutual credibility of offers; Social altruism
promises of strong party adherence to normative
framework
Cooperation Benevolence by strong, Normatively specified Concern for unit welfare
Compliance by weak
Outcome Structure Constant-sum game Variable-sum game No Conflict game
Conflict Scarce resources Resource allocation Individual vs. dyad welfare;
nature joint policies
Additional Characteris- Persuasion Offers-Counter Offers Mutuality
tic Constructions Promises Trade Continuity
Threats Equity Consensus
Ingratiation Reciprocity Social Welfare
Control Contract Joint policies
Manipulation Superordinate goals
Coercion Problem solving
basic social relationships. At the other extreme, applying the same modal anal-
ysis to all interactions between the same two parties may lead to erroneous
conclusions in special instances. Here it is conceived that a given interaction
mode applies to most, but not all interactions, between two parties.
We have argued strongly in this section that social interaction theory and
research should recognize different interaction modes between parties. More-
over, we have offered a threefold typology for distinguishing between inter-
action modes. This analysis suggests that research will proceed most effectively
when an appropriate middle range theory is applied to a given interaction mode.
A middle range theory for the dominance mode can be constructed from various
bases in psychology, sociology, or political science. For the exchange mode an
appropriate middle range theory can be derived from the classic work in social
exchange theory by Gouldner (1960), Homans (1961), Thibaut and Kelley
(1959), and Blau (1964). The problem is more acute for the commitment mode
where social science theories are less developed and more diffuse. However, the
work on the psychology of commitment by Kiesler and his colleagues should be
a useful point of departure (Kiesler, 1971).
Although we believe that research will be more productive when greater
emphasis is placed on identifying the mode of social interaction and then select-
ing a middle range theory, there is another fundamental issue that should be
addressed. In dynamic social systems two types of change may occur. A given
middle range theory, to the extent it involves process, deals with the first type of
change-that which occurs within a given interaction mode. An understanding of
this intramode change is intrinsically useful and also complementary to under-
standing the second type of change-that which leads to a shift from one inter-
action mode to another. Such intermode change represents a fundamental
restructuring of the parties' interaction with consequent influences on behavior
and outcomes. Unfortunately, little attention has been given to intermode
change despite its potential importance. In the remainder of this paper we will
analyze intermode change from the perspective of a new development in mathe-
matical theory.
A CATASTROPHE MODEL ANALYSIS OF

SOCIAL INTERACTION
The notion of a "qualitative" difference implies that distinctive attributes will

separate the three modes of social interaction. Although identifying interaction
modes is an important first step, it leaves unanswered the question of how these
modes are related. We might conceptualize a continuum with dominance at one
end, commitment at the other end, and exchange in the middle. Such a con-
tinuum would suggest three features of the relationship between the interaction
modes. First, on the extreme ends of the continuum and in the center we could
note the distinctive attributes that separate the three modes. That is, we can
rather easily distinguish between some areas along the continuum with respect to
the characteristic attributes discussed earlier. Second, when we move to the left
(or right) of center, but not to the end, on our continuum, it becomes difficult
to say whether dominance or exchange (exchange or commitment) exists. There
is some smooth transition between modes in which we might presume that some
attributes are declining, others increasing, and some mixture exists. Third, the
dominance and commitment modes are inaccessible without moving through the
exchange mode. If two parties are in a dominance interaction and if the nature
of that interaction were to change, it must move to an exchange relationship.
Similarly, change from commitment implies movement to an exchange interac-
tion. Thus, the assumption of a continuum from dominance to commitment
leads to testable propositions. Particularly interesting are the propositions tha~:
(a) smooth transitions occur between modes of social interaction, and (b) transi-
tions must occur from dominance to exchange to commitment or the reverse.
Although these are plausible propositions, we wish to examine a different, and
perhaps more descriptive, conceptualization.
Rene Thom, a French mathematician, has recently applied typology to
develop a mathematics applicable to discontinuous events, which has been
labelled catastrophe theory. This innovation has particular relevance to the be-
havioral sciences where differential calculus is inadequate for describing the
sudden breaks in behavior that are often observed in social interaction. At this
point, the mathematics may be less important than the conceptual view offered
by catastrophe theory. In the future, when important behavioral problems have
been conceptualized as catastrophes, then a powerful mathematics will become
available for their solution. Here we will apply a graphical analysis to show how
movements between modes of social interaction can be conceived as catastrophes.
More important, we will also show how provocative hypotheses can be derived
from the catastrophe theory conceptualization. The theoretical work by Bould-
ing, Walton and McKersie, and Bonoma suggests that the catastrophe model is
appropriate for explaining movements between interaction modes. Recall that in
each of the three typologies two categories were closely related, while the third
differed in kind rather than degree. With respect to our composite typology,
dominance and exchange are similar in their focus on an individual party in
interaction with another party. In contrast, commitment interactions center on
the inseparable dyadic unit. Thus, if the interaction mode were to change from
either dominance or exchange to commitment, we would expect a major break
or "catastrophic" change in the basic nature of the interaction.
Catastrophe Model Attributes
The three-dimensional graph in Figure 1 depicts the relationship between six

constructs: the three modes of social interaction (dominance, exchange, commit-
ment) and competition, cooperation, and community. Competition and coopera-
tion are viewed as independent parameters (Le., with perpendicular axes)
defining a horizontal plane. This plane is called the control surface and com-
petition and cooperation are referred to as control parameters. The assumption
is made that behavior represents an equilibrium between the two control param-
eters, cooperation and competition, which in effect determine behavior. Thus,
cooperation and competition do not refer to sets of behaviors, but rather to
basic underlying motivations. Under conditions of social interdependence there
exist mixed motives with respect to the existence and nature of this interdepen-
dence. That is, the parties have a cooperative motive to act in concert for the
benefit of the dyad and a competitive motive to obtain outcomes most congruent
with their individual goals. In summary, cooperation and competition are not
B
conun'itment
Conununity
Figure 1. Three-dimensional Graph Showing the Relationship between Six Constructs.

ends on either a behavioral or motivational continuum, but independent motiva-

tional factors. They may vary from high to low without reference to the other,
and, most interestingly, may simultaneously be high.
The behavior that is controlled by cooperation and competition we have
labelled community. By community we mean not a single behavior but a set or
repertoire of behaviors stemming from a party's sense of oneness with the other,
the degree to which the dyad is viewed as a unit, and the party's identification
with superordinate goals. In Figure 1 community is depicted by an axis perpen-
dicular to the control surface (defined by cooperation and competition) and is
termed a behavior parameter. For many points on the control surface, deter-
mined by the values of both cooperation and competition, there exists a most
likely or probable sense of community. For some points on the control surface
there may be two highly probable degrees of community. Thus, for each point
on the control surface there exists either a most probable community value or
two highly probable values for community.
When unique values of cooperation and competition are taken with an
associated value or values of community, they defme a point or two points in
three-dimensional space. Joining all such points produces a smooth surface called
the behavior surface as shown in Figure 1. The most interesting feature of this
behavior surface is the double fold or pleat, which results from the fact that in
some area of the control surface there is bimodality of the most likely com-
munity behavior. Actually, there should be a break in the behavior surface where
it folds under to make the pleat because in this region we are projecting three
points rather than Simply the two points representing the two highly probable
community behaviors. In order to maintain the smooth behavior surface, this
middle point is taken to represent the least likely community behavior. The
folded surface between the upper and lower surfaces thus represents least likely
behavior and is, as we shall see, inaccessible and therefore irrelevant to our
analysis. In summary, community is a probable set of behaviors controlled by
the parties' cooperative motivation and competitive motivation. The probabilistic
rather than deterministic nature of community is similar to the conceptualiza-
tion of most dependent variables in the behavioral sciences. Not only is
community probabilistic but it is unimodal in some regions and bimodal in other
regions with respect to its most likely value. It is this bomodality of community
behavior which gives the catastrophe model of social interaction its most inter-
esting property.
Examining the behavior surface in Figure 1, we can distinguish between
different areas on the behavior surface by their height above the control surface
(i.e., by the general level of community). Thus, dominance represents a low,
exchange a moderate, and commitment a high level of community. The catas-
trophe model therefore suggests a more complicated relationshIp between
interaction modes than the simple dominance-exchange-commitment continuum

we discussed earlier. To better understand this relationship we will conduct a
comparative statics analysis that suggests the dynamics inherent in the model.
Let us begin our analysis by introducing a simple example of an organizational
interaction between two divisions. Atlanta Industries is a large conglomerate
with several autonomous, profit-centered divisions. Among these divisions is
Birmingham Manufacturing, which produces orezes. Atlanta Industries purchases
Canton Products, which also makes orezes, and operates it as a separate division.
However, because it is much larger and has greater political influence with
Atlanta's top management, Birmingham is able to successfully dominate Canton.
This dominance relationship is depicted by point A in Figure 1. In an effort to
escape Birmingham's dominance, Canton begins to develop new product lines.
As the new products are introduced, the competitive attitude between Canton
and Birmingham gradually declines. Eventually, some of Canton's products
become raw material inputs for Birmingham. Exchange thus becomes the basis
for the interaction between the two divisions. Competitive attitudes continue
to decline and cooperation begins to increase. As Canton grows, personnel are
transferred from Birmingham. These new personnel retain ties with Birmingham,
which continues to build a cooperative feeling. However, the new personnel also
have a strong competitive motivation to exceed their former division and com-
petition therefore increases. Finally, Birmingham and Canton become committed
as represented by point C in Figure 1. Our example illustrates that movements
from dominance to commitment can occur by tracing a curved path from A to
B that goes above the pleat and avoids crossing the bimodal region. This path
must move through the exchange region before reaching the commitment region.
It is therefore possible to have a smooth transition between the different modes
of interaction, as we discussed earlier with respect to the continuum view.
However, the catastrophe conceptualization offers an important refmement to
the continuum notion. Remember that movements along the behavior surface
result from changes in the control parameters, cooperation and competition. To
go above the pleat requires a marked reduction in competition. Moreover, this
reduction in competition must be accomplished without a large enough increase
in cooperation to pull the path into the bimodal region. We can thus deduce
testable hypotheses from the catastrophe model. First, smooth transition from
dominance to commitment requires movement through an exchange relation-
ship. Second, this smooth transition requires that a maximal level of competition
not be exceeded. Third, the smooth transition will be broken if cooperation
increases too rapidly relative to the decrease in competition. To fully understand
the meaning of these hypotheses, particularly the third, it is necessary to examine
a second basic way in which interactions can move from dominance to commit-
ment (Le., from point A to point B).
Let us return to our earlier example, but with a different scenario. Suppose
that soon after acquiring Canton Products, the top management at Atlanta
Industries had taken a more active stance. Birmingham and Canton are directed
by Atlanta to work closely together in developing a new product (Le., a super-
ordinate goal is provided). Atlanta supplies monetary resources, extensive staff
support, and top management time. The coordination between Birmingham and
Canton is closely monitored. In short, the high level of competitive motivation
between Birmingham and Canton is taken as given and efforts are directed
toward increasing cooperative attitudes. Finally, Atlanta management is satisfied
that they have built a "real team."
The relationship between Birmingham and Canton moves from point A to
point B in Figure 2 by a different path than in the previous example (compare
with Figure 1). Initially, movement would occur toward the fold along path AC.
Such movement might represent some change in community, but would still fall
within the dominance region. However, what happens at point C is very different
community
Figure 2. Three·dimensional Graph of Catastrophe Model.

from the previous movement. At C we reach an edge in the pleat; continued

movement in the dominance region is no longer possible. If the control param-
eters (cooperation or competition) continue to increase their pressure, the edge
is passed and the point moves suddenly or "catastrophically" to point B in the
commitment region. It is this sudden break or catastrophe which is the most dis-
tinctive feature of the model and lends it its name. Note that the path cannot
travel the surface representing the underside of the curve, since this region repre-
sents the least likely behavior and is therefore inaccessible. Movement alongACB
through the pleat (see Figure 2) has four important contrasts with the pathAB
around the pleat (see Figure 1): (1) change from a dominant mode to a com-
mitted mode occurs without passing through an exchange mode; (2) a sharp
break, rather than a smooth transition, in the nature of the interaction occurs;
(3) no maximum value of competition need be enforced; and (4) the transition
can be effected by increasing only the level of cooperation.
It will be instructive for our purposes to examine more closely the bimodal
region of the behavior surface. First, note that the two edges of the pleat
(termed the fold curve) may be projected onto the control surface where they
form a cusp. The control surface from Figure 2 is highlighted in Figure 3. The
area within the cusp (XYZ) represents the area where community behavior is
bimodal and is called the bifurcation set. The path A'C' on the control surface
represents the changes in competition (if any) and cooperation that forced the
movement ACB on the behavior surface (see Figure 2). Notice how the first fold
curve (XY) is crossed and the bifurcation set entered without any change in the
interaction mode, which continues to be one of dominance. Remember, we are
moving along the lower behavior surface and travel underneath the fold curve
along the upper surface. However, when we reach C' we reach the portion of the
fold curve (YZ) on the lower surface and further movement results in a sudden
transition to a commitment mode. Now, if we should reverse the direction of
movement from F' to A', we would cross the fold curve YZ while remaining in
the commitment mode. That is, we are moving along the upper fold as, for
example, in the path BD in Figure 2. Only whenD' is reached on the fold curve
XY do we return to the dominance region. This asymmetric property of move-
ment across the bifurcation set is termed hysteresis and is a major feature of the
catastrophe model. We wish to explain two other important features of the
catastrophe model before summarizing the implications of the model.
Again referring to Figure 3, we can see that the axes of the control surface
have been placed differently with respect to the cusp. The competition axis
bisects the cusp and is called a splitting factor. Increases in competition in-
crease the size of the split between the upper and lower portions of the behavior
surface. Cooperation is called a normal factor because increases in cooperation
correspond to smooth movements along the back of the control surface. Both
><1
.;
U
II>
't
:::J
e...
CI)
c:
0
U
.....
0
&.
Q.
f!
t!l
M
E
:::J
.~
~I II..
298
control parameters may be treated as normal factors. However, we feel that

modeling competition as a splitting factor will be more descriptive of most
social interaction.
Consider again the fold in Figure 2. If we are at point P, just before entering
the fold at its beginning, then as we enter the cusp small movement in either
direction implies moving either to the upper or lower surface. For example, the
path PQ leads to the lower surface, while path PR leads to the upper surface.
Yet, the underlying control variables, cooperation and competition, need to
change little to produce this divergence in community behavior. Divergence is
thus a major descriptive property of the cusp catastrophe model of social inter-
action that we have been describing.
Research Implications
We have discussed the basic features of the catastrophe model: sudden breaks,
bimodality, hysteresis, and divergence. It will now be helpful to examine the
model as a whole and its implications for understanding social interaction. The
major purpose in presenting the catastrophe model is to derme the relation-
ship between the dominance, exchange, and commitment modes. In order
to truly derme this relationship, it is necessary to consider the dynamics of social
interaction.
Dominance, exchange, and commitment were modeled as three areas on the
behavior surface of a catastrophe model. Within a given area the slope of the
behavior surface is relatively flat. This suggests that movements within a partic-
ular interaction mode involve only small changes in community. That is, the
notion of an interaction mode implies a stable interaction pattern or equilibrium
between the parties. When we refer to an equilibrium, we also imply there is
some pull or attraction between the two parties that binds the interaction to a
familiar pattern. The same basic attractor may operate in different contexts;
for example, a magnetic attractor in three different physical systems. However,
in social interactions we believe that three different simple attractors exist, one
for each interaction mode. In the dominance mode, it is the high power party
that uses his or her power to establish and enforce the prescribed interaction
pattern. For exchange, the interaction pattern is determined through many past
exchanges of valued factors. Interactions in a commitment relationship are
circumscribed by a network of norms, beliefs, and values. The attractor in
dominance is power, in exchange it is reinforcement, and in commitment the
attractor is cohesiveness. As we discussed earlier, understanding and predicting
social interaction within a given context requires selecting and applying an
appropriate theoretical model. The power, reinforcement, and cohesiveness
attractors are the social glue in the influence model, social exchange model, and
commitment model, respectively.
Once a given interaction mode (dominance, exchange, or commitment) is
accepted by the parties, the attractor (power, reinforcement, or cohesiveness)
acts to maintain stability within the mode. However, stability can be over-
come and the parties may enter a new relationship. Such change from a stable
equilibrium can only occur when new forces enter the situation. The effects of
such forces are summarized by changes in the underlying control parameters,
cooperation and competition. Later we will consider why and how changes in
cooperation and competition occur, but for now we will assume that enough
change has taken place so that two attractor forces are affecting the interaction.
In an area where the two attractors are nearly equal in their pull, subtle changes
in either direction may dictate movement back to the old mode or change to a
new interaction mode. Unlike objects in a physical system, the parties do not
simply passively react to attractors. Rather, the parties sense the pull toward a
new interaction mode and react to this pull.
We will call the areas where the pull between attractors is nearly equal conflict
zones. In these conflict zones the parties recognize that a basic change in their
interaction is eminent. We can distinguish two forms of conflict: outcome con-
flict and relationship conflict. Outcome conflict is the conflict that ensues
within a given interaction mode. This usually involves differences in preferences
for outcomes where the outcome itself will have no effect or only marginal
effect on the nature of the future interaction mode. That is, the basic interaction
mode endures unchanged by some momentary conflict. Relationship conflict
develops when the parties are in the conflict zone. Here, conflict results not
from different preferences over outcomes but from different desires concern-
ing the basic social relationship that should exist and how that relationship
should be defined. Outcome conflict and relationship conflict are not
independent, since continued outcome conflict may result in change in the
underlying control parameters ultimately leading to a conflict zone. However,
the immediate effects of outcome conflict are neither as serious nor as pervasive
as those for relationship conflict. When outcome conflict breaks out, the attrac-
tor force acts to pull the relationship back into stable equilibrium. Moreover,
the parties have vital interests at stake in the present relationship and anticipa-
tions of how their interests will be met in a new relationship. Thus, because the
equilibrium is weak and the issues vital, we would expect any relationship con-
flicts to be more violent than the more typical outcome conflicts. Because the
parties can foresee the strong possibility of serious relationship conflict occurring
as the conflict zone is entered, we would expect more intensive bargaining
between the parties. Successful bargaining will move the relationship to a new
interaction mode or back into the previous equilibrium. The failure of bargaining
will lead to an overt relationship conflict between the parties.
The conflict zones occur on the behavior surface near areas of transition
between social contexts. Thus, there is a conflict zone on: (1) the lower surface
between exchange and dominance, (2) the lower surface near the fold curve
marking the break between exchange and commitment, (3) on the lower surface
near the fold curve marking the break between dominance and commitment, (4)
on the upper surface from commitment to exchange, and (5) on the upper
surface from commitment to dominance. These five conflict zones can be
distinguished by the degree to which community behavior would shift should a
transition through the zone to a new social context occur. This difference
suggests that the intensity of bargaining behavior will vary with the magnitude of
the shift in community behavior. That is, we would expect bargaining would be
more intense and the probability of relationship conflict higher as the difference
between the current and anticipated interaction situation increases. With respect
to the five conflict zones noted above, bargaining intensity and relationship
conflict potential are greatest in the dominance-commitment and commitment-
dominance zones, moderate in the exchange-commitment and commitment-
exchange zones, and low in the dominance-exchange zone.
SUMMARY
We have elected to view social interaction in terms of three modes whose

relationship can be described by a cusp catastrophe. This perspective has allowed
us to deduce certain testable propositions about the three modes of social inter-
action. Among the most important propositions are:
1. Transitions between dominance and commitment are accompanied by

sudden breaks in community.
2. Transitions between exchange and commitment occur smoothly if an
upper value for competition is not exceeded. When competition exceeds a
maximum value during the transition, a break occurs.
3. Transitions between dominance and exchange occur smoothly.
4. In the area near the beginning of the bimodal region slight changes in
cooperation and competition will result in either exchange or commit-
ment (divergence).
5. Change to a new interaction mode implies establishing a stable relation-
ship (hysteresis).
6. Bargaining and the potential for relationship conflict are greatest in the
regions where the attractor forces for two interaction modes are nearly
equal (Le., in the conflict zones).
7. Bargaining intensity varies directly with the anticipated levels of change in
community.
Although catastrophe theory can be useful in a purely phenomenological

sense, it promises to make even greater contributions. First, Thorn has proven
that if a maximum of four control variables influences a behavior, then this
behavior can be described by one of the seven elementary catastrophes. The cusp
catastrophe applied in our analysis, although providing many insights, is only
one of these seven elementary catastrophes. The six other catastrophe models
also merit attention in examining social interaction. Thus, we are fortunately
limited to only seven models when examining most social situations (Le., those
with four or less control parameters). Second, there is a powerful mathematics
underlying catastrophe theory, which is finding increased application. If his-
torical data can be obtained on the relationship between parties, then catastrophe
theory may be able to provide more accurate predictions than traditional sta-
tistical techniques.
Our analysis has indicated that catastrophe theory can be tested and grounded
at strategic points within widely varying theoretical domains. We further
contend that catastrophe theory permits the consolidation of more specific
theories of social interaction, thus enabling empirical observation to be tied to a
smaller number of linking constructs. This emphasis on consolidation counters
the frequent criticism that middle range theorizing lacks a sound theoretical
basis (Denzin 1970). This conclusion stems from the belief that too restrictive
and limited an approach produces splintering and parochialism. However, if this
consolidation is to occur, it requires validation of the assumption that plausible
inferences can be drawn from the model and empirically tested.
REFERENCES
Blau, P.M. Exchange and Power in Social Life. New York: John Wiley and
Sons, 1964.
Bonoma, T. V. Conflict, cooperation and trust in three power systems. Behavioral
Science, 1976,21,499-514.
Boulding, K.E. Economics as a Science. New York: McGraw-Hill Book Com-
pany, 1970.
Denzin, N.K. The Research Act. Chicago: Aldine Publishing Company, 1970.
Gouldner, A. The norm of reciprocity: A preliminary statement. American
Sociological Review, 1960,25, 161-179.
Homans, G.c. Social Behavior: Its Elementary Forms. New York: Harcourt,
Brace and World, 1961.
Kiesler, C. A. The Psychology of Commitment. New York: Academic Press, 1971.
Thibaut, J.W., and Kelley, H.H. The Social Psychology of Groups. New York:
John Wiley and Sons, 1959.
Thorn, R. Structural Stability and Morphogenesis (translated by D.H. Fowler).
W.A. Benjamin, Inc., Reading, Massachusetts, 1975.
Walton, E., and McKersie, R.B. A Behavioral Theory of Labor Negotiations.
New York: McGraw-Hill, 1965.
19, 357-367.
Zeeman, E.C. Catastrophe theory. Scientific American, 1976, (April), 65-83.
25
THE APPLICABILITY
OF MIDDLE RANGE THEORIES
TO THE STUDY OF ORGANIZATIONAL
EFFECTIVENESS
Paul M. Muchinsky and Paula C. Morrow,
Iowa State University
Middle range theorizing has been proposed as an efficient means by which sets of
hypotheses can be integrated and synthesized into submolar theory, such that
the hypotheses are grounded in empirical confirmation while lending themselves
to a higher·order theoretical abstraction. This approach can be construed as an
epistomological strategy of moderation or middle ground, on a continuum of
abstraction above organized descriptive data but beneath comprehensive theories
that attempt to explain large-scale social phenomena. According to Merton
(1967), middle range theories consist of limited sets of assumptions from which
specific hypotheses are logically derived and confirmed by empirical investiga-
tion. Advocates of middle range theories (Merton, 1967; Weick, 1974) have
outlined three particular characteristics of these theories that are of special con·
cern to this paper. First, middle range theory involves abstractions, but these
abstractions are close enough to observed data to be incorporated in proposi-
tions that permit empirical testing. Secondly, a given theory of the middle range
having a measure of empirical confirmation can often be subsumed under com-
prehensive theories, which are themselves discrepant in certain respects. Thirdly,
an effective strategy of scientific inquiry is to develop theories of the middle
range and consolidate them periodically_ The focus of this paper is on the
applicability of middle range theorizing to the construct of organizational
304
THE STUDY OF ORGANIZATIONAL EFFECTIVENESS 305
effectiveness, a construct that has met with limited success when conceptualized
via traditional modes of analysis, and accordingly may be more profitably
studied through the use of middle range theories.
The feasibility of applying middle range theories to the construct of organiza-
tional effectiveness will be examined in part by focusing upon the three cited
characteristics of middle range theories and the potential they have for reducing
some of the empirical and theoretical ambiguities associated with organizational
effectiveness research. The format of this paper consists essentially of two parts:
a review of three pronounced problems in organizational effectiveness research,
and an examination of how middle range theories might alleviate some of
these problems.
THE CONSTRUCT OF ORGANIZATIONAL EFFECTIVENESS
As Steers (1975) has discussed, there are numerous conceptual problems with
the construct of organizational effectiveness. These problems range from a lack
of convergence among theorists as to what the construct of organizational ef-
fectiveness should entail, to data that offer weak empirical confirmation of the
construct. While a complete review of all the problems currently plaguing orga-
nizational effectiveness researchers is beyond the scope of this paper, three
substantive problems will be presented to which the concept of middle range
theories may offer some answers. More detailed discourses on the problems asso-
ciated with organizational effectiveness research are presented in Spray (1976)
and Goodman and Pennings (1977).
1. Global Theoretic Conceptualizations. As Steers (1975) commented, most

theories of organizational effectiveness attempt to be explicated in ways that can
be applied to a wide variety of organizations. Child (1974) has termed these
"universalistic" models. Caplow (1964) set as his goal the development of a
"single theoretical model, although rough and incomplete, (that) can be used to
analyze organizations of any type or size, regardless of their cultural or historical
location ... " (p. v). Georgopoulos and Tannenbaum (1957) proposed a gener-
alized model when they argued that the three evaluative criteria they set forth
could "potentially apply to nearly all organizations." The problem with these
theoretical frameworks is that they are too broad and too abstract to be of much
value to enhance our empirically based understanding of organizational effective-
ness. These theories receive only very mild empirical verification for two major
reasons: (1) there is too much "distance" between the theorizing and the data
(with the concomitant increase in measurement error typically associated with
increased distance), and (2) organizations are sufficiently different to escape cap-
ture under one theoretical umbrella, that is, by their nature some organizations
just don't readily fit into one theoretical framework.
While the universalistic framework seems to be a popular one, other types of
theories have also been proposed to explain organizational effectiveness. Con-
ceptualizing organizational effectiveness as a series of goals has been advocated
by Hall (l972), Price (1972), and others. Alternately, Yuchtman and Seashore
(1967) have pursued the study of organizational effectiveness via the system
resource approach. Some investigators (Molnar and Rogers, 1976) have empiri-
cally compared the two approaches, although Hannan and Freeman (1977) have
described the problems in trying to make across-study comparisons. Puzzlements
exist even within theoretical frameworks, as a lack of consensus within an orga-
nization as to what the goals of the organization are or should be (Dubin, 1976).
When the empirical data do not support the propositions of a theory, one can
attribute the lack of confirmation to poor theory or poor measures of the con-
cepts under investigation. In most cases the question of the quality of the mea-
sures can be resolved through traditional psychometric criteria (e.g., statistical
reliability and validity). The question of poor theory is a tougher issue to re-
solve, as there are many determinants of theory quality (Marx and Hillix, 1963).
It does seem likely, however, that in the area of organizational effectiveness,
theories may not be so much an issue of "poor" as they are "inappropriate"
given the stage of development in organizational effectiveness research. Many of
the more universalistic theories may be trying to accomplish too much, given the
relatively uncertain state of organizational effectiveness research. More modest
degrees of theorizing may culminate in greater degrees of success in explaining
the construct of organizational effectiveness, if for no other reason than the
theoretical domain will be less heterogeneous. The current state of the art on
organizational effectiveness (e.g., Spray, 1976; Goodman and Pennings, 1977)
illustrates the lack of agreement among scholars as to what is meant by "orga-
nizational effectiveness," how it should be studied, and even whether the concept
is a useful one or not (pfeffer, 1977). Given this situation it seems reasonable for
scholars to retrench from the molar and grandiose theories of organizational
effectiveness (that have not met with much professional favor) to less ambitious
theories reflecting in part our current limited comprehension of the topic.
2. Construct Validity. The issue of construct validity entails a whole host of

dimensions. Briefly, construct validation consists of: (I) identifying the domain
of the relevant criteria and (2) determining the extent to which these dimensions
are similarly related. Steers (1975) and Campbell (1976) have presented exten-
sive lists of variables that have been used as measures of organizational effective-
ness, including such variables as productivity, profitability, survival, employee
satisfaction, adaptability, and so forth. A major problem is that different re-

searchers select the measures they wish to examine, and there is not much con-
sistency in the measures selected. The problem is further illustrated in a study by
Friedlander and Pickle (1968) who found that different reference groups (e.g.,
owners, employees, customers) identify with different types of effectiveness
measures (e.g., profitability for owners, satisfaction for employees, etc.). Such
diverse frames of reference explain in part the lack of consensus concerning
appropriate evaluative criteria.
In addition to the identification problems, there have been problems in the
interrelationships among the measures that supposedly define effectiveness. If
these measures are all indicators of the same concept (organizational effective-
ness), then these measures should all be positively interrelated. However, the
empirical data have revealed that some of the measures comprising organizational
effectiveness are unrelated to each other while others are actually negatively
interrelated. An example of this latter case might be the relationship between
productivity and satisfaction, in the case where the organization pressures the
employees to increase productivity which in turn engenders lowered feelings of
job satisfaction. Independent and negatively interrelated measures of supposedly
the same construct do not meet the traditional standards for establishing con-
struct validity.
In a somewhat different vein there is an awkward juxtaposition of using
profitability as a standard measure of organizational effectiveness when the orga-
nization(s} in question are nonprofit. A similar problem occurs in trying to use
measures of productivity as measures of effectiveness in organizations that are
service, rather than production, oriented. Previous authors have similarly decried
some of the seemingly inherent problems in establishing the construct validity
of organizational effectiveness. Goodman and Pennings (1977), in addition to
citing the problems of constituencies (different reference groups), organizational
types, and the domains of effectiveness, also describe the problem that some
researchers view variables to be indicators of effectiveness while other researchers
view those same variables as determinants of effectiveness. Additionally, as
Steers (1975) comments, the time dimension is also a major consideration;
short-run measures of effectiveness may not be equivalent to long-run measures
of effectiveness. In short, the problem of construct validity has not been satis-
factorily met in current organizational effectiveness research. The degree of
thorniness of this problem appears to be exacerbated by universalistic theories
that try to embrace multiple perspectives simultaneously. Middle range theories
may offer a partial promise of relief to this problem by identifying logically and
empirically derived sets of hypotheses that could serve as a nucleus for integrat-
ing the fmdings on organizational effectiveness.
3. Inductive-Deductive Models of Inquiry. The investigation of organizational

effectiveness can be initiated at a highly abstract, theoretical level where general
principles are formulated and then tested through the careful application of de-
ductive reasoning and formal logic. Conversely, one can proceed inductively,
putting aside preconceived notions about organizational effectiveness, and allow
the data alone to guide the theory construction process. The organizational
models of Bennis (1962) and Schein (1970) represent the deductive approach,
while the model by Yuchtman and Seashore (1967) represents more of an induc-
tive approach. There are, however, advantages and disadvantages associated with
each approach. It may be possible for middle range theories to retain the advan-
tages of both approaches while simultaneously reducing the impact of their
respective disadvanages.
The deductive process of developing indicators of organizational effectiveness
follows the model by first selecting a theoretical perspective, explicating the
component dimensions specified in the concept, and then developing measures
for each of the dimensions posited. The validity of measures derived through this
process is primarily based on the extent to which the intended meaning sphere
of each dimension is reflected in the measure. These points are stressed in order
to facilitate theory testing in diverse organizations and hence contribute to a
general theory of organizational effectiveness. A major disadvantage of this de-
ductive approach is that the measures of effectiveness generated may fail to
demonstrate statistical reliability and validity since there is considerable "dis-
tance" between the operational measure and the theoretical concept requiring
inference (Le., low epistemic correlation).
Researchers who choose the inductive approaches tend to argue that effec-
tiveness criteria should be sought with reference to the organization under
investigation. In this approach the achievement of reliability and validity in the
measures of effectiveness is facilitated by the more grounded orientation. The
heavy reliance on data as a theoretical building block requires that the organiza-
tional effectiveness measures employed manifest desired pyschometric attributes
prior to any theoretical interpretation. There tends to be a high degree of
isomorphy between the effectiveness measures and the concepts used to describe
them in this approach since the concepts are not assigned until after the data
collection process is completed. These advantages of the inductive approach are
offset by: (1) the potential narrowness of the emergent concepts and (2) the
difficulty of using the same measures in different organizations.
In summary, inductive approaches seem to be characterized by strong mea-
sures but weak generalizability, while deductive approaches offer the promise of
generalizability but frequently suffer from weak empirical verification. A middle
range theory perspective might be a reasonable solution to the dilemma. A mid-
dle range theory would "exist" (in conceptual space) somewhere between the
empirical data (with little or no abstractions) and the overriding theoretical per-
spective (at a very high level of abstraction). The middle range theory would be
"close" enough to the data to deter the problems of increased measurement
error that occur when attempting to traverse great distance between theory and
data. Thus, a major advantage of the inductive approach would be retained. Per-
haps research would show that one middle range theory would not be sufficient
to explain organizational effectiveness, thus necessitating the use of multiple
middle range theories to explain the construct. These multiple middle range
theories would then become the conceptual target of a larger, more general
theory of organizational effectiveness. Thus, the deductive approach would work
down to the content of the middle range theories, which was derived via the
inductive approach. It appears that working through a middle range theory per-
spective might be a most efficient way to study the construct of organizational
effectiveness, a construct that has not met with much success when studied via
one singular perspective (inductive or deductive).
UTI LlTY OF MIDDLE RANGE THEORIES
This paper has focused upon three particular issues currently facing researchers
in the area of organizational effectiveness. Many other additional types of prob-
lems are also present as has been documented by Steers (1975), Hall and Clark
(1977), and Goodman and Pennings (1977). The remainder of this paper will be
devoted to exploring how middle range theories might alleviate some of these
problems, particularly in regard to the three characteristics of middle range
theories cited earlier in the paper.
Because the problems associated with organizational effectiveness are inter-
twined, it is difficult to address them separately. This is especially true with the
issues of reduced theoretical range and construct validity, as a decrement in the
scope of a theory would in all likelihood produce a decrement in the diversity of
variables used to demonstrate its construct validity. This dependency among
problems notwithstanding, a reasonable place to begin would be to abandon the
goal of developing the (general) theory of organizational effectiveness. Research
has shown that this objective has to date not been met with any great degree of
success. An alternative strategy, therefore, would be to develop a series or set
of theories that address a subset of the content matter of a general theory. We
are therefore proposing that the (general) theory of organizational effectiveness
be supplanted with a number of middle range theories. The number and type of
middle range theories would depend upon the manner in which scholars wished
to partition the construct of effectiveness.
This might be accomplished by one of two ways. One way would be to seg-
3lO MIDDLE RANGE THEORY AND THE STUDY OF ORGANIZATIONS
ment the scope of the theory's content, as developing a (middle range) theory
of organizational effectiveness that addresses effectiveness only in the short run,
or effectiveness as judged by one constituency (e.g., employees). Such an ap-
proach would serve to reduce the complexity of the construct by temporarily
excusing the scholar from the need to embrace both short- and long-run per-
spectives, multiple constituencies, and so on. After logically sound and em-
pirically supported middle range theories (i.e., theories of reduced scope) are
acquired, the scholar's next task is to reunite the segmented middle range
theories into a higher-order theory of broader scope. This approach to generating
larger-scale theories is fundamentally different from one where a scholar reduces
the scope or range of a construct under investigation without the intent of
studying the other elements or units comprising the construct, and without the
intent of integrating the segmented bodies of knowledge into a broader theo-
retical framework.
A second middle range approach to the study of organizational effectiveness
would be to selectively group the elements of analysis (i.e., organizations) into
homogenous categories. For example, separate middle range theories could be
devoted to private versus public sector organizations, production versus service
oriented organizations, and so forth. Greater refmement could be attained by
limiting the types of organizations embraced by a theory as hospital, prison,
school, and military organizations. It would again be the researcher's objective
to first empirically support each theory within organizational types, and then
seek to consolidate and integrate the findings across organizational types, striv-
ing for generalizability and cross-validity. It would seem that the range of vari-
ables needed for inclusion in the nomological net of construct validity would be
far less via these approaches than for a general theory of effectiveness. If the
focus of a given middle range theory is, for example, nonprofit hospitals, vari-
ables dealing with financial gain can be discarded as being inappropriate to the
theory, as well as variables that are irrelevant to a service-oriented organization.
Certain variables would seemingly be relevant for all types of organizations as
stability (or survival) and employee satisfaction. These types of variables might
serve as "marker" variables, measures that can be used for comparison of orga-
nizations that span the focus of different middle range theories. However, our
initial concern should be the development of relatively clean and tight theories
of effectiveness that possess a high degree of internal clarity and conciseness.
The best path to this goal would seem to be one of limiting or narrowing our
focus of investigation, going from developing the general theory of effectiveness
(applicable to all organizations) down to the middle range theory aimed at ex-
plaining effectiveness in just a certain type of organization, or effectiveness at
one point in time. It seems the construct validity waters have been muddied by
incorporating variables that simply "must" be considered because of their
presence in some overarching theoretical framework. Forcing diverse variables

into a conceptual space adds unique variance, which de facto does not increase
common variance. Given a middle range perspective, variables are added to the
criterion space as justified by the specific theory, not forced into the model as
dictated by some global theoretical perspective. Middle range theories served to
cluster logical groups of criterion variables in a selective way that global theories
do not and can not. Accordingly, the probability of attaining success in the em-
pirical validity of a homogenous battery of variables is far greater than when
working with a heterogenous mixture of variables that are related more in spirit
than in fact.
APPLICABILITY OF MIDDLE RANGE THEORIES
The three characteristics of middle range theories presented earlier in this paper
can be used to assess the potential applicability of middle range theories to the
study of organizational effectiveness. The first characteristic involves the use of
abstractions that are close enough to observed data to permit empirical testing.
This characteristic seemingly could be met without much difficulty, because the
levels of abstraction are not that distant from the observed data. In contrast,
there is considerable distance between the abstract concepts in a general theory
and the data. For example, in a middle range theory of effectiveness a theoreti-
cal abstraction might be employee stability, as operationalized by rates of turn-
over. However, a more general theory might be forced to use a higher level of
abstraction to account for the phenomena under investigation, such as negen-
trophy (Katz and Kahn, 1966), i.e., the capacity of a system to resist death
and disorganization. Thus, because middle range theories do not attempt to
explain large-scale phenomena, they are permitted to develop concepts at lower
levels of abstraction. Additionally, if one adopts an inductive approach, the
theoretical abstractions emerge directly out of the observed data and are thus
literally borne out of empirical verification. Such a system seems most desirable
in an area of research that has spawned a multitude of (primarily untested)
theoretical abstractions. It would thus appear that the first property of middle
range theories (Le., empirically testable abstractions) would be a reasonable
starting point for giving some systematic direction to the study of organiza-
tional effectiveness (Goodman and Pennings, 1977).
The second property of middle range theories that was presented involves
subsuming empirically confirmed middle range theories under general theories,
which are themselves discrepant in certain respects. Two such general theories
of effectiveness are the goal-centered and the natural systems models (Ghorpade,
1971). The general goal-centered model assumes that organizations have a set of
goals to pursue. Assuming that the goals can be identified, it should be possible
to plan the best organizational strategies for attaining them. With this frame-
work, therefore, the way to assess organizational effectiveness would be to
develop criterion measures to assess how well the goals are being achieved
(Campbell, 1977). The natural systems approach rejects the notion of specific
goal attainment and rather focuses upon maintaining the organization through
time. Organizational effectiveness as assessed by this approach partially entails
studying whether the organization's resources are being distributed over a variety
of coping (adaptation) mechanisms. It would seem that a middle range theory of
organizational effectiveness that operationalizes effectiveness in part in terms of
participative decision making or shared control would have something to offer
each of these two general theories, that is, goal-centered and natural systems.
Adopting the goal-centered approach, it could be argued that shared control
is a strategy for goal attainment, and those organizations (or types of organiza-
tions) that invoke shared control are more effective than those that don't (on
the basis that participative decision making is instrumental for attaining other
goals of the organization, as increased job satisfaction, decreased absenteeism,
etc.). If one adopts the natural systems perspective, it could be argued that
shared control is a necessary coping mechanism to ensure the viability of the
organization. Those organizations or types of organizations that share control
will survive longer than those that don't. While the example of shared control is
but one of many possible examples, it does seem that empirically verified middle
range theories may be encompassed by larger general theories, which themselves
are discrepant or adopt initially incompatible theoretical perspectives.
The third characteristic of middle range theories is that they should periodi-
cally be consolidated. This appears to be a most appealing property, especially
with regard to the study of organizational effectiveness. To be sure, researchers
are not interested in nor would they be satisfied with a multitude of middle range
theories of effectiveness, one theory for nonprofit hospitals, one for voluntary
organizations, one for prisons, and so on. Eventually, we would like to collapse
across theories, consolidating them into larger theoretical units. However, given
the problems that have occurred in attempting to verify global or general theories
of organizational effectiveness, the consolidation of many middle range theories
is probably a long-range goal. Given our current limited knowledge of organiza-
tional effectiveness, the development of a series of empirically strong middle
range theories of effectiveness seems to be (at least at this point in time) a chal-
lenging goal in itself. The eventual consolidation of middle range theories may
be accomplished over the time dimension (short run versus long run), type of
organization (product versus service, public sector versus private sector, etc.),
and perhaps over constituencies (employees versus customers). It also seems
likely that we will reach a point in time when we cannot consolidate all of the
theories into one general theory due to irreconcilable differences among orga-
nizations. To attempt a complete consolidation may serve to reduce the explana-
tory power and conceptual accuracy of the theories. While Bierstedt (1960) has
referred to the implementation of middle range theories as an "anemic ambi-
tion," the feelings of epistomological anemia may be offset by the increased
empirical verification that such midrange theories provide.
In conclusion, the applicability of middle range theories to the study of
organizational effectiveness seems to hold considerable promise. The observed
problems of global theoretical orientations, construct validity, and inductive
versus deductive approaches to organizational effectiveness appear to be miti-
gated by the construction of middle range theories. Midrange theories seem
particularly attractive as theoretical building blocks because they provide the
opportunity for grounded empirical verification of abstraction, which in turn
can be consolidated into larger theoretical frameworks. As Merton (1967)
stated, middle range theories specify the limits of our ignorance. Given the
current state of the art of organizational effectiveness research, middle range
theories seem to accord us the means by which we can increase our under-
standing of this vexing construct.
REFERENCES
Bennis, W.G. Toward a "truly" scientific management: The concept of organiza-

tional health. General Systems Yearbook, 1962,7, 269-282.
Bierstedt, R. Sociology and humane learning. American Sociological Review,
1960,25, 3-9.
Campbell, J.P. Contributions research can make in understanding organizational
effectiveness. In S. Spray (Ed.), Organizational Effectiveness: Theory, Re-
search, and Application. Kent, Ohio: Kent State University Press, 1976,29-
45.
--. On the nature of organizational effectiveness. In P.S. Goodman and J.M.
Pennings (Eds.), New Perspectives on Organizational Effectiveness. San Fran-
cisco: Jossey-Bass, 1977, 13-55.
Caplow, T. Principles of Organization. New York: Harcourt, Brace, and World,
1964.
Child, J. Managerial and organizational factors associated with company
performance -Part 1. Journal of Management Studies, 1974,11, 175-189.
Dubin, R. Organizational effectiveness: Some dilemmas of perspective. In S.
Spray (Ed.), Organizational Effectiveness: Theory, Research, and Application.
Kent, Ohio: Kent State University Press, 1976,7-14.
Friedlander, F., and Pickle, H. Components of effectiveness in small organiza-
tions. Administration Science Quarterly, 1968,13, 289-304.
Georgopoulos, B.S., and Tannenbaum, A.S. The study of organizational effec-

tiveness. American Sociology Review, 1957,22, 534-540.
Ghorpade, J. (Ed.). Assessment of Organizational Effectiveness. Pacific Palisades,
California: Goodyear, 1971.
Goodman, P.S., and Pennings, J.M. (Eds.). New Perspectives on Organizational
Effectiveness. San Francisco: J ossey-Bass, 1977.
Hall, R.H. Organizations: Structure and Process. Englewood Cliffs, N.J.:
Hall, R.H., and Clark, J.P. Organizational effectiveness: Some conceptual,
methodological and moral issues. Paper presented at the American Sociologi-
cal Association Meetings, 1977.
Hannan, M.T., and Freeman, J. Obstacles to comparative studies. In P.S. Good-
man and J.M. Pennings (Eds.), New Perspectives on Organizational Effective-
ness. San Francisco: Jossey-Bass, 1977, 106-131.
John Wiley, 1966.
Marx, M.H., and Hillix, W.A. Systems and Theories in Psychology. New York:
McGraw-Hill, 1963.
Merton, R.K. On Theoretical Sociology. New York: The Free Press, 1967.
Molnar, J.J., and Rogers, D.C. Organizational effectiveness: An empirical com-
parison of the goal and system resource approaches. SOCiological Quarterly,
1976,17,401-413.
Pfeffer, J. Usefulness of the concept. In P.S. Goodman and J.M. Pennings (Eds.),
New Perspectives on Organizational Effectiveness. San Francisco: Jossey-Bass,
1977,132-145.
Price, J. L. The study of organizational effectiveness. SOCiological Quarterly,
1972,13,3-15.
Schein, E.A. Organizational Psychology. Englewood Cliffs, N.J.: Prentice-Hall,
1970.
Spray, S.L. (Ed.). Organizational Effectiveness: Theory, Research, and Applica-
tion. Kent, Ohio: Kent State University Press, 1976.
Steers, R.M. Problems in the measurement of organizational effectiveness. Ad-
ministrative Science Quarterly, 1975,20, 546-558.
19, 357-367.
Yuchtman, E., and Seashore, S.E. A system resource approach to organizational
effectiveness. American SOCiological Review, 1967,32, 891-903.
26
ON RESEARCH
IN ORGANIZATIONAL SOCIALIZATION
The Case for Middle Range Theory
Daniel C. Feldman, Northwestern University
In The Structure of Scientific Revolutions, Thomas Kuhn argues that it is the

abandonment of critical discourse among competing schools of theory that
marks the transition of a field to a science. "To be accepted as the authority,"
writes Kuhn, "a theory must seem better than its competitors, but it need not,
and in fact never does, explain all the facts with which it can be confronted"
(Kuhn, 1962, p. 17). The acceptance of one paradigm ends constant reiteration
of fundamentals, gives scholars confidence they are on the right track so they
undertake more precise sorts of work, allows for ongoing communication among
researchers, and gives direction to both theory articulation and data collection.
In contrast to Kuhn, Paul Feyerabend holds that the existence of pardy over-
lapping, mutually inconsistent and yet empirically adequate theories is not only
possible but required (Peyerabend, 1962). Feyerabend argues that "theoretical
pluralism" will reduce the tendency to create ad hoc hypotheses in order to
handle falsifying instances of hypotheses and theories, and will enable scientists
to uncover new and unsuspected phenomena or "facts" (Lashchyk, 1970).
While certainly no theory can be considered the paradigm of research in
the field of organization behavior today-at least in the sense that Kuhn uses
"paradigm" -open systems theory is becoming a dominant framework for both
theory articulation and data collection. Despite open system theory's usefulness
315
in specifying the classes of variables and the relationships between variables

important to examine in organization behavior, some are wondering if the field
would make more progress if there were a greater number of "middle range"
theories being pursued. Each midrange theory might attempt to predict and
explain only a subset of all organizational phenomena, make different sets of
assumptions about organizations, require very different research strategies, and
lead to varying practical implications. These theories might be individually less
ambitious than more comprehensive theories in the short run, but they might
collectively make a larger contribution to the understanding of behavior in
organizations in the long run.
This essay considers four issues about the role of midrange theories in the
field of organization behavior today:
1. the meaning and nature of middle range theories

2. the criteria on which to judge the usefulness of middle range theories
3. the type of follow-through research that middle range theories require
4. some ideas on the appropriateness of middle range theorizing at this par-
ticular point in the development of organizational behavior as a field.
Throughout the paper, illustrative examples are taken from the author's
"A Contingency Theory of Socialization" (Feldman, 1976a). In that research, a
middle range theory of individual socialization into organizations is presented
and tested. The model: (a) identifies three distinct stages of socialization, (b)
specifies the activities engaged in by an individual at each stage, and (c) specifies
the personal and organizational contingencies that control an individual's move-
ment through the stages. Interview and questionnaire data collected from 118
hospital employees-nurses, nurse's aides, radiology technologists, tradesmen,
and accounting clerks - were used to develop, refine, and test the model. The dif-
ferences between the socialization experiences of professional, paraprofessional,
and nonprofessional workers are identified and explained as well.
THE MEANING AND NATURE OF MIDDLE RANGE THEORY
The term "middle range theory" is best known from Merton's Social Theory
and Social Structure: "theories that lie between the minor but necessary work-
ing hypotheses that evolve in abundance during day to day research and the all-
inclusive systematic efforts to develop a unified theory that will explain all the
observed uniformities of social behavior, social organization, and social change"
(1968, p. 39). While Merton does not give a comprehensive list of criteria on
ON RESEARCH IN ORGANIZATIONAL SOCIALIZATION 317
which to classify theories as middle range, some characteristics of middle range

theories emerge throughout his essay:
1. Middle range theories involve abstract concpts (Merton, 1968, p. 39).

2. These abstractions are close enough to observable data so that empiri-
cal testing of propositions may occur (Merton, 1968, p. 39). Middle
range theories involve more than just organized descriptive data or em-
pirical generalizations (p. 61).
3. Middle range theories involve sets of confirmed hypotheses; the hy-
potheses must not be "logically disparate or unconnected" (Merton,
1968, p. 61).
4. Middle range theories deal with limited aspects of social phenomena
(Merton, 1968, pp. 39-44).
5. Middle range theories cut across distinctions between micro and macro
problems (Merton, 1968, p. 68).
Merton also makes an important distinction between two types of middle

range theories. At one level, middle range theories can be "special theories"
from which to derive hypotheses that can be empirically investigated (p. 51).
Ultimately, however, Merton hopes that progressively more conceptual schemes
will evolve to consolidate groups of these "special theories" into a broad scien-
tific base; Merton argues for the "strategic importance of a graded series of
empirically confirmed theories" (p. 58).
Middle range theories may be seen to differ from more comprehensive the-
ories in five major ways: (1) the type of phenomena under investigation, (2)
the type of concepts used, (3) the kind of relational statements made between
concepts, (4) the methodology of the research to test the theory, and (5) the
rigor of empirical analysis used to evaluate the results of the research.
Phenomena. Middle range theories generally examine fewer variables and fewer
relationships between variables than overarching theories. Their focus is often on
one process in organizations - the variables that initiate, sustain, and cessate
some behavior-or on a small subset of organizational variables. In contrast,
more overarching theories try to provide frameworks in which many processes
and many variables can be understood. Rather than fOCUSing on several levels of
analysis-individual, group, organization, and environment-most middle range
theories focus on a problem that cuts across only one or two of these levels.
The type of phenomena that middle range theories try to explain may also be
different from those considered by overarching theories. A middle range theory
may examine a phenomenon that cuts across previously identified phenomena,
or it may set forth propositions that subsume previously stated middle range
theories. In "A Contingency Theory of Socialization" (Feldman, 1976a), for
example, socialization is the phenomenon under investigation; it cuts across pre-
viously identified phenomena in the selection, training, orientation, and role
development literature, and integrates some existing theories in those areas.
Concepts. At least in the first type of middle range theory that Merton discusses
-the more exploratory research from which empirically testable hypotheses can
be derived-the most important way in which the concepts used in middle range
theories differ from concepts in more comprehensive theories is their precision.
Concepts may be more global in nature, or aggregates of several facets of a vari-
able. For example, in "A Contingency Theory of Socialization," the concept
"congruence" is defined globally as "the extent to which the organization's
resources and individual needs and skills are mutually satisfying." The concept
"realism" is defined as "the extent to which individuals have a full and accurate
picture of what life in the organization is really like" -realistic expectations
about the job itself, the work group, and the organization are combined (Feld-
man, 1976a). Secondly, the reliability and validity of these operational defmi-
tions may be lower. At least in the initial formulation of the theory, there may
not be similar questionnaire items or interview schedules from previous research
which could be used in data collection. As a result, there may be enough conver-
gent and discriminant validity to establish the existence of the concepts but not
very high interrater or test-retest reliability.
As such theories get subsumed under more comprehensive middle range the-
ories, concepts become much more meticulously specified. Overlapping domains
of concepts should become less problematic as both further theory articulation
and data collection occur. Too, the psychometric quality of the instrumentation
should increase with larger, more complete data sets and validation studies.
Relational Statements. Establishing causality is central to scientific enterprise,

to identify the factors that make phenomena happen or change. It is equally
important to determine whether these cause-effect relationships are one-way or
two-way, direct or inverse (Weick, 1971). There are two ways in which middle
range theories can, and have, fallen somewhat short in dealing with causal
relationships.
First of all, middle range theories have often fallen short in dealing with
causality in the empirical testing of hypotheses. At least in more exploratory
research, the relational statements may be under-formalized; classes of variables
that should be related are identified, but the specific paths of causality are not
predicted. For example, in Feldman (I976a), the variables important at each
stage in the socialization process are identified and the relationships between
stages are predicted, but not the specific bivariate relationships between each
two variables in the study. Often, the initial test of the theory is cross-sectional
rather than longitudinal; the theory is often developed correlationally rather
than experimentally.
Secondly, the middle range theories often inadequately test for feedback, or
two-way, influence processes. For example, in the area of leadership research,
much of the early research-much of it middle range theory-concentrated on
the impact of leader behavior on subordinate behavior (e.g., Halpin and Winer,
1957; Fleishman, Harris, and Burtt, 1955), and did so through correlational
data. It has only been more recently (Lowin and Craig, 1968; Farris and Lim,
1969) that the mutual influence processes in leader-member relations became
investigated and done so through more experimental methodologies. Similarly,
in the socialization literature, much theoretical work has been on the impact of
"socialization agents" on new group members (Caplow, 1964); it has only been
more recently (Van Maanen, 1975; Feldman, 1976a) that more emphasis has
been put on the two-way direction of the influence process during socialization.
The comprehensive, overarching theories are often more conscientious about
acknowledging potential feedback causal processes.
At the minimum, middle range theories must establish associations of some
magnitude and consistency to establish solid covariance. The size of the associa-
tions must be large enough and the circumstances under which the associations
will occur must be identifiable enough to encourage further investigation. In the
evolution of larger-scale middle range theories, even more attention should be
given to the question of causation. There should be greater scrutiny of potential
feedback influence; there should be greater concern with research designs where
the independent causal variable either occurs first or changes prior to the depen-
dent variable (Labovitz and Hagedorn, 1971) and/or where two-way influence
processes can be teased out.
Methodology and Empirical Analysis. Middle range theories often are tested
with a research methodology rigorous enough to establish the plausibility of the
theory but not rigorous enough, at least in the early stages of theory formula-
tion, to reject the null hypothesis (if one exists). The size of the sample may be
too small. The variety of settings in which the theory is initially tested may be
too limited, (e.g., there may be only one organization, one type of organization,
or one employee classification in the study). As mentioned above, the refine-
ment and measurement of concepts may not be adequate, or the research design
of the initial theory tests may be suboptimal.
Two issues dealing with empirical analysis are of particular concern to the
midrange theorist. The first is the amount of variance in the phenomenon under
investigation that the theory will explain. The lower reliability and validity of
the concept measurements are one constraint on the amount of variance ac-
counted for. Also, because by definition middle range theories do not consider
all the variables that might possibly influence the phenomenon, they may
account for less variance in the phenomenon than might be expected with more
comprehensive theories.
Secondly, middle range theory testing often must use less powerful statistics
and tools of data analysis because of the lower formality of the relational state-
ments and the imperfections of the research methodology. Ideally, the contin-
gency theory of socialization should have been tested longitudinally through
path analysis; because of the cross-sectional data, it is tested with partial corre-
lations (Blalock, 1964).
CRITERIA TO EVALUATE MIDDLE RANGE THEORIES
The second problem to be considered in this essay is the criteria on which to

evaluate midrange theories. Traditionally, three criteria have been used to eval-
uate the utility of theories (Hempel, 1966; Labovitz et al., 1971):
1. Simplicity. The better theory uses the fewest number of variables and
hypotheses to explain a set of phenomena; a good theory should not need
a set of ad hoc hypotheses.
2. Prediction and control. A theory is considered important to the extent
that it can accurately predict the occurrence of a phenomenon or allows
the researcher or practitioner to control the phenomenon in actual
situations.
3. Precision of supporting evidence. The better theory has more reliable
and valid data to its support, and has evidence of its explanatory power
in a variety of situations.
The set of criteria on which to judge the utility of midrange theories, while it
might include the criteria listed above, could also include some different -and
perhaps less rigorous-criteria as well. Additional criteria on which to judge the
utility of middle range theories would include:
1. The establishment of the existence of a phenomenon. The convergent and

discriminant validities of new concepts should be large enough to esta-
blish their plausibility. The magnitude of associational/correlational/causal
data should be large enough to encourage further investigation.
2. Probability of hypotheses. The credibility of a hypothesis depends in
large part on all the evidence relevant to that hypothesis, all the hy-
potheses and theories currently accepted that have any bearing on it. The
new theory should either be consistent with an already present nomologi-
cal network or cogently explain the differences in its predictions (Hem-
pel,1966).
3. Testability-in-principle. While middle range theories may not have been
tested in the most appropriate manner-or, in fact might be completely
"logical" or "nonempirical" in nature -no middle range theory can be
significantly proposed unless it is amenable to objective empirical test
(Hempel, 1966), at least in principle (e.g., the technology to test the
theory may not currently be available).
4. Theoretical understanding. A middle range theory must increase our un-
derstanding of a phenomenon in some Significant way. It should either
offer a systematically unified account of diverse phenomena, or specify
the conditions under which previously formulated empirical laws hold
true, or predict and explain phenomena that were not known about be-
fore the theory was formulated (Hempel, 1966).
In evaluating "A Contingency Theory of Socialization" as a middle range

theory, the questions that should be asked are:
1. Do the data suggest the existence of a phenomenon of "socialization"

that is worth pursuing further?
2. To what extent are the results of the research consistent with previous
research in socialization and other related areas (e.g., selection, orienta-
tion, role theory)?
3. Could there be a longitudinal, empirical test of the contingency model of
socialization?
4. Does "A Contingency Theory of Socialization" clarify what we know
about the entry of individuals into organizations, or specify the condi-
tions which will make the process successful or unsuccessful for indi-
viduals?
THE ROLE OF FUTURE RESEARCH
The next issue concerns identifying the types of research that are most called for
to further elaborate, articulate, and refme midrange theories of organization
behavior. The two traditionally suggested studies for further research are:
I. Crucial tests and key experiments. If two rival theories concerning the
same phenomenon have so far stood up equally well in empirical tests,
then a decision between the two may be reached if a certain set of test
conditions can be established so that their two hypotheses yield mutually
exclusive outcomes (Hempel, 1966).
2. Confirmation by "new" test implications. When a theory is formulated,
it is generally done to explain certain already observable phenomena. It
is highly desirable for a theory to be confirmed also by "new" evidence
-by facts that were not known or not taken into account when the hy·
pothesis was formulated (Hempel, 1966).
Just as the criteria on which to judge the utility of middle range theories may
be different from those used to evaluate more global theories, so, too, are the
types of research that must follow the initial theory articulation and data collec-
tion. Some additional types of follow-up research that midrange theories might
demand are:
1. Refinement of concepts. If concepts are global (such as "realism") then

the dimensions of the concept (realism of job duties, work groups, orga-
nizations) must be specified, the relationships between those dimensions
identified, and the algorithm for combining those dimensions justified.
The reliability and validity of measurement by both questionnaire and
interview should be increased.
2. Formalization of relational statements. Midrange theorists will also have
to specify the existence (or nonexistence) of relationships between vari-
ables, to specify the direction and magnitude of those relationships, and
to specify the conditions under which those relationships occur. In "A
Contingency Theory of Socialization" an important next research step is
to state hypotheses about (and empirically test) the existence, direction,
and magnitude of the relationship between each two variables in the
socialization process. Moreover, more sophisticated hypotheses about the
conditions under which those relationships will occur (e.g., for what job
categories, for what types of individuals, in what type of training pro-
grams) should be made.
3. Elaboration of research design. The next research design under which the
middle range theory is tested should be maximally fit to the phenomenon
under investigation. In the case of the socialization theory in our exam-
ple, the next research must be longitudinal as well as cross-sectional in
nature. It is critical to determine how the socialization process operates
over time.
4. Increasing the power of statistical tests. As the concept operationalization
becomes more reliable, the relational statements more formal, and the
research design more elaborate, it should become possible to use more
sophisticated statistics to isolate both the main effects and the more sub-
tle patterns of the results. In the research on socialization, the appropriate
test of the socialization model calls for path analysis: the existence of
causal relationships can be more firmly established (or disputed); the dif-
ferences between socialization experiences in different job categories can
be more fully understood; alternative models of the socialization process
can be compared and contrasted.
CONCLUSIONS
In weighing the merits of middle range theories, two key considerations emerge.
The first is the current "state of the art" of organization behavior as a social
science endeavor. The second, and perhaps the more important issue, is how
facilitative a collective or consensus approach to research would be in extending
our knowledge in the field.
The field of organization behavior is both interdisciplinary and young. Orga-
nization behavior is an amalgamation of theoretical perspectives from other
fields. As researchers and practitioners, all of us do not share the same profes-
sional association; many of us do not value the same methodologies; some of us
disagree on what phenomena should be the focus of study; a few of us even dis-
agree on basic defmitions.
Our progress in the field is a much more interactive process of theory formu-
lation, data collection, theory refmement, and methodological improvement
than the pure deductive scientist would be comfortable with. Our hypotheses
contain theoretical terms whose instances have not as yet been observed. "The
operations and measurements that a scientist undertakes in the laboratory,"
writes Kuhn, "are not 'the given' of experiment but rather 'the collected with
difficulty.' They are not what the scientist sees -at least not before his research
is well advanced and his attention focused" (1962, p. 126). We can not pursue a
truly inductivist path either, for we can not enumerate all the instances of occur-
rence of a phenomenon under investigation and lack criteria for determining
how many instances could or should be studied to guarantee representativeness
(Feldman, 1972).
The field of organization behavior is clearly beyond the "craft" stage, where
shared theory is only adequate to establish the plaUSibility of the diScipline and
to provide a rationale for the various craft rules that govern practices, and where
individual failures are uninformative and do not reflect on the competence of
the prognosticator in the eyes of coprofessionals (Kuhn, 1962). On the other
hand, neither the state of our theorizing nor the state of our methodology sug·
gests the utility of a comprehensive theory to guide research. Given the diversity
of our assumptions about individuals, groups, and organizations and the diversity
of the phenomena under investigation, the field of organization behavior may
progress faster from the growth of middle range theories bottom-up than from
the pursuit of a more global theory top-down.
The second issue is that of how important consensus among scientists about
assumptions, methodologies, and theories is in increasing the body of knowledge
in a field. Kuhn, in The Structure of Scientific Revolutions (1962), sees research
as "a strenuous and devoted attempt to force nature into the conceptual boxes
supplied by professional education. Research could not proceed without such
boxes, whatever the element of arbitrariness in their historic origins, and occa-
sionally in their subsequent development .... Observation and experience can
and must drastically restrict the range of admissible scientific belief, else there
would be no science" (pp. 3-5, emphasis added).
Certainly, perceptually and sociologically, organization behavior scholars
must have some shared references on which to base their communications with
each other. However, collectivism can exist in scientific communities in a dif·
ferent, and perhaps a more appropriate, sense than Kuhn means (Feldman,
1972). As Joseph Agassi writes in The Journal of the History of Philosophy,
collectivism can operate, not in the sense that everyone belongs to the same
paradigm, but in that everyone agrees on what is Significant enough to deserve
critical attention (1966, pp. 351-354). Given the origins and the age of the
field, that is probably the most we can hope for in the near future. Middle
range theories might move us faster and more efficiently to that collectivism.
REFERENCES
Agassi, J. Book review of the structure of scientific revolutions. Journal of the

History of Philosophy, 1966,4, 351-354.
Blalock, H.M., Jr. Causal Inferences in Nonexperimental Research. New York:
W.W. Norton, 1964.
Caplow, T. Principles of Organization. New York: Harcourt, Brace, and World,
1964.
Farris, G.F., and Lim, F.G., Jr. Effects of performance on leadership, cohesive-
ness, influence, satisfaction, and subsequent performance. Journal of Applied
Psychology, 1969,53, 490-497.
Feldman, D.C. A contingency theory of socialization. Administrative Science
Quarterly, 1976a,21, 433-452.
--. The reluctant revolutions: An analysis of Thomas Kuhn's history and phi-
losophy of science and its application to biological and sociological theory.
Unpublished honors thesis, University of Pennsylvania, Philadelphia, 1972.
Feyerabend, P.K. Explanation, reduction, and empiricism. In H. Feigl and G.
Maxwell (Eds.), Minnesota Studies in the Philosophy of Science Vol. III.

Minneapolis: University of Minnesota Press, 1962.
- - . Consolations for the specialist. In I. Lakatos and A. Musgrave (Eds.),
Criticism and the Growth of Knowledge. Cambridge, England: Cambridge
Fleishman, E.A., Harris, E.F., and Burtt, H.E. Leadership and Supervision in
Industry. Columbus, Ohio: Ohio State University, Bureau of Educational Re-
search,1955.
Halpin, A.W., and Winer, B.J. A factorial study of the leader behavior descrip-
tions. In R.M. Stogdill and A.E. Coons (Eds.), Leader Behavior: Its Descrip-
tion and Measurement. Columbus, Ohio: Ohio State University, Bureau of
Business Research, Monograph No. 88, 1957,39-51.
Hempel, C.G. Philosophy of Natural Science. Englewood Cliffs, N.J.: Prentice-
Hall, Inc., 1966.
Kerlinger, F.N. Foundations of Behavioral Research. New York: Holt, Rinehart,
and Winston, 1964.
Kuhn, T. S. The Structure of Scientific Revolutions. Chicago: University of Chi-
cago Press, 1962.
- - . Logic of discovery or psychology of research. In I. Lakatos and A. Mus-
grave (Eds.), Criticism and the Growth of Knowledge. Cambridge, England:
Cambridge University Press, 1970.
Labovitz, S., and Hagedorn, R. Introduction to Social Research. New York:
McGraw-Hill, 1971.
Lashchyk, E.M. Scientific revolutions: A philosophical critique of the theories
of science of Thomas Kuhn and Paul Feyerabend. Unpublished Ph.D. thesis,
University of Pennsylvania, Philadelphia, 1970.
Lowin, A., and Craig, LR. The influence of level of performance on manage-
ment style: An experimental object lesson in the ambiguity of correlational
data. Organizational Behavior and Human Performance, 1968, 3, 440-458.
Merton, R.K. Social Theory and Social Structure (2nd ed.). New York: Free
Press, 1968.
Popper, K.R. Normal science and its dangers. In I. Lakatos and A. Musgrave
(Eds.), Criticism and the Growth of Knowledge. Cambridge, England: Cam-
bridge University Press, 1970.
Reynolds, P.D. A Primer in Theory Construction. Indianapolis: Bobbs-Merrill
Company, 1971.
Runkel, P.J., and McGrath, J.E. Research on Human Behavior. New York: Holt,
Rinehart, and Winston, 1972.
Van Maanen, J. Breaking in: Socialization to work. In R. Dubin (Ed.), Hand-
book of Work, Organization, and Society. Chicago: Rand-McNally, 1975.
Watkins, J.W.N. Against normal science. In I. Lakatos and A. Musgrave (Eds.),
Criticism and the Growth of Knowledge. Cambridge, England: Cambridge
19, 357-367.
27 EXAMPLES OF MIDDLE RANGE
THEORY
Discussion
Thomas A. Mahoney, University of Minnesota
The papers presented here are intended as "examples of middle range theory" of
organizations and organizational behavior. Quite different approaches were
taken in the development of the various papers, yet a discussant must search for
common dimensions on which to compare the papers. Common dimensions that
appeared in my analysis of the papers relate to issues and topics that were the
subject of earlier sessions. Repeated examination of certain of these issues is
inevitable and, at this stage, probably desirable. The dimensions that appeared
common to all five papers are outlined below as questions and provide the
framework for my comparison and evaluation of the different papers. Certain
authors addressed these questions explicitly, other authors addressed them only
inferentially. My inferences based upon these papers are open to challenge and
I'm confident the authors will exercise this right to challenge. Each of the papers
represents a condensation of some larger body of thought, and my inferences
about that larger body of thought based upon the condensed papers may well
be faulty.
1. What Is Middle Range Theory? This question, discussed formally in the

first session of the conference, probably bothered everyone writing for this con·
ference. Every author was forced to answer this question at least implicitly in
326
EXAMPLES OF MIDDLE RANGE THEORY 327
the selection and development of topics for submission to the conference; cer-
tain of the papers in this session also address the question explicitly_Merton 's
characterization of middle range theories as lying somewhere between "minor
working hypotheses" and a "master conceptual scheme" is accepted by all of
the authors as a general defmition of middle range theory. Minor working hy-
potheses are relatively easy to identify as one boundary, but, as noted by Beres
and Price, the other boundary of a master conceptual scheme is not as
easy to identify.
What constitutes a master conceptual scheme for one class of phenomena
may constitute a middle range theory for another, broader class of phenomena.
A theory of employment determination at the level of the firm appears middle
range relative to a theory of the firm, which also appears middle range to a
theory of capitalism. It would appear that the concept of middle range theory
varies with the class of phenomena of concern. It struck me that most of the
theories with which I am familiar are middle range theories, whether intended
as such or not. Further, it appears that middle range theory cannot be defined
without specification of the phenomena that are the subject of the theories,
and that characterization of middle range theories of organizations and organiza-
tional behavior might better focus upon definition of the intended domains of
these constructs than upon the concept of middle range. We probably ought to
distinguish between middle range theory and middle range phenomena, two
independent concepts, and then seek to identify what might be classed as middle
range phenomena of organizational behavior.
The papers by Feldman and by Beres and Price explicitly address the question
"What is middle range theory?", develop definitions of middle range theory, and
provide examples from what they would term middle range theories to illustrate
their points. The paper by Muchinsky and Morrow answers the question inferen-
tially as they discuss middle range theorizing relative to the concept of organiza-
tional effectiveness. The papers by Miner and by Slusher and Roering answer the
question only indirectly as they present what they consider to be specific middle
range theories.
Feldman, Beres and Price, and, less directly, Muchinsky and Morrow, indicate
that middle range theory can be characterized as follows:
• Somewhat more general than simple, relational hypotheses and somewhat

less general than overall, comprehensive conceptual schemes
• Involving constructs relatively close to observable data or experience, sug-
gesting that middle range theory tends to be inductive in nature
• Applicable within limited domains and, presumably because of the induc-
tive approach, derived from or applied to small sample research
• Focused upon process rather than structural relationships.
While these characteristics may apply to certain middle range theories, it is not
apparent that they need be exclusionary in nature. The economic theory of the
firm, for example, might be cited as an example of middle range theory in eco-
nomics. From one standpoint, this theory does present a conceptual scheme
comprehensive of allocational decisions within the firm, yet it is not sufficiently
comprehensive to explain market phenomena; in short, it addresses middle range
phenomena of economics. The constructs employed in this theory while less
abstract than the construct of general equilibrium, often are relatively abstract
and not easily induced from experience; many students would assert that the
construct of marginal revenue product is quite difficult to induce from observa-
tion. The domain within which the theory is restricted is indicated quite clearly
in the assumptions upon which the theory is based, assumptions of rationality,
competition and profit maximizing behavior, yet this domain limitation in no
way suggests limitation to small sample or case study research. Also, the theory
of the firm analyzes the structure of equilibrium conditions rather than the
process through which equilibrium is achieved. Finally, although the theory of
the firm probably was induced in part from experiential observations, it is today
stated as a deductive theory in which propositions are derived from a specific
set of assumptions. The characteristics of middle range theory inferred from
these papers may thus appear descriptive of present-day theorizing about orga-
nizational behavior, but they are not necessary or defmitional characteristics of
The papers by Miner and by Slusher and Roering present intended examples
of middle range theories, examples that can be analyzed in terms of middle range
theory characteristics specified in the other papers. Miner's presentation of
middle range theory illustrates certain of these characteristics. The concepts
specified in the paper, for example, appear to have been induced from observa-
tion of organizational roles rather than deduced from a comprehensive concep-
tual scheme of organizational energy. The domain of variables considered
also is relatively restricted, focusing upon organizational role characteristics and
inferred rewarding or reinforcing aspects of these roles; other influences of
motivation and energization are excluded from consideration. Relational hy-
potheses are not stated explicitly in the paper and can only be inferred from the
discussion. The example from Miner is more easily characterized as an elabora-
tion of constructs of middle range phenomena presumed relevant to motivation
and performance in organizational roles than as a theory of motivational pro-
cesses. The theory in which these constructs are applied may be developed else-
where in Miner's writing, but it is not stated explicitly in this condensed version.
The paper by Slusher and Roering illustrates relatively fewer of the character-
istics of middle range theory cited earlier. This paper presents a relatively ab-
stract model developed in topographical mathematics as it might apply to a
limited domain of organizational phenomena. While the intended application is

relatively specific and middle range, the model itself is intended as a general
model applicable to a wide range of phenomena. This application of a general
model to organizational phenomena is an interesting illustration of middle range
theorizing, a general model is applied to middle range phenomena. Middle range
theory, in this context, is theory directed toward middle range phenomena, not
a stage or type of theory distinguishable from general theory directed toward the
same phenomena.
The lack of consensus among these papers regarding the characteristics of
middle range theory is neither surprising nor disappointing. Too precise a defmi-
tion of middle range theory would require that we invent new labels and classes
for those theories excluded from our definition of middle range theory, theories
which are otherwise quite useful in guiding research and practice relating to orga-
nizations. The differences among the papers do suggest clearly the need to distin-
guish between middle range theory as a stage between ad hoc hypothesizing and
general theory, and as theory directed toward middle range phenomena. Middle
range theory as a stage in theory development can be understood and developed
within the context of traditional methods of science, building upon single-order
relationships among variables or elaborating a general theory for application in
a specific setting. Theory of middle range phenomena is distinctly different, how-
ever, and need not share constructs or relationships with theories of micro or
macro phenomena; the constructs relevant to explanation of behavior of the firm
(micro) may be irrelevant to explanation of behavior of an economy (macro).
Theory development for middle range phenomena thus is distinctly different
from middle range theory development for some specified phenomenon.
2. What Is the Value of Middle Range Theorizing about Organizations and

Organizational Behavior? This question is addressed most explicitly in the
paper by Muchinsky and Morrow; other papers address the question only by
implication. The paper by Muchinsky and Morrow appears to be more concerned
with the process of middle range theorizing than with any specific middle range
theory. They focus upon the concept of organizational effectiveness and argue
that an acceptable construct of organizational effectiveness is more likely to be
achieved if approached within the context of some limited domain than if ap-
proached as a construct of general applicability regardless of level or type of
organization. While sympathetic to this view, I looked in vain for suggestions for
domain limitation or phenomenon specification in the development of a con-
struct of organizational effectiveness. The problems associated with research
concerning organizational effectiveness reviewed by Muchinsky and Morrow
appear related to construct formulation in the absence of a theory. Organiza-
tional effectiveness is a construct, and the relevant attributes of that construct
would appear to vary with the theory in which it is employed. Thus, profit is
an appropriate attribute of the construct of organizational effectiveness in the
economic theory of the firm, and job satisfaction might be an appropriate attri-
bute in a theory of organizational commitment. Partitioning of the domain for
elaboration of the construct of organizational effectiveness might thus be based
upon the specific theory in which the construct is employed. I find it difficult to
elaborate a construct in the absence of a tentative theory, model, or purpose to
provide motivation for the construct. Muchinsky and Morrow appear to be
examining the desirability of formulating a general construct of organizational
effectiveness without considering the generality of the model or theory in which
the construct is to be employed. Perhaps definition of the construct ought to be
abandoned until necessary in development of a specific model or theory of
organizations and organizational behavior.
The papers by Feldman and by Beres and Price do not state explicitly but
rather imply that the value of middle range theorizing lies in its usefulness as a
strategy toward achieving general theory, that general theory is more likely to
emerge from well conceived and validated middle range theories than from
deduction from global propositions. Middle range theorizing is justified as a
strategy, not an end, of scholarship; general theory appears as the end sought.
While middle range theorizing may be a very fruitful strategy for the develop-
ment of knowledge about organizations, particularly at the current stage of
scholarship, I doubt whether the authors would argue that it is the only appro-
priate strategy, although they imply it is the predominant strategy. Improved
understanding of organizations and organizational behavior is most likely to
develop if each interested scholar applies the craft at which he or she is most
skilled, not by forcing all to apply a single craft.
The paper by Miner suggests that general theory may require additional
specification and elaboration for application in relatively specific domains, and
that these elaborations can be viewed as middle range theories just as eyert
and March's behavioral theory of the firm elaborates the economic theory of the
firm for more accurate prediction of specific decisions. One might infer from
Miner's discussion that a general proposition to the effect that individual perfor-
mance is a function of task characteristics perceived as reinforcing by the indi-
vidual underlies development of his theory, a general proposition consistent with
various theories of motivation and performance. Miner elaborates this general
proposition through specification of constructs of organizational roles and by
reinforcing role characteristics. Miner seeks elaboration of constructs from gen-
eral theory for more precise prediction within limited domains. He thus appears
to suggest that middle range theory can be derived from general theory and that
the value of middle range theory is to provide more specific and precise predic-
tion within limited domains, a rationale for middle range theorizing counter to
that offered by Feldman and by Beres and Price who appear to seek general
theory from middle range theory. Miner's paper does not provide any explicit
motivation for development of the theory presented and the specific constructs
elaborated in the theory. One can only infer Miner would argue that the specific
construct elaborations he proposes ought to be more predictive of individual
performance in organizational roles than other motivation and reward variables
such as contingency compensation, need achievement, or performance goals. He
proposes construct dimensions more specific than the general constructs of role
and reinforcement, but he does not include in this paper the motivation for
elaboration of these more specific dimensions.
Both rationales for middle range theorizing inferred from these papers appear
reasonable, the development of middle range theory as an advance over ad hoc
hypotheses and the development of middle range theory as elaboration of gen-
eral theory for more accurate prediction within limited domains. The search for
middle range theory as an advance over ad hoc hypothesizing is easily rationalized
as a process in the development of general theory, an accepted end of scientific
endeavor. The development of middle range theory as an elaboration of general
theory differs both in rationale and form. The rationale depends upon recog-
nized imprecision of predictions from general theory and the value of improved
prediction within limited domains. The rationale for theorizing about coalitional
decisions in resource allocation, for example, depends upon the degree of error
observed in predictions about resource allocation decisions derived from the
theory of the firm. Middle range theory developed as limited domain elabora-
tions of general theory also differs in form from middle range theory developed
from ad hoc hypotheses; it is concerned more with construct specification than
with the identification of relationships as illustrated in comparison of the papers
by Miner and Feldman.
The paper by Slusher and Roering considers what might be classed as middle
range phenomena, modes of interaction between organizational units, and ap-
plies a general model in analysis of those phenomena. The model applied is not
a theory. The model is quite abstract and general, specifying relationships but
lacking constructs with substantive content. They attempt to use the model in
developing a theory of middle range phenomena, and illustrate the development
of theory of middle range phenomena, not the development of middle range
theory. This paper demonstrates quite clearly that theory of middle range phe-
nomena need not be approached as the development of middle range theory
from a base of single order hypotheses.
3. Do These Papers Exemplify Middle Range Theories of Organizations or

Organizational Behavior? This question may appear to be redundant given the
comments already offered. However, it provides opportunity to express a num-
ber of reactions to the papers and to issues raised by this conference, which are
not particularly germane to the questions posed earlier. Note that there are three
aspects of this question, one relating to middle range, another to theory, and a
third relating to subjects of the papers.
Certainly all of the papers relate to something that might be termed middle
range, some more than others. This is not surprising since Beres and Price ob-
serve that what is middle range depends upon one's view of general. Most theories
are middle range in the sense that a domain more general than that addressed by
the theory can always be postulated.
Whether or not these papers present theories as such is debatable depending
upon one's terminology and approach to the philosophy of science. Certain
differences among the papers can be noted, however, without engaging in such a
debate. The paper by Muchinsky and Morrow, for example, is a discussion of
theorizing and does not present a theory of organizational effectiveness; they
urge a particular approach to development of such a theory. The papers by
Feldman and by Beres and Price do not explicitly present theories; rather, they
discuss middle range theorizing and present illustrations from middle range
theories. Citations in these papers presumably refer readers to more complete
statements of the theories. It appears from the papers that Feldman refers to
an explicit statement of a theory, whereas Beres and Price refer to constructs
in role literature that might be integrated into a theory, a theory that is not
yet explicit. Miner's paper presents a framework for construct elaboration but
does not explicate a theory. Theory, viewed as a set of constructs defined with
linkages among themselves and with rules of correspondence to the observable
world, is more comprehensive than the discussion in the paper and, presumably,
is more explicit in works cited in the paper. The paper appears to present an
elaboration of a construct of reward through specification of role characteristics
as relevant attributes of the construct, a construct that presumably achieves rele-
vance through incorporation in some theory of performance. The paper by
Slusher and Roering is at a preliminary stage of theory development. They pro-
pose a general model lacking substantive content and causal relationships for
application in analysis of specific organizational phenomena. It remains to
specify constructs within the framework of this model and to develop causal
explanations of the predicted relationships. Application of the catastrophe
model in developing a theory of modes of interaction between organizations
provides a helpful discipline to this theory development. Theory development
using a general model as framework for the theory appears to be contrary to the
order implied by Merton, an alternative and probably fruitful approach to
theory development.
Having distinguished between middle range theory and theory of middle
range phenomena earlier, let me note that the former was the subject of most
of these papers. Only the paper by Slusher and Roering and, to a lesser degree,
the paper by Muchinsky and Morrow focus upon what might be termed middle
range phenomena of organization behavior. Organization behavior, viewed as
middle range between concepts of individual behavior in organizations and be-
havior of social systems, suggests theory development with the concept of orga-
nization as the unit of analysis. Most of these papers address middle range
theorizing about individual behavior in organizations rather than theory develop-
ment about the behavior of organizations as middle range phenomena. It is by
no means apparent that the development of theory of behavior of organizations
will proceed best if approached as aggregation of or abstraction from theories
about individual behavior in organizations. In fact, examples from other sciences
suggest that variables and models appropriate for analysis at one level do not
easily aggregate for analysis at another level. New models and variables appro-
priate to that level are required.
Finally, I commend the authors of these papers. Examination of their papers
has helped me to understand better the need for theory development at the level
of organizational analysis. The analysis stimulated by these papers also suggests
to me that theory development for organizations as middle range phenomena
might well proceed independently of micro theories of individual behavior in
organizations and macro theories of the behavior of social systems. I propose the
topic "theories of organizations as middle range phenomena" as the natural sub-
ject for a second annual conference at the University of British Columbia.
28DOMAIN
A RATIONALE FOR THE LIMITED
APPROACH TO THE STUDY
OF MOTIVATION
A reading of Tom Mahoney's discussant comments suggests that he is not really

sure what I am trying to do with my four limited domain theories of organiza·
tional energy and that he has serious doubts that I am engaged in theory con·
struction in any event. The first problem is the easiest to handle. Although my
paper for the Midrange Theory Conference does not include a statement of
rationale, this has been done elsewhere as Mahoney correctly surmises. One ex-
pression of what I am attempting to do is the following:
Our concern in the field of management is not with the totality of human
motivation, but with the special case of the relationship between motivation
and various job requirements. It seems quite possible that an approach that
yields a reasonable approximation to a solution of the job or employee
motivation problem might be developed long before the total complexity of
human motivation comes to be understood.
We might start with various job requirements (role prescriptions), with
either those that are formally stated, or perhaps more fruitfully with those of
an informal nature. Then job-specific motivation theories might be developed
to indicate what kinds of motivational patterns make a difference in that par-
ticular type of job. There would be little concern with specific motives; much
more concern with the sum total of motivational inputs that make for satisfy-
ing a particular job requirement .... It would be assumed that a great variety
334
A RATIONALE FOR THE LIMITED DOMAIN APPROACH 335
of different combinations of more basic motives would produce much the

same job behaviors in different people. Ideally, we could spell out all of these
patterns, but for the moment it may be more fruitful to bypass this problem
and look only at the broader question of motivation to meet various role
requirements (whatever the specific motives involved) .
. . . the study of motivation in its occupational and organizational context
need not be the same as the study of human motivation generally. The former
is a somewhat more limited problem, and we should take advantage of the
fact that a theory of employee motivation can be less complex and less gen-
eral than a total theory of psychological motivation. Past efforts have tended
to try to apply drive theory, expectancy theory, and balance or homeostatic
theories from general psychology to management problems. Perhaps this is
the long way around. Useful results for management may be more likely to
emerge from applying shortcuts that apply to the specific problems that
management seeks to solve (Miner, 1974, pp. 575,577-578).
Further elaborations of this strategy have been set forth in other publications:
An alternative approach would be to shift the research emphasis away from
the specific needs and motives of most content theories to the motivational
requirements of tasks and to informal and formal job role requirements. Thus
the major concern would be with the sum total of motivational inputs needed
to satisfy job requirements .... If this approach can be extended further to
cover a whole network of key jobs in varying organizational contexts, the
possibility exists that a base for meaningful organizational change can be de-
veloped long before the total complexity of specific human motives is under-
stood. What may well emerge is a highly parsimonious but comprehensive
theory which takes job requirements rather than human motives as its starting
point (Miner and Dachler, 1973, p. 397).
The main virtue of limited domain theories is that they simplify the theo-
retical task to a point where it is amenable to precise conceptualization and
intensive research investigation. Key variables delimiting the domain, which
may later be expanded into true contingency variables, can be clearly iden-
tified. Over time as various domains are explored the matrix of possible do-
mains is gradually filled in, until a comprehensive contingency theory emerges.
Ideally the domains explored initially will be those where the prospect of
establishing valid theories is greatest and where the most important practical
contribution can be made. One definition of the latter would be a domain in
which the greatest number of organizations fall for the longest periods of
time (Steiner and Miner, 1977, p. 313).
Additional discussions of this approach or strategy for theory construction

are contained in Miner (1977) and MacKenzie and House (1978). The key aspect
is that many of the variables and relationships that have characterized other
theories to understand and predict the same or similar phenomena are simply
bypassed; as knowledge develops they may well be filled in later, but for many
current purposes they are not needed.
Whether the result of this strategy as presented in my paper constitutes a true
theory is a legitimate matter for speculation. Certainly it serves in the role of a
theory. The situation is much like that with regard to goal setting theory (Locke,
1975). My own view, elaborated elsewhere, is that such efforts are best treated
as theories, with full acknowledgement that they are markedly, and often inten-
tionally, incomplete (Miner, 1980).
REFERENCES
Locke, E.A. Personnel attitudes and motivation. Annual Review of Psychology,

1975,26,457-480.
MacKenzie, K.D., and House, R.I. Paradigm development in the social sciences:
A proposed research strategy. Academy of Management Review, 1978, 3,
7-23.
Miner, J.B. The organization for motivation. In J.W. McGuire (Ed.), Contem-
porary Management: Issues and Viewpoints. Englewood Cliffs, N.J.: Prentice-
Hall,1974.
Miner, J.B. Motivation to Manage: A Ten Year Update on the "Studies in
Management Education" Research. Atlanta, Ga.: Organizational Measure-
ment Systems Press, 1977.
Miner, J. B. Theories of Organizational Behavior. Chicago: Dryden Press, 1980.
Miner, J.B., and Dachler, H.P. Personnel attitudes and motivation. Annual Re-
view of Psychology, 1973,24, 379-402.
Steiner, G.A., and Miner, J.B. Management Policy and Strategy. New York:
Macmillan, 1977.
V COUNTERPOINTS
AND ALTERNATIVES
29 THEORY DEVELOPMENT
IN ORGANIZATION BEHAVIOR
A Systems' Perspective
David Cooper, University of British Columbia
and University of Manchester;
and Frank Wolf, University of British Columbia
There is considerable agreement that the open systems model has dominated
thinking and research in the field of organization theory over the last decade.
Recently, however, there has been a call for a paradigm change in organization
theory and proponents of such a change have referred to the possibility of a
paradigmatic revolution in the near future (Scott, 1974). The growing discontent
with the state of organizational theory has been largely responsible for the
renewed interest in middle range theorizing (pinder and Moore, 1977; Weick,
1974). However, the reasons for such discontent have not been subject to
unambiguous interpretation.
This paper will focus upon the attributes of a systems methodology! and will
indicate that recent contributions, which have seemingly called for something
new, have pleaded for little more than the exploitation of systemic notions in
the science of organization. Much of the recent concern with a systems orienta-
tion will be shown to have arisen largely because of our own failure to adequately
exploit the implications of a systems methodology in organizational behavior.
In this paper we explore the potentialities of systems thinking in the science
of organizations. In the next section we will consider the idea of a revolution in
the study of organizational behavior and will clarify the Kuhnian notion of a
revolutionary change. The applicability of the revolutionary metaphor to the
339
current state of organizational theory will be evaluated. Drawing on recent de-

velopments by Sneed (1971) and Stegmilller (1976) we will show that middle
range theorizing can be explained in terms of the existing paradigm and that
such theorizing represents nothing more than one set of potential models within
the systems framework.
The potentialities of a systems methodology will then be elaborated and the
use of systems analysis in organizational behavior will be exemplified. The
importance of systems thinking to middle range theorizing will be illustrated by
considering the impact of a systems methodology to taxonomic development.
Finally, we will summarize the arguments developed in this analysis and will
examine the implications of a paradigmatic change when the existing paradigm
is not yet fully exploited. A plea for the reevaluation of existing methodologies
will be advanced.
THE NOTION OF A REVOLUTION IN ORGANIZATION THEORY
The notion of a revolution in organization theory is taken from the work of

Kuhn (1970) who contends that science does not progress through the accumu-
lation of theories. Instead, a series of tradition-shattering revolutions occurs in
which one time-honored scientific theory is rejected in favor of another that is
incompatible with it.
The new theory is unique and is seldom just an increment to what is already
known. In moving from the old set of ideas to the new, the community of scien-
tists follows a number of identifiable stages: First, there is a recognition of
anomalies, then a period of insecurity, which in turn is followed by the develop-
ment of alternate sets of ideas. At the fourth stage there is an identification of
schools of thought, and the fmal stage in the Kuhnian revolution is the domina-
tion of new practices or ideas.
The first step is a precursor to the entire process. As anomalies amount, a
crisis stage is reached during which scientists become increasingly dissatisfied
with the existing theoretical framework. Many of the deviations from the theory
are handled in an ad hoc way but a search for alternatives also commences,
schools of thought emerge, and one set of ideas gradually gains ascendency over
the alternatives.
The stages in the revolutionary process are both mutually interactive and
cyclical. Once a new theory has been accepted the stage is set for the cycle to
repeat itself.
While Kuhn has been criticized for the looseness of his terminology (Shapere,
1964; Lakatos and Musgrave, 1970), particularly in the earlier edition of his
THEORY DEVELOPMENT IN ORGANIZATION BEHAVIOR 341
work, it is clear that he did not carefully distinguish between theories and para-
digms in his description of the stages of the revolutionary process.
Ritzer (1975), drawing upon the ideas developed by Kuhn, suggests that
a paradigm is a fundamental image of the subject matter within a science. It
serves to define what should be studied, what questions should be asked, how
they should be asked, and what rules should be followed in interpreting the
answer obtained. The paradigm is the broadest unit of consensus within a
science and serves to differentiate one scientific community from another
(Ritzer, 1975, p. 157).
The paradigm change, according to Kuhn, is unlikely to be completed quickly
and the new paradigm will only be accepted if it is able to resolve some out-
standing and generally recognizable problems that cannot be met in some other
way. Furthermore, a paradigm change will only be successful if the new para-
digm is at least potentially able to preserve a large part of the concrete problem-
solving ability that has accrued to science through its predecessors.
Recently, Sneed (1971) and subsequently Stegmuller (1976) and Mattessich
(1978) have provided a formalization of Kuhn's theory of paradigm revolution,
which provides a framework within which recent developments in organizational
theory can be constructively viewed. It is argued that a theory consists of a
fairly stable core and, in addition, a set of intended applications. Thus, a theory
can be represented by the following equation:
T= f(K,l)
where: T= theory
K = the core of a theory
I = the set of intended applications
Sneed and Stegmtiller specify the core elements in some detail. They indicate
that the core consists of a set of models (m), possible models (mp), and possible
partial models (mpp). Within the core there are also constraints (c) and a number
of variables (v). Thus, the core of the theory can be described functionally as:
K = gem, mp, mpp, c, v)
and hence, a theory may be functionally described as follows:
T = f(m, mp, mpp, c, v, l)
According to Sneed and Stegmtiller, normal science (the process of accumula-
tion) consists of the elaboration of the core. Core elaboration takes place as
more variables and constraints are added and as model development ensues.
Hence, partial models and partial partial models may be added to the core
dimension. In a Kuhnian sense, revolution occurs only when the core can no
longer be extended and is overthrown. However, theory development can pro-
ceed by altering the models, the constraints, the variables, or the set of intended
applications of the theory. In the ensuing analysis it will be demonstrated that
calls for a paradigm change in organizational theory are merely calls for core
elaboration-no paradigm upheaval is involved.
Nevertheless, while calls for a paradigm change in organizational behavior
may be misplaced, the recent concern about the usefulness of the systems para-
digm has been important. First, it has drawn attention to the need to defme just
what the core of organizational theory actually is. In so doing, it has placed
demands on the researcher to specify the nature and extent of the variables and
constraints underlying the core construct. In organizational behavior science
careful elaboration and specification of the core elements seems to have been
lacking. We have spoken loosely of organizations as systems without defining the
parameters implicit in such an observation. A fundamental concern of this
paper is a call for a more careful and precise specific language in organizational
behavior science.
Second, dispute over the usefulness of the systems paradigm has fulfilled the
function of drawing attention to the need to specify the set of intended appli-
cations of organization theory. A theory relates to a specific domain of reality:
organization theory does not seem to have clearly specified its domain.
Associated with, but nevertheless distinct from, the need to specify the in-
tended applications of organization theory has been a need to specify the basis
on which we may judge the success of the science of organization. To argue that
the open systems approach has been unsuccessful necessarily presupposes criteria
of success. It is not clear that discontent amongst scientists is, of itself, evidence
of the failure of systems thinking in organization theory. The discontent may be
a result of the current organization of science that encourages artificial disputes
and "innovation." In a system where promotion and recognition seem to be
related to novelty and theory "development" (rather than elaboration or rep-
lication of tests of the theory) the existence of disputes and "innovation" is
hardly surprising.
Popper (1959) proposes a set of criteria for the evaluation of science:
scientific theories are universal statements ... theories are nets cast to catch
what we call the 'world': to rationalize, to explain and to master it (p. 59).
The science of organization is also concerned with the explanation, predic-
tion, and mastery of organizational behavior. 2 As a positive science, explanation
and prediction may be sufficient. But, if organization theory is intended to
enable the effective management of organizations (i.e., to act as a normative or
instrumental science) then it requires a notion of "good" and "bad" organiza-

tional behavior. However, such issues have not been addressed in the literature.
We are unaware of research that, one way or the other, considers the value of
the system paradigm of organization theory in terms of a well defmed set of cri-
teria. It would appear though, that systems thinking has high potential in that it
focuses on purpose and aims to utilize insights from other sciences. The empha-
sis on purpose encourages concern with the explanation of the continued ex-
istence of a contrived phenomenon, like organizations. Systems thinking also
encourages the use of insights gained about reality in other sciences. We are
not advocating theorizing by analogy, but rather the sensitive use of the insights
from all branches of science in the study of organizational behavior. The model
for organization theory may be just "organizations" but it is possible that there
is regularity in nature and in human contrivances; surely a fundamental assump-
tion of science is a notion of pattern and order across phenomena (Robinson,
1947), and the existence of Gestalten properties that are common to the differ-
ent levels of organization of living matter (Emery, 1969).
THE SYSTEMS PARADIGM
The advantages of the systems approach have been emphasized in the literature
on organization theory (most notably by Katz and Kahn, 1966 and Thompson,
1967). Even a brief review of the literature in organization behavior or theory
indicates the pervasiveness of the systems paradigm. For example, Pondy and
Mitroff (1978) indicate that thirty-five percent of the articles in a recent one-
year period in Administrative Science Quarterly refer to Thompson's work. Yet,
although lip service has been paid to the concepts of systems thinking, it does
not seem that they have been carefully utilized or developed in organizational
theorizing. In this section we will briefly indicate the basic features of any sys-
tem and illustrate that systems thinking has not been fully utilized in organiza-
tion theory. We show that while many of the current interests in organization
theory are entirely consistent with systems thinking, the use of the systems
approach has not been rigorously applied in the study of organizations. The sys-
tems paradigm has not been fully exploited.
Boulding (1956) has provided us with a useful elaboration of a general systems
theory. He identifies nine levels of analysis. In his hierarchy each level of analysis
contains a111evels below it and adds a new dimension that was not included in
the lower levels. Accordingly, the higher the level, the greater the complexity
and generality of the analysis. Pondy and Mitroff (1978, p. 12), using Boulding's
classification of systems, suggest that "our formal models and data collection
efforts are rooted at levels 1 and 2." These lowest levels are concerned with
structural attributes of systems, whereas the level of social organization must be
analyzed at a higher level of complexity.
This criticism can be leveled at Thompson's rather engineering-oriented view
of organizations (see the critiques by Silverman, 1971 and Pondy and Mitroff,
1978) yet Thompson's definition of an organization as an open system indicates
the major concepts we wish to emphasize and develop:
The complex organization is a set of interdependent parts which together
make up the whole because each contributes something and receives some-
thing from the whole, which in turn is interdependent with the larger environ-
ment. Survival of the system is taken to be the goal, and the parts and their
relationships presumably are determined through evolutionary processes
(Thompson, 1967).
In short, Thompson draws attention to the basic features of a system: struc-
ture, environment and function (or purpose).
Structure
"The structure of a system is the arrangement of its subsystems and components

in three-dimensional space at a given moment in time. This always changes over
time" (Miller, 1971, p. 294). In any definition of a specific system it is necessary
to indicate the components of the system, the relationship of the components,
and the "space" in which the arrangement exists. The components of the system
may be concrete, observable, and measurable. The components may, however,
be abstract-a construction of the observer. "It may well be that the systems
detected in the world outside ourselves are models -mapping of our own brains
onto the world" (Beer, 1966, p. 243). These abstract constructions may be
tested by checking the predictive capability of the insight. Organization theory
can be scientific even if it utilizes abstract components (e.g., role, power, values,
profits) since predictions from the observer's abstract constructs may be tested.
The structure of a system includes not only the components but also the
relationships between the components. Relationships between components may
be strong or weak, and may be spatial, temporal and/or causal (Miller, 1971).
Interdependence is generally assumed; however, the important issue is not only
the recognition of interdependence but also the specification of the nature and
strength of the relationship. An extremely weak relationship-sometimes re-
ferred to as loose coupling-is one possible value in the range of values indicating
the strength of relationships.
We would therefore suggest that strength of couplings is seen as one major
variable in a general theory of organization. The nature, strength, and time

dimension of any relationship must be indicated in any careful analysis of a
system. For a general theory of organization, consideration needs to be given
both to relationships and their strength and to the way these relationships
change over time. The partiality of the traditional, structurally oriented and
static models of organization has been demonstrated by the study of educational,
artistic, and innovative organizations (Hedberg et al., 1976; March and Ohlsen,
1976; Weick, 1976). Understanding the behavior of such organic organizations
involves the recognition of the problematic nature of the relationship between
components. Thus, the recent models of organic or loosely coupled organiza-
tions are also partial models within the core of organization theory and are
entirely consistent with the systems paradigm. The interest in loosely coupled
organizations-organized anarchies-is a necessary step in the construction of a
general theory of organization; that necessarily is apparent when a systems
framework is adopted.
Environment
Concern with the environment of a system is fundamental to systems thinking.

Variables that are uncontrollable (or weakly controllable) by the system, yet
affect its performance or behavior, are components in the substantial environ-
ment or space of a system. Systems researchers are careful to specify not only
the physical dimensions of space but also the time dimension (Miller, 1971).
Thus, concern with the process of organizing is consistent with systems thinking
(Weick, 1969). The idea of specification of time also opens up the consideration
of equilibrating mechanisms. Natural systems seem to be homeostatic: There
appears to be order in nature (although natural failures do occur). Organizations,
in part because they are contrived, may have a greater tendency towards failure:
Organizations construct mechanisms to combat the tendency not to be self-
regulating. There are organizational "failures," and we can undoubtedly learn
about viable organizations from these cases of failure. Yet, given our concern
for mastery of organizations, it seems appropriate to recognize the homeostatic
and adaptive nature of organizations. Systems may regulate themselves, but they
may also change their structure and their environment in the pursuit of survival
or viability (Buckley, 1968). Organizations are likely to be complex adaptive
systems. There is a large variety in their potential states and they seek stability
in these states (Ashby, 1956, Chapter 5).
Beer (1966) and Weick (1977) have effectively drawn attention to the need
to explicitly recognize the observer, and his or her perceptions, in any descrip-
tion of the organization and its environment. The enactment process that Weick
(1977) so graphically discusses also emphasizes that the domain of a system is

more pervasive than the characterization of system boundaries in a purely physi-
cal sense. The distinction between an organization and its environment has to be
depicted in terms of causal relationships rather than physical structures.
Note also that the boundaries of a system need not coincide with physical
boundaries. The boundaries of any system are essentially arbitrary: A system is
in the eye of the beholder, so to speak (Simon, 1962, p. 468; Weick, 1969). One
can view any branch of an organization's hierarchy as a system itself. Similarly,
the lowest level terminal point in a system's hierarchy is also essentially arbitrary.
One could trace the component of an organization down to groups, individuals,
or even atomic structures. At some point, the component must-for the sake of
parsimony-be treated as a black box with known or assumed transformations
between inputs and outputs but with unknown or unconsidered internal struc-
ture (Ashby, 1956, pp. 86-117). The study of organizations seems to have been
dominated by the view that the boundaries of an organization are physical: the
organization of a factory, for example, is bounded by the factory perimeter. The
definition of the boundary of a system needs to be given careful consideration in
organization theory, and explicit recognition must be made of the analyst (or
theorist) who chooses the specific defmition.
The analysis of organizational environments may also be enriched by con-
sidering two of their characteristics that are rarely given extensive attention in
organization theory. First, the causal texture of organizational environments
implies that the variables in an environment will also be interdependent (Emery
and Trist, 1965). As Klir and Valach (1967) indicate, the relevant or substantial
environment of a system incorporates those potentially interdependent variables
that at some point of time may affect the system.
Second, variety in an environment and in the system is necessary for the sys-
tem to adapt and change over time. As we have previously indicated, analysis of
a system necessitates the analysis of the environment of the system. Not only
the defmition of the boundary of the system, but also the distinction between
the system and its environment are arbitrary analytical necessities. We cannot
explain or predict a system's behavior if we do not "close our analysis" to
include all relevant variables (both spatial and temporal) and exclude all other
general environmental variables that are not expected to affect the system's
behavior. At one point in time, we may identify the causal relationship between
a system and its environment as flowing from the environment to the system.
Yet, the system may allocate resources to the environment ("enact" its environ-
ment) to maintain organizationally desirable environmental conditions or to
change organizationally undesirable environmental conditions for future time
periods. This phenomenon is rarely considered by organization theory. Organiza-
tions engage in social and political activity to maintain or alter the environment.
Contributions to political parties, support for disequilibrating groups in coun-

tries where "hostile" regimes exist, and political lobbying over such diverse
issues as education, health, welfare, and foreign policy are all examples of orga-
nizations enacting their environments (Child, 1972). It is rare for organization
theorists to acknowledge the political character of organization activity and
incorporate that character into their theories.
Function
One aspect of systems that does not seem to have deeply permeated into orga-
nization theory is the notion of function or purpose. Beer (1964) and Churchman
(1968) both emphasize their concern for function. To paraphrase Churchman
only slightly "the way to describe an organization is first by thinking about what
it is for, about its function, and not the list of items that make up its structure"
(Churchman, 1968, p. 13). Purposive behavior, or teleology, is fundamental to
systems thinking-a system must be defined in terms of its purpose. This pur-
pose cannot be identified by asking the components of the system for their
view, but rather by considering the environment or metasystem and hence,
specifying the reasons why the environment sustains a system. Organization
theory has clearly been concerned with objectives, but little emphasis has been
placed on environmental demands (see Dill, 1958; Thompson and McEwen,
1958). Instead of asking what an organization is, systems thinking focuses on
what an organization does (Ashby, 1956).
The focus on the function of an organization in systems thinking does indeed
form the basis of Katz and Kahn's (1966) classification of organization. More
recently, Carroll and Tosi (1977) similarly use function as a way of identifying
subsystems within an organization -viz production, adaptive, boundary spanning,
maintenance, and managerial subsystems. Yet, the discussion of function has
little influence on the majority of the chapters of these two books. Issues of con-
flict, change, decision making, and policy are discussed without a functionalist
perspective. Perhaps this is because the authors acknowledge the criticisms of
Parsonian functionalism-its overemphasis on stability and maintenance of the
status quo. "Organization theory has been a theory of order" (pondy and
Mitroff,1978).
Thompson in his above cited definition of a complex organization emphasizes
corrective devices, but a more general notion of function in systems thinking
embraces change: "As a fundamental principle, it can be stated that a condition
for the maintenance of a viable adaptive system may be a change in its particular
structure. Both stability and change are a function of the same set of variables,
which must include both the internal state of the system and the state of its
significant environment, along with the nature of the interchange between the
two" (Buckley, 1968, p. 510). It is important to indicate the feedback loops in
any "correction" device, and to indicate the time horizon of the loop. Disequi-
librium in a system may indeed occur at one point in time; systems analysis
suggests that organizations will attempt to move over time to a stable state. The
fact that organizations fail indicates that stability is not "natural" in organiza-
tions; organization theory should specify the mechanisms of adaptation.
These examples hopefully indicate the superficial treatment of systems in
organizational theorizing. We have not been exhaustive in our discussion (e.g.,
Pondy and Mitroff, 1978 place great emphasis on language in systems analysis),
yet this section has indicated both the richness of the paradigm and its ability to
raise a set of questions about any organizational analysis. Elaboration of the
paradigm is still necessary in the application of system thinking to organiza-
tional theorizing.3
Organization theorists, while they have seen themselves as working within a
systems framework, do not seem to have carefully utilized or applied that
framework. Our call is for careful elaboration of the systems paradigm, the core
of organization theory. So our concern to place many of the insights from the
science of organization into a systems frameworl is not merely a need by the
faithful to see the world through the eyes of a favored paradigm. As Pioggi
states, "A way of seeing is also a way of not seeing" (1965). Instead, systems
thinking offers the science of organization a useful way of seeing that empha-
sizes multiple relationships, coherence, and purpose. Multiple relationships are
emphasized in the interdependencies of structural components in producing
an organizational outcome. Coherence is emphasized by the similarities of con-
cern within science in general and the insight offered to any particular science
by other sciences. Coherence is also recognized in the set of questions asked of
a theory or a test thereof; what is the structure, environment, and purpose of the
organization that is the object of theory or investigation? Indeed, has the re-
search identified the systemic features? And fmally, systems thinking emphasizes
purpose. What is the purpose of the theory and by what set of criteria might it
be regarded as important or significant? Any system is a subset of a metasystem,
and it is the metasystem that provides the metaphysical justification for the
science and its theories.
SYSTEMS THINKING AND TAXONOMIC THEORIZING
In the preceding sections of this analysis the potentiality of systems thinking has
been alluded to and the usefulness of a systems methodology for organizational
behavior science has been described. We now illustrate further the richness of the
systems approach and at the same time provide a concrete illustration of the
Sneed-StegmUller formalization of the nature of theory by considering the appli-
cation of taxonomic developments in organizational behavior science.
In this section of our analysis we will show that middle range theorizing is
consistent with a systems methodology and need not imply the abandoment of
our search for a general theory. Indeed, failure to have a general theory that can
be used for the purpose of guiding the development of middle range theories
may lead to a morphological nightmare whence no prescriptions about organiza-
tions will be realizable.
McKelvey (1977) has carefully drawn a distinction between organizational
taxonomy and organizational typology and his distinction will be adhered to in
this analysis. Taxonomy is used to refer to the theory and method of classifica-
tion, while typology is used to refer to classification schema that are founded on
heuristic criteria. Taxonomic developments are neither new in the area of orga-
nizational behavior nor for that matter to science in general.
Taxonomic developments in organizational behavior have been advanced by
Etzioni (1961), Blau and Scott (1962), Thompson (1967) and others. More
recently, McKelvey (1975, 1977) and Pinder and Moore (1978) have called for a
reexamination of taxonomizing as a means by which theory development in
organizational behavior could be advanced. In particular, it has been argued that
taxonomy development may provide a mechanism by which middle range
theories can be formulated (Pinder and Moore, 1978).
However, taxonomies in the area of organizational behavior science can be
created on a multitude of dimensions since in organizational theory there exists
no commonly held view of what the basic unit of organizational structure is, and
hence the many variables and relationships with which organizational theory is
concerned lack integration (Miller, 1972, pp. 7-14). Clearly, for theory develop-
ment to proceed efficiently a meta-theory is needed to guide taxonomic formu-
lation. Failure to proceed in such a way can only lead to confusion. The need
for a general meta-theory to guide taxonomizing was well argued two hundred
and fifty years ago by Linnaeus (1737):
All the real knowledge which we possess, depends on methods by which we
distinguish the similar from the dissimilar ... The more numerous the objects
which employ our attention the more difficult it becomes to form such a
method and the more necessary.
For we must not join in the same genus the horse and the swine, tho' both
species had one hoof'd nor separate in different genera the goat, the reindeer
and the elk, tho' they differ in the form of their horns. We ought therefore by
attentive and diligent observation to determine the limits of the genera, since
they cannot be determined a priori. This is the great work, the important
labour, for should the genera be confused, all would be confusion.
A systems methodology can provide a unifying conceptual framework for orga-

nizational theory and a description of the concepts that differentiate organiza-
tions from other living systems. Furthermore, systems thinking can be used to
classify organizations on the basis of structural, functional, environmental,
and/or process-related characteristics in an integrated and consistent way.
Proposals for developing taxonomies in the area of organizational behavior
have suggested alternative criteria for such theorizing. Unless our taxonomic
developments are generally perceived within a holistic model of organizations,
we have no direct criteria for selecting between differing taxonomic formula-
tions. Several dimensions may simultaneously satisfy the criteria considered by
Hempel (1965), McKelvey (1975), and others as a means for selecting the
taxonomizing base. If we wish theorizing to advance in an internally consistent
way, it is essential that taxonomic development proceed within the context of
a systems methodology.
It is possible to illustrate that taxonomic theorizing represents nothing more
than an extension of the existing paradigm in which the set of intended applica-
tions of our existing theory is expanded. Recall that a theory is functionally
related to its core elements (K) and the set of intended applications (1). By
considering taxonomic structures of organizations, we are merely explicating
the set of intended applications of the theory. Taxonomizing provides an illus-
tration of scientific accumulation-accumulation within the bounds of our
existing paradigm. No paradigm upheaval is involved. More importantly, how-
ever, by focusing on the constructs developed within the paradigm to date,
taxonomizing can address unresolved but nevertheless recognized problems. In
that way we can be more certain that researchers are not recognizing problems
merely because they have the requisite skills to resolve them.
Middle range theorizing within the systems framework calls for the reevalua-
tion of the direction and the strength of the relationships between the parameters
of the model. The question as to which variables to manipulate and which
parameters to alter can then be addressed within a holistic frame of reference.
A PARADIGM CHANGE? SOME CONCLUSIONS
Recent discontent in the literature has led some to call for a paradigm change in
organization behavior science. It has been our intent to illustrate that such calls
are misplaced at the current time. Calls for a paradigm change are nothing more
than calls for the exploration of the core of our existing paradigm. If calls for a
paradigm change are taken seriously, we may well discard the accumulative gains
we have reaped to date. Further, since a paradigmatic change is revolutionary
and painful, its occurrence at a time when the core of our existing paradigm has
not been fully elaborated may lead to turmoil and confusion as students in the
area do not know where to turn.
Certainly, the development and elaboration of systems thinking within orga-
nizational behavior has not been easy. Further elaboration of the constructs
in terms of the set of intended applications, the set of models, variables, and
constraints may at this stage seem a forbiddingly complex task. Yet this gives
us little excuse for not confronting this task.
We can conclude no better than by considering and acting upon the following
observation:
The concepts presented ... were not selected to be admired. Rather, they are
meant to be used in connection with the stubborn phenomenon of human be-
havior in organizations . . . They provide some answers but more guidance as
to what questions it might be well to ask next to gain the understanding that
is necessary for the effective resolution of particular situations. When they do
not provide such understanding, the serious student of phenomena in orga-
nization will do well not to cast these concepts aside quickly. Rather, he may
still find them useful to throw into relief the aspects of the situation he fails
to understand. Then new questions for inquiring may suggest themselves to
him. For chance still favors the prepared mind." (Lombard's foreword to
Seiler, 1967, p. viii).
NOTES
1. In this paper we use the terms systems "methodology," "thinking," and "analy-
sis" synonymously. We prefer to avoid using the term "theory" since this suggests
that the approach we adopt is refutable. Instead, we regard the systems' approach as an
insightful way of viewing the world.
2. Some readers may be disturbed by the notion of "mastery" -alternatively influence,
control, or management. Yet, in an instrumental or applied science (such as organization
theory) we would argue that understanding is required so that it may be used-hopefully
to improve the "performance" (measured across all dimensions of behavior) of organizations.
3. Notice that Pondy and Mitroff's critique (1978) of open system models works within
a general systems framework and that their "sketch of possible futures" is a call for analysis
consistent with Boulding's higher levels of systems analysis.
REFERENCES
Ashby, W.R. An Introduction to Cybernetics. London: Chapman and Hall,

1956.
Beer, S. Cybernetics and Management. New York: Wiley, 1964.
Beer, S. Decision and Control. Chichester, Eng.: Wiley, 1966.
Blau, P., and Scott, W. Formal Organizations. San Francisco: Chandler, 1962.
Boulding, K. General systems theory - The skeleton of science. Management
Science, 1956, vol. 2, 197-208.
Buckley, W. Society as a complex adaptive system. In W. Buckley (Ed.), Modern
Systems Research for the Behavioral Scientist. Chicago: Aldine, 1968.
Carroll, S., and Tosi, H. Organizational Behavior. Chicago: St. Clair Press, 1977.
Child, J. Organizational structure, environment and performance: The role of
strategic choice. Sociology, January 1972, vol. 6, no. 1, 1-22.
Churchman, C.W. The Systems Approach. New York: Dell Publishing, 1968.
Dill, W.R. Environment as an influence on managerial autonomy. Administrative
Science Quarterly, 1958, vol. 2,409-443.
Emery, F. Systems Thinking. Harmondsworth, Middlesex: Penguin, 1969.
Emery, F., and Trist, E. The causal texture of organizational environments.
Human Relations, 1965, vol. 18,21-32.
Etzioni, A. A Comparative Analysis of Complex Organizations. New York: Free
Press, 1961.
Hedberg, B., Nystrom, P., and Starbuck, W. Camping on seesaws: Prescriptions
for a self-designing organization. Administrative Science Quarterly, 1976,
vol. 21,41-65.
Hempel, C. Aspects of Scientific Explanation. New York: Free Press, 1965.
Katz, D., and Kahn, R. The Social Psychology of Organizations. New York:
Wiley, 1966.
Klir, G., and Valach, M. Cybernetic Modelling. London: Iliffe Books, 1967.
Kuhn, T. The Structure of Scientific Revolutions (2nd enlarged edition). Chi-
cago : University of Chicago Press, 1970.
Lakatos,I., and Musgrave, A. (Eds.). Criticism and the Growth of Knowledge.
Cambridge: Cambridge University Press, 1970.
Linnaeus. Genera Plantarum. 1737 as cited in B. Everitt, Cluster Analysis. Lon-
don: Heinemann, 1974.
McKelvey, W. Guidelines for the empirical classification of organizations. Ad-
ministrative Science Quarterly, 1975, vol. 20, 509-525.
- - . Organizational Taxonomy: Lessons from Biology. Unpublished paper, Uni-
versity of California (Los Angeles), 1977.
March, J.G. and Ohlsen, J. Ambiguity and Choice in Organizations. Oslo: Uni-
versitetsforlaget, 1976.
Mattessich, R. Instrumental Reasoning and Systems Methodology. Dordrecht:
Reidel, 1978.
Miller, J.G. The nature of living systems. Behavioral Science, 1971, vol. 16, 277-
301.
- - . Living systems: The organization. Behavioral Science. 1972, vol. 17, 1-
182.
Pinder, C., and Moore, L. The Inevitability of Multiple Paradigms of Organiza-
tional Behavior. Faculty of Commerce Working Paper 513, University of
British Columbia, 1977.
- - . The Resurrection of Taxonomy for the Development of Middle Range

Theories of Organizational Behavior. Unpublished paper, University of British
Columbia, 1978.
Pioggi, G. A main theme of contemporary sociological analysis: Its achievements
and limitations. British Journal of Sociology, 1965, vol. 16,263-294.
Pondy, L. and Mitroff, I. Beyond open system models of organization. Paper to
appear in B. Staw and L.L. Cummings (Eds.), Research in Organizational Be-
havior. Greenwich, Conn.: JAI Press, 1978.
Popper, K. The Logic of Scientific Discovery. London: Hutchinson, 1959.
Ritzer, G. Sociology: A multiple paradigm science. American Sociologist, 1975,
vol. 10, 156-167.
Robinson, D. The Principles of Reasoning. New York: Appleton Century Crofts,
1947.
Journal, 1974, vol. 17, 242-254.
Seiler, J.A. Systems Analysis in Organizational Behavior. Homewood, Ill.: R.D.
Irwin, 1967.
Shapere, D. The structure of scientific revolutions. Philosophical Review, July
1964,374-383.
Silverman, D. The Theory of Organizations. London: Heineman, 1971.
Simon, H.A. The architecture of complexity. Proceedings of the American
Philosophical Society. December 1962, vol. 106, no. 6,467-482.
Sneed, J.D. The Logical Structure of Mathematical Physics. Dordrecht: Reidel,
1971.
Stegmiiller, W. The Structure and Dynamics of Theories. New York: Springer
Verlag, 1976.
Thompson, J. Organizations in Action. New York: McGraw Hill, 1967.
Thompson, J. and McEwen, W-. Organizational goals and environment. American
Sociological Review. 1958, vol. 23, 23-31.
Weick, K. The Social Psychology of Organizing. Reading, Mass.: Addison-Wesley,
1969.
- - . Middle range theories of social systems. Behavioral Science, 1974, vol. 19,
357-367.
ence Quarterly, 1976, vol. 21,1-11.
- - . Enactment processes in organizations. In B.M. Staw and G.R. Salancik
(Eds.), New Directions in Organizational Behavior. Chicago: St. Clair Press,
1977.
30
HAVING ONE'S CAKE
AND EATING IT TOO
Middle Range Content and Generalized
Process as Ways of Understanding
Organization
Peter J. Frost and David C. Hayes,
Debate on the relative usefulness of general theory versus middle range theories
of organization has taken place in recent years (e.g. Melcher, 1975; Pinder and
Moore, 1977, Weick, 1974). The intent of this paper is to argue for a dual ap-
proach to studying and understanding organization, involving the development
of partial, middle range theories to describe and explain the content of organiza-
tion, and the development of general theory to capture the essence of the pro-
cess of organizing. We use the concept of middle range in this paper to refer to
discrete, confirmed theory that " ... applies to limited ranges of data, consoli-
dates segregated hypotheses and is itself available for consolidation with other
theories" (Weick, 1974, p. 357). By general theory we mean that the elements
of such a theory (the constructs and relationships between them) are not bound
by particular settings, but represent a deductive approach based on phenomena
that are assumed to be universal in application.
We begin with an outline of our assumptions about organization, follow with
a discussion of the content-process distinction referred to above, and then pre-
sent our arguments for a middle range theory approach to organization content
and a general theory approach to the organizing process. We conclude the paper
with speculations on some likely constructs of a general theory of organizing.
354
HAVING ONE'S CAKE AND EATING IT TOO 355
ASSUMPTIONS ABOUT ORGANIZATION
We assume, as do some other students of organization (e.g., Bittner, 1965;

lehensen, 1973; Pondy and Boje, 1976; March and Olsen, 1976; McCall, 1977;
Schutz, 1967; Silverman, 1971; and Weick, 1969, 1977) that organizational
participants try to make sense of an ongoing stream of (organizational) ex-
periences they encounter. We assume further, with Weick (1977) that as individ-
uals generate and/or bracket ongoing experiences and punctuate the result,
they create causal or cause maps (Bateson, 1968; Bougon et al., 1977; Weick,
1977) that contain variables and assumed reasonable connections among the
variables, and which provide the sense making for individuals of what they per-
ceive to be "out there" in what are labelled organizations. We assume that the
behavior of organization members is interlocked through the organizing process
(Weick, 1969) and that members negotiate, through an exchange process, collec-
tive causal maps that allow for shared attitudes and actions (Day and Day, 1977;
Strauss et aI., 1963). We believe those models of organization which reflect a
concern with subjective, phenomenological, interactive, and inductive ap-
proaches to describing and understanding organization alert us to the importance
of the "eye of the beholder" and thus the role of the organization member in
determining the content of organization. We believe also those approaches to
organization which emphasize detached, objective, positivistic, and deductive
descriptions of organizational phenomena can be harnessed to explore and de-
velop a generalizable theory of the process by which organizing takes place
(Benson, 1977; Driggers, 1977).
ORGANIZATION CONTENT AND ORGANIZING PROCESS
The content of organization in our framework refers to the substance, to those

specific variables and connections among variables which constitute an organi-
zation at a given point in time, or over a finite period of time. Such variables
and their relationships are frequently the focus of interest of theorists and re-
searchers when they discuss and describe variables such as goals, structure,
technology, environment, and so forth. We shall argue in the next section that
the variables and connections among variables, which constitute the content of
organization, are largely derived from the sense that organizational partici-
pants make of the stream of experiences they either enact or react to during the
ongoing process (Weick, 1977). Consequently, there is no such thing as the
organization, in a "concrete social entity" sense, but rather there are coexisting
and frequently competing images of what the substance of organization is at
any point in time (Benson, 1977; Pfeffer and Salancik, 1978).
Organizing process refers to the mechanisms or types of behaviors involved

in the combining of individuals and groups; it is ongoing, inevitable, and a crit-
ical process for the formation and evolution of organization. Theorists who
focus on organizing process (such as Weick, 1969; Benson, 1977) frequently
assume content as given, exploring "content free" models of this process. Sim-
ilarly, theorizing about and research on organization content typically occurs
with the (often unrecognized) assumption that organization is a given concept
and the process of organizing is treated as constant. We believe the study of both
content and process to be vital for a fuller understanding of organization and
perceive them in a figure-ground relationship such that the flow of the orga-
nizing process is the ground when content is in focus, the content giving sub-
stance to the organizing process in a given situation, while content becomes the
"ground" when questions of organizing are pursued. We will discuss in the
remainder of the paper the concept of middle range approaches to understanding
content and a general theory approach to understanding organizing.
THE PARTICULARISM OF ORGANIZATIONAL CONTENT
Much of the current theorizing about "the organization" seems to be a descrip-

tion of the cause maps of dominant coalitions in various settings, particularly the
dominant coalition of managers or administrators in business or administrative
settings (Benson, 1977). (In fact, we would go one step further and say that
what we have at present frequently are the cause maps of organization theorists
writing about organization.) We think it will be more helpful in the long run for
an understanding of organization if the possiblity were explored further of
multiple images of organization at one point in time, or over time. Such an
approach would better equip the theorist to handle some of the apparent com-
plexity of organization.
Based on the notion of cause maps and of how they may be generated, it
seems reasonable to argue that the content of maps of organizational partici-
pants, the variables and reasonable connections among these variables, may
differ in a number of ways. Ifwe take a snapshot (Barnard, 1938; Day and Day,
1977), or freeze the ongoing process of organization in a given setting at one
point in time, we can derive the content of organization through the cause maps
of participants. Several possibilities emerge:
1. Cause maps of individuals and coalitions of individuals may be observed

that are identical or very similar in terms of the variables identified, the
weightings given to the variables, and the connections judged reasonable
by participants. We suspect that such an outcome reflects the negotiated
order among participants when the flow of organizational experiences is

perceived to be relatively unambiguous (March and Olsen, 1976), unequiv-
ocal (Weick, 1969), or certain (Lawrence and Lorsch, 1969), or where
beliefs about cause-effect linkages and preferences regarding possible out-
comes are certain (Thompson, 1967).
2. Cause maps may be observed that reflect variation among individuals or
coalitions of individuals. Maps may be similar in terms of the variables
identified, though weightings and/or perceived reasonable connections
among variables are different. Alternatively, the variables themselves may
differ for different individuals or coalitions so that they perceive essen-
tially different organizational worlds; their causal maps would be markedly
different. We suspect that differences in cause maps among organizational
participants are increasingly marked when the flow of experiences is per-
ceived to be highly ambiguous (March and Olsen, 1976), equivocal (Weick,
1969), uncertain (Lawrence and Lorsch, 1969), or beliefs about cause-
effect relationships and preferences among desired outcomes vary in their
certainty (Thompson, 1967), so that the stream of experience can be
generated and punctuated in a variety of ways.
The notion of similarities and differences in causal maps of participants can

be accentuated further by considering the organizational snapshot (the freezing
of organization processes) at more than one point in time and in different or-
ganization settings. We expect that there would be variation in the stability of
causal maps of individuals and coalitions of individuals over time, and that the
variables, and connections perceived among variables, would not necessarily be
the same in different settings. A recent study of a jazz orchestra provides some
interesting and useful insights into the cause map view of organizational content
(Bougon et ai., 1977). In addition, several other studies, such as those discussed
below, have reported findings that allow us to make inferences about the likely
cause maps of various participants.
The jazz orchestra study of Bougon et al. provides an example of a negotiated
collective cause map at one point in time in which members of the jazz orchestra
report an apparently common set of variables (such as satisfaction with rehearsal,
time spent rehearsing, availability of new numbers) and showed similar reason-
able connections among variables (e.g., agreement on interpretation) as means
linked to ends (such as quality of band's performance and enthusiasm about
attending) (Bougon et aI., 1977, p. 614). The content of organization as re-
flected in the cause maps identified in the study appears different, at least in its
raw form, from that which would be inferred from other situations, such as
business, university, or voluntary settings. The following discussion of inferred
"cause maps" illustrates this point. (It is an empirical question, of course,
whether the contents are meaningfully different, and if different, in what

respects they vary and what effects such variation may have.)
Organizational participants in the same organizational setting may have dif-
ferent cause maps. For example, Johnston (1976) reports that two generations
of employees in a consulting firm placed different emphasis (weightings) on
variables such as organization climate, structure, goals, participation in decision
making, and personal values as to what is good and bad. The two generations
apparently punctuated the stream of organizational experiences in different
ways. Interestingly, Johnston comments that the first generation had been
members of the firm at a time when environmental demands and task definition
were perceived to be highly uncertain, and "spontaneous ingenuity" (Johnston,
1976, p. 45) was required. Second generation employees apparently perceived
more certain environment and task definition than had the initial group. In a
sense, they perceived a stream of organizational experiences already largely
enacted and bracketed by the first employees. Nevertheless, they developed
"cause maps" in which relationships among variables were different from those
perceived by the initial group.
Other studies of business organization suggest that the sensemaking or organi-
zational participants may be different for individuals and coalitions of individuals
in formal subunits. Lawrence and Larsch (1969), for example, in their treatment
of the differentiation concept, provide evidence suggesting that in settings where
organizational experiences (the perceived environment) are uncertain, different
groupings of individuals (such as sales, production, applied research, and funda-
mental research groupings) generate different causal maps in terms of interper-
sonal orientation (people versus task), time, and goal orientations. In a more
recent study of organizational effectiveness Mahoney and Frost (1974), using
path analysis to allow causal inferences, identify different "cause maps" for
managers in departments with different technologies. These differences in cause
map content refer to the variables generated and the connections perceived
reasonably by managers when reporting on departmental performance. For
example, the perceived organizational content for departments with a long-
linked technology (Thompson, 1967) included linkages between variables such
as planning, reliability, performance, and effectiveness of the unit, whereas the
organizational content for intensive technology units included linkages between
cooperation, development, performance, and effectiveness.
Individuals' ideal images (cause maps) of organization, that is, the imagined
or pref~rred content of organization, may also differ. In an interesting explor-
atory study of organization images, Mitroff and Kilmann (1976) identified
differences in ideal organizations constructed by managers as well as those
negotiated by groups of managers, such differences being associated with vari-
ation in personality type. Some differences were in terms of weightings given

to similar variables and in terms of the connections perceived as reasonable; in
other cases, different variables were generated. One statement of Mitroff and
Kilmann is worth noting here. Comparing organizational images of managers
separated by personality type, they observe " ... One is tempted to say that the
ideal of one type is the absolute hell of the other and vice versa" (Mitroff and
Kilmann, 1976, p. 193).
Based on the sorts of findings reported here, which suggest that different
organizational participants in the same as well as in different settings may enact
or respond to different variables and/or patterns of variables, one might expect
that the attitudes and behaviors accompanying these organizational contents
may be very different and in turn may lead to different organizational outcomes.
The possible consequences of different cause maps in the same organization set-
ting is dramatized in an analysis of the failure of the merger of the Pennsylvania
and Central Railroads in the 1960s (Daughen and Binzen, 1971). The negotiated
reality of the Pennsylvania Railroad differed from that of the Central Railroad
on content such as acceptable cab signals, computers, even railroad spikes, as
well as preferred structure, attitudes toward change, and innovation and methods
of communication. Failure to negotiate a revised shared reality apparently con-
tributed heavily to the collapse of the merged Penn Central.
A Role for Middle Range Theory in Understanding Organization. The perspec-

tive developed thus far suggests to us a far more diverse world of content of
organization than has perhaps been explicitly acknowledged to date. We believe
that one consequence of not having this notion of variable content explicit is
the tendency for organization theorists to fall into the trap of talking about and
reporting research about the organization, as if such an entity existed. At this
stage in the understanding of organization we argue that there is no apparent
generalizable content of organization, and that it may be more fruitful to think
of organization content in rather particularistic terms. Middle range theorizing
then would suggest that different theories should be sought to describe, explain,
and predict the content of organization in limited settings or circumstances. This
implies a need to study the "cause maps" of different individuals and coalitions
in the same as well as in different settings, and to do this at several points in
time. To the extent that this is an accurate assessment of the situation, abstrac-
tion, and theory building to understand organization content will be arduous,
requiring painstaking "site by site" observation, documentation, and analysis.
We see no other viable path if content is the focus of interest.
BEYOND MIDDLE RANGE THEORIZING
Is there no pOint in searching for a general theory of organization? Is there

nothing portable or general about organization that would allow us to anticipate
comprehensive theories of organization? We believe that there is a role for
general theory, that there is a need to develop a generalized formula of organiza-
tion to which, paraphrasing Bittner (1965), the substance of a given organization
setting or cluster of settings can be referred. We perceive, with Benson (1977)
that such a generalized formula, or general theory can be developed from a study
of the process through which organizing occurs. While a dialectical framework
utilized by Benson (1977) is not the only way to approach the study of orga-
nizing, we find his comments on what a general theory of the organizing process
should provide most pertinent to our arguments. Thus, following Benson (1977,
p. 2) we suggest that a processual perspective and the accompanying generalized
framework would allow for an explanation of the process leading to the emer--
gence, continuation, and decline of organizational forms, would enable analysis
of the processes through which organizational participants make sense of or-
ganizational events, and finally, would provide a basis for explaining " ... the
empirical grounding of conventional theories because it deals with the social
processes which conventional theories ignore" (Benson, 1977, p. 2).
A study of the process of organizing is concerned with mechanics, with how
organizing takes place. We perceive the mechanisms of organizing to be content
free in that they are likely to occur regardless of the characteristics of organiza-
tional members of the particular situation. Weick (1969) illustrates this point
through his use of the concept of interlocking behavior, a notion we discuss in
some detail below. The interlocking mechanisms in Weick's framework are
assumed to occur whenever the process of organizing takes place and provide
much of the continuity observed in given organizations. A general theory of
the organizing process would contain other such concepts, and we speculate on
some likely ingredients in the next section of the paper. In keeping with our
figure-ground analogy for the content-process distinction used earlier, we expect
that the particular sense-making of participants in an organization provides the
relatively unique substance to that particular organization or cluster of or-
ganizations. The mechanisms generalize, the outcomes do not.
We are not aware of anyone theory of organizing which has general accep-
tance, but we do perceive several frameworks that have a "general theory"
intent, which implicitly or explicity are concerned with trying to explain and
predict organizational process, and these will be discussed later in the paper.
There appear to be convergences in these writings providing the basis for some
speculation, some playfulness, to provide a sense of what concepts and relation-
ships might find their way into an overarching general theory of organizing.
What follows are our deductive speculations on possible ingredients of a general

theory of organizing.
SPECULATION ON INGREDIENTS OF A GENERAL THEORY

OF ORGANIZING
Our speculations on possible concepts and relationships to be incorporated in a

general theory of organizing are based on the literature dealing with exchange
theory (Emerson, 1962, 1976; Georgiou, 1973; Nord, 1976; Pfeffer and Salancik,
1978; White, 1974), with negotiated reality theory (Day and Day, 1977); with
conflict theory and with action theory (Silverman, 1971). In addition, we have
drawn from one model that is directed explicitly toward describing the orga-
nizing process (Weick, 1969).
Weick defines the process of organizing in terms of " ... resolving equivocality
in an enacted environment by means of interlocked behaviors embedded in con-
ditionally related processes" (1969, p. 91). These interlocked behaviors, assumed
content free, are for Weick the essential elements of organizing.
We perceive three aspects to the organizing process as Weick develops it
(Weick, 1969, 1977). First, there is the development of individual cause maps
through sense-making as individuals attempt to remove equivocality from im-
pinging streams of experience. Second, there is the interlocking of behaviors of
individuals and coalitions of individuals through which participants attempt to
remove residual equivocality. This is a process in which efforts are made by par-
ticipants to negotiate a collective cause map, a common negotiated content of
organization. The negotiation process may produce such a map, or it may pro-
duce an absence of commonality reflected in competing cause maps of collec-
tives and of individuals. Third, there is a process of attempted implementation of
content contained in the "presumably more homogeneous causal maps" (Weick,
1977, p. 296) either in similar future circumstances, as Weick suggests, or,
looked at somewhat differently, in the implementation in the present of the
outcomes implied by the casual maps.
Several aspects of the process of organizing strike us as important:
1. The process of removing equivocality by individuals and coalitions of

individuals appears analogous to the notions of removing the uncertainty
that writers such as Thompson (1967) and Lawrence and Lorsch (1969)
describe as organization process. Their perspectives appear to be restricted
to the uncertainty related actions of one "group" only, typically the dom-
inant coalition.
2. The process of removing equivocality can be an individual as well as a col-
lective process. The collective process involving interlocked behaviors gives
us the essential organizing process. Further, it appears sensible in the light

of both individual and group sense-making behavior to treat interlocked
behavior as a variable, so that such behavior can be viewed in terms of
loose or tight coupling.
3. The negotiation of a shared negotiated order of organization should also
be treated as a variable. Such negotiations can succeed or fail in varying
degree, allowing the possibility of a single homogeneous shared cause map,
competing maps shared among different groups and individuals, and idio-
syncratic cause maps of individuals, whose behavior is perhaps loosely
coupled in the organizing process. The negotiation of order should be
characterized by both cooperation and conflict, given this perspective
(Weick, 1969; Day and Day, 1977; Manning, 1977).
The process of organizing, involving the removal of equivocality by individ-

uals and by negotiation within and among groups of individuals, is accomplished
in the framework outlined above through interlocking behaviors. We perceive
two major generalizable dimensions of interlocking behavior which, made ex-
plicit, enable us to get a better understanding of the richness of this concept.
The first dimension refers to the perceived style of the interlocked behaviors
involved in the process of negotiating a common or shared cause map. The at-
tempts to accomplish (as well as to resist) this can be made through behaviors
characterized as either consensus or nonconsensus by participants in the process,
that is, the degree of acceptability of the behaviors. Weick suggests, with Simmel
(1950), that interlocking behaviors can, and perhaps should, reflect selfish and
altruistic, cooperative and conflictful behaviors. Applying an exchange frame-
work allows further development of the notion of consensus and nonconsensus
interlocked behaviors (Frost and Hayes, in press: Georgiou, 1973; Nord, 1976;
White, 1974). It is argued that organizational participants interlock behaviors in
an effort to negotiate common or shared cause maps of organization. The negoti-
ations that occur involve the exchange of information, as well as other resources,
between individuals and groups of individuals. In the process the variables and
assumed causal connections between variables are established as common and/or
are disputed to varying degrees. So the maps negotiated provide the guidelines
for ''who does what, with and to whom, why, when and how." The organizing
process, manifested in the interlocked behaviors, provides the vehicle or the
medium for ongoing organizational membership.
The other dimension of interlocked behavior refers to the degree of routineness
involved in the process of equivocality removal. Highly unequivocal inputs
should be handled through invoking rules (Weick, 1969) or standard operating
procedures (Nystrom, et at, 1976). On the other hand, highly equivocal inputs
should be handled by nonroutine, unpredictable behaviors (Weick, 1969) such as
obtaining feedback on an ad hoc basis.
We perceive, therefore, that the interlocked behavior of any organizing pro-

cess can be characterized in terms of their routineness and in the degree to which
they are perceived as consensual by organizational participants. We believe that
such a view provides a means to expand our understanding of what interlocked
behaviors "look like" across a wide variety of organizational settings. This in
turn should assist in developing general theory about organizing processes.
Illustrating Potential Aspects of the Organizing Process. We have conjectured

above that the two-dimensional perspective on interlocking behaviors provides
part of a framework for interpreting the organizing process. The generality of
the approach cannot be demonstrated in this paper, given the need to report on
a multiplicity of studies, in a variety of settings, over a number of time periods,
and the absence of such studies at this point in time. As a way to illustrate the
concepts discussed in this section as well as emphasizing the mix of interlocking
behaviors and conveying a sense of the flow of the organizing process, we have
included a description of a series of events, taken from a study of organization
(pettigrew, 1973) (Figure 1).
Figure 1. The Purchase of a Computer. *
Throughout 1963 and early 1964 Ramsbottom pressed for further computeriza-
tion of the business. When the issue of replacing the outdated SE 100 arose, he
argued that the new machine should be used to do both the work the SE 100
was doing and the work of one of the clothing administrative offices. It was his
insistence on the latter point that finally lost him the support of the directors
and allowed Jim Kenny to move back into formal control of the 0 & M
department.
Ramsbottom's explanation of his lack of success was as follows:
... The directors weren't too keen on having computers in the administrative offices.
This was the main side of the business where feelings were strongest and where the
most entrenched systems were to be found. To me it was a prolific area for computeriza-
tion, but in the minds of the entrenched this was the heart and soul of the business ...
. . . Neil Turner offered the most complete description of what happened in

the Newton 350 decision process:
Ramsbottom raised the issue of the administrative offices and got himself in trouble
with the management. But then they asked him to look at one of the administrative
offices and to replace the SE 100. They then changed their minds and told him to
forget about the administrative office and see about replacing the 100. We didn't take
any part in this. We just said something ought to be done about replacing the 100 ...
. . . Fred (Ramsbottom) kept on fiddling about with the administrative office plus the
SE 100, and eventually completed a report recommending BDC. The very next day he
*Summarized from Pettigrew, 1973, pp. 113-115.

saw a Victor representative and proceeded to change the name on the report to recom-
mend Victor while leaving the main body of the report untouched.
At this point in the interview, I could no longer maintain my neutral, interest-

ing interviewing face. Turner continued:
It's absolutely true. I've seen the stencil. Eventually he was told by the management to
stop buggering about and get on with replaci ng the 100 ....
We- Parkin, Jim, and I-had a meeting to decide between BCD and Newton.
Ramsbottom wasn't invited. He was left to make his own decision. Parkin thought
Newton was the best, which, of course, Jim was all for then. I didn't care either way ...
We stated right from the very beginning, we wouldn't have NTL under any circum-
stances. This caused some real rows with NTL and Mills, but this time we had plenty
of experience to prove that their access device was too unreliable. We recommend
Newton and left Fred (Ramsbottom) to keep mumbling about BCD, but nobody took
any notice of him ...
. . . Ramsbottom suggested that the directors rejected NTL for technical

reasons and because the company did not wish to become too dependent on one
supplier ...
. . . The Michaels board had to sign the cheque, but the crucial intervention
in the Newton 350 decision process was made by Kenny.
The following is Kenny's version of what happened:
Ramsbottom was putting up a BDC 1700 with a painless transfer to a BCD 250. I said,
"Not on your life". I mentioned this to one of the directors. He said he wouldn't go
along with it either. Ramsbottom changed his mind for a Victor 300. I wasn't happy
about this either. I knew their maintenance was poor. I rang up the managing director
and told him I was worried. He said, "Sit back and wait".
It was in this worried state that I met Henry de Ville .... De Ville said, "I understand
you're considering buying BCD. If you don't buy Newton you're mad. The BCD isn't
off the production line yet." This was just sales talk. I hear it every day. I wanted to
have a look at some of their installations. I called in Neil Turner and asked him what
Newton were like. He said he knew nothing about them ...
De Ville made a wise move. He sent an experienced software man called Sanderson
around with Turner. They were away a week. Turner was impressed when he came back.
I saw de Ville again and all the top people in London. They convinced me I would get
good backing. I finally came back convinced.
We feel that the case illustrates the essence of our arguments about inter-
locked behavior and the organizing process. Interlocking appears as a variable,
sometimes loose, sometimes tight; behaviors are evidenced that represent at-
tempts to remove equivocality/uncertainty by individual action as well as
through behaviors aimed at negotiating and resisting negotiation of an organiza-
tional order. The dimensions of routineness and consensus are clearly reflected
in these behaviors, as are differences in causal maps of individuals and coalitions
participating in the process. The content and process of this organization de-
cision are illustrated in Table 1.
Table 1. The Purchase of a Computer.
A. ORGANIZATIONAL CONTENT
Common Organizational Images of Participants: Replacing an outdated
computer SE 100 will improve flow of information and the decision
making of managers
Competing Images of Participants:
1. Ramsbottom:
• Applying a new computer to the administrative offices is an efficient
use of computer and must be pushed
• Top management resists this move because of its entrenched views
• I prefer a BDC or a Victor Computer
• Top management rejected NTL for technical reasons
2. Top Management:
• Keep the new computer in existing areas and out of our domain
• Search best handled by delegation to next level of managers
3. Staff Managers (coalition):
• Ifwe try to push computer into administrative offices we are threat-
ening the power of gp-so don't tamper
· We don't want NTL, we will choose between BDC and Newton
• Ramsbottom doesn't count
4. DeVille (Sales Manager Newton);
• I want Michael's organization to buy our computer
• Experienced sales representatives and personal touch will influence
the outcome
B. ORGANIZING PROCESS
The setting: an organization deciding to replace a computer.
Removal of equivocality (what computer shall it be?) via negotiated
reality which involved interlocking behavior characterized by both
degrees of routineness, and of consensus.
Interlocking Behaviors Involved:
• Ramsbottom (low degree of interlocking)
• Turner, Kenny, Parkin (high degree of interlocking)
• Kenny with directors (moderate degree of interlocking)
• Sanderson (sales representative) with Turner (high degree of inter-
locking)
Behaviors as:
I. Routine • Completing a report recommending BCD
II. Nonroutine Telephoning MD to voice misgiving
Allocating sales representative to spend a week
with manager
Ill. Consensus • Discussion with sales people
IV. Nonconsensus Changing report to suit personal preferences
Excluding key person from meetings
CONCLUDING REMARKS
We have attempted to argue that if one is interested in studying what we refer

to as the content of organization, the "stuff' of cause maps (e.g., structure,
goals, technology, hierarchy, internal processes), then a middle range approach
is the appropriate means of bringing such variables into focus. However, if the
main figure or focus of the research is study of organizing process, we have sug-
gested that a more appropriate strategy is to try to develop a general theory of
process. It is our belief that the mechanisms of organization, of how the orga-
nizing process occurs, is not situation specific.
Clearly, one of the difficulties in studying organizing process is that in order
to infer the means by which coalitions form, combine, and disintegrate, the per-
ceptions and understandings of all parties need to be tapped. Then, inferences
may be possible concerning the nature of the cause maps, the methods of nego-
tiation employed, the means by which equivocality is removed, and the reasons
why these behaviors occurred. Such inferences are more easily drawn from fic-
tion, from novels, where the author punctuates the world for all the characters
and develops motives and explanations for behaviors. Wheels is a good example
of a rich, seemingly true-to-life description of organizing processes in fictional
form (Hailey, 1971).
Methodologies for such study that show promise for the study of organizing
in a nonfiction context include the Bougon, et al. (1977) mapping of causal
maps, Mitroff and Kilmann's (1976) study of myths and image making and
Pettigrew's (1973) interview study of a computer acquisition decision.
The richness of such studies and the insights into people's perceptions of
variables and connections between them strongly suggests that further study
of organizing processes is likely to be rewarding. In particular, the search for
commonalities in interlocking behaviors may be extremely fruitful in terms of
the development of generalizable theory. Mere hints of such behaviors and their
generalizability have been presented here, yet we believe that they are extremely
encouraging leads in such theory development.
Our view of the path to a better understanding of organization, in fact, re-
quires the development of two maps addressing different questions, but which
are clearly interrelated. We perceive the need to consider the content of organi-
zation as a particularistic phenomenon, amenable to theories that are rather
limited in range, but which may also be capable of consolidation with other
partial theories of content. Interwoven with this mapping of content, we per-
ceive the process of organizing to be a more generalizable phenomenon, appli-
cable across organization settings and across the dimension of time. We see the
notion of interlocking behaviors as a crucial ingredient of this process and we
believe that an exploration of such behaviors, testing for generality, will be
helpful in the development of general theory of the organizing process.
REFERENCES
Barnard, C.1. The Functions of the Executive. Cambridge, Mass.: Harvard

Bateson, G. Steps to an Ecology of Mind (1972 reprint). New York: Ballantine,
1968.
Benson, J.K. Organization: A dialetical view. Administrative Science Quarterly,
1977,22, 1-21.
Bittner, E. The concept of organization. Social Research, 1965,32,239-255.
Bougon, M., Weick, E., and Binkhorst, D. Cognition in organizations: An analy-
sis of the Utrecht Jazz Orchestra. Administrative Science Quarterly, Dec.
1977,vol. 22,no.4,606-639.
Cyert, M., and March, J.G. A Behavioral Theory of the Firm. Englewood Cliffs,
N.J.: Prentice-Hall, 1963.
Daughen, J.R., and Binzen, P. The Wreck of the Penn Central. Little, Brown &
Co., 1971.
Day, R.A., and Day, J. V. A review of the current state of negotiated order
theory. In J.K. Benson (Ed.), Organizational Analysis. Beverly Hills: Sage,
1977,128-144.
Driggers, P. F. Theoretical blockage: A strategy for the development of organi-
zational theory. In J.K. Benson (Ed.), Organizational Analysis. London: Sage,
1977,147-159.
Emerson, R.M. Power-dependence relations. American Review, 1962, 27,
31-41.
- - . Social exchange theory. In Annual Review of Sociology. Palo Alto: Annual
Reviews, Inc., 1976.
Frost, P.J., and Hayes, D.C. An exploration in two cultures of political behavior
in organizations. In A. Neghandi (Ed.), Cross-Cultural Studies in Organization
Functioning. Kent, Ohio: Kent State University, in press.
Georgiou, P. The goal paradigm and notes towards a counter-paradigm. Admin-
istrative Science Quarterly, 1973,16, 216-229.
Hailey, A.H. Wheels. New York: Bantam, 1973.
Jehenson, R. A phenomenological approach to the study of the formal or-
ganization. In G. Psathas (Ed.), Phenomenological Sociology: Issues and
Applications. New York: Wiley, 1973,219-247.
Johnston, H. Interactions between individual predispositions, environmental
factors, and organization design. In R.H. Kilmann, L.R. Pondy, and D.P.
Slevin (Eds.), The Management of Organization Design (Vol. II). New York,
North-Holland, 1976,31-58.
Lawrence, P. R., and Lorsch, J. W. Organization and Environment. Homewood,
Ill.: R.D. Irwin, 1969.
McCall, M. W. Making sense with nonsense: Helping frames of reference clash.
In P.C. Nystrom and W.H. Starbuck (Eds.), Prescriptive Models of Organi-
zations. Amsterdam: North-Holland, 1977, 111-123.
Mahoney, T.A., and Frost, P.J. The role of technology in models of organiza-
tional effectiveness. Organizational Behavior and Human Performance,

February 1974, vol. 111, 122-138.
Manning, P.K. Rules in organizational context. In J.K. Benson (Ed.), Organi-
zational Analysis. Beverly Hills: Sage, 1977, 46-63.
March, J.G., and Olsen, J.P. Ambiguity and Choice in Organizations. Olso:
Universitetforlaget, 1976.
and realizations. Organization and Administrative Sciences, 1975,6, 3-10.
Mitroff, 1.1., and Kilmann, R.H. On organization stories: An approach to the
design and analysis of organizations through myths and stories. In R.H.
Kilmann, L.R. Pondy, and D.P. Slevin (Eds.), The Management ofOrganiza-
tionDesign (Vol. I). North-Holland, N.Y., 1976, 189-208.
Nord, W.R. The Study of Organizations Through a Resource-Exchange Para-
digm. Unpublished manuscript, University of Washington, St. Louis, 1976.
Nystrom, P.C., Hedberg, B.L.T., and Starbuck, W.H. Interacting processes as
organization designs. In R.H. Kilmann, L.R. Pondy, and D.P. Slevin (Eds.),
The Management of Organization Design (Vol. I). North-Holland, N. Y.,
1976,209-230.
Pettigrew, A.M. The Politics of Decision Making. London: Tavistock, 1973.
Pfeffer, J. and Salancik, G. R. The External Control of Organizations. New York:
Harper and Row, 1978.
Pinder, C.C., and Moore, L.F. The Resurrection of Taxonomy for the Develop-
ment of Middle Range Theories of Organizational Behavior. Working Paper
No. 512, University of British Columbia, 1977.
Pondy, L.R., and Boje, D.M. Bringing Mind Back in: Paradigm Development as
a Frontier Problem in Organization Theory. Unpublished working paper,
University of Illinois, 1976.
Schutz, A. The Phenomenology of the Social World. Evanston, Ill.: North-
Silverman, D.S. The Theory of Organizations. New York: Basic Books, 1971.
Simmel, G. The Sociology of Georg Simmel (trans. K.H. Wolff). New York:
Free Press, 1950.
Strauss, A.L., Schatzman, L., Ehrlich, D., Bucher, R., and Sabshim, M. The
hospital and its negotiated order. In E. Freidson (Ed.), The Hospital in
Modern Society. London: Free Press of Glencoe, 1963, 147-169.
Wesley, 1969.
- - . Middle range theories of social systems. Behavioral Science, 1974, vo1.. 19,
357-367.
- - . Enactment processes in organizations. In B.M. Staw and G.R. Salancik
(Eds.), New Directions in Organizational Behavior. Chicago: St. Clair Press,
1977,267-300.
White, P.E. Resources as determinants of organizational behavior. Administra-
tive Science Quarterly, 1974,19, 366-379.
31
DISCUSSION OF "THEORY
DEVELOPMENT IN ORGANIZATION
BEHAVIOR'; BY COOPER AND WOLF
AND "HAVING ONE'S CAKE
AND EATING IT TOO"
BY FROST AND HAYES
Bill McKelvey, University of California,
Los Angeles
There is a common thread between the two papers I am to discuss. Both papers
argue that there is a need for general theory despite the legitimate calls for mid-
dle range theory. However, since the arguments take rather different forms and
comprise differing proportions of the total length of the papers, I will discuss the
papers separately. Both papers are thought provoking and contain many valid
points. In my remarks I have tended to focus on points that were important
issues but not as well developed as one might like.
"THEORY DEVELOPMENT IN ORGANIZATION BEHAVIOR:

A SYSTEMS PERSPECTIVE" BY DAVID COOPER AND
FRANK WOLF
A Paradigm Shift?
It would seem that one person's paradigm shift is another person's pebble polish-
ing. Like the watched pot that never boils, it is possible that a paradigm shift
happens only when scientists aren't looking, meaning that they are spending
their time in discovery rather than in idle chatter. I fear the time when one will
369
have to claim a paradigm shift to feel worthy of submitting a paper to a top

journal. Needless to say, like the rumors of Mark Twain's death, such claims are
usually greatly exaggerated!
Scott's (1974) paper would appear to be an example of such an exaggeration
and consequently offers a shaky point of departure for the Cooper and Wolf
paper. It is probable that the shift from classical management theory to open
systems theory did rank as a paradigm shift because the former could not handle
such anomalies as: two-way causal flows, adaptation to environmental forces,
many interdependencies, conflict, multiple goals, subordinate and group choice-
fulness, and what Gouldner (1959) termed natural system forces, and so forth.
Unfortunately, Scott failed to identify a parallel set of anomalies to support
his contention that another paradigm shift is about to do away with systems
theory. I see nothing in the items Scott listed, such as declining reserves, anti-
pollution, humanism, and lack of consensus that denies the need for systems
concepts such as: inputs and outputs, conversion technologies, feedback, nega-
tive entropy, equifmality, and the like. If my view of the situation is true,
Cooper and Wolf are thereby cast in the position of mounting a long argument
rebuilding something that was never really torn down in the first place.
Without explicit discussion of anomalies I am inclined to believe that any
talk about paradigm shifts is fruitless. In my view Cooper and Wolf could greatly
strengthen their arguments by focusing on whether certain anomalies exist in
conjunction with the use of systems theory.
One of the difficulties I have with both the Scott and the Cooper and Wolf
papers is the failure to clearly distinguish between kinds of systems and kinds of
causality. To me the element of paradigm shift significance in the move from
classical to systems theory was the shift from the reductionist to the holistic
principle of enquiry (see Schwab, 1960, and McKelvey [1978a], the paper pre-
sented at this conference). In this view the 1958 March and Simon book was the
last major treatise clearly based on reductionism. Now we see, with the explica-
tion of systems theory by Ackoff (1971) that there are several kinds of systems.
Holism, with its view of causality as a two-way flow up and down hierarchical
levels, becomes a general case with reductionism and rationalism becoming
opposite causal flows applicable in special circumstances.
Many people feel that systems theory has not lived up to its promise. One
probable reason is that scientific methodology remains rooted in what Kaplan
(1964) termed the deductive or hierarchical or logical mode, best suited for
either reductionism or rationalism. The pattern or concatenated or causal mode
never developed to the point where criteria separated good and bad practice
(see Wallace, 1971, for a quick review of these points). In short, the shift in the
theoretical paradigm was not matched by a shift in the methodological paradigm.
It is possible that a subparadigm shift is underway at the moment in the
DISCUSSION OF THEORY DEVELOPMENT IN ORGANIZATION BEHAVIOR 371
method of explaining organizations as entities-what traditionally has been

termed organization theory and recently is sometimes referred to as macro-
organizational behavior. For March and Simon (1958) the attempt was to ex-
plain macro-organizational behavior largely if not entirely by looking to lower
levels-managers especially and employees in general-a good case of reduc-
tionism. With systems theory it la Katz and Kahn (1966) the idea was to look to
the whole and to the parts together-causality bouncing around all over the
place within the organization's skin. (Though "open systems" was a popular
term, in fact the environment and any thought of environmental determinism
gave way to need theory and individual determination.) With the appearance of
the contingency theorists, dating back to Woodward (1958) and currently being
extended by "evolutionists" such as Campbell (1969), Hannen and Freeman
(1977), McKelvey (1978a, b, c), and Aldrich (1979), much more credit is given
to environmental determinism and the underlying causal model, rationalism.
While holism and individual choice are tolerated in the short term as forces
affecting the behavior of individual members of a specie, just as individuals'
choices lead to obesity, suicide, or self-actualization, organizational form is in
the long run driven solely by environmental forces, in this view. Systems theory
remains as the overarching conceptual framework but within it the subparadigm
shift is toward the simpler rational causal view, with the result that the theoreti-
cal paradigm is back in sync with the methodological paradigm.
What is Good Organization Science?
One of the many instructive points of the Cooper and Wolf paper is their bring-
ing to our attention the absence of criteria for deciding what is good and bad
organization science. However, in passing I should note that I thought the authors
pulled their punch by not attempting to list or discuss possible criteria. Hempel
(1965) suggested two goals for science: (1) to satisfy curiosity; and (2) to help
the human race gain leverage on its environment. The authors cite Popper (1959)
as proposing scientific theories are "to rationalize, to explain, and to master." I
was struck by the footnote in which the authors more or less were apologizing
for their inclusion of the word "mastery." It made me realize that there are good
and bad words in organizational behavior. Some bad words are: manipulation,
direction, control, determinism, and behaviorism. Some good words are: free-
dom, choice, growth, and needs. Don't get me wrong; I think it is good that we
now recognize and value choicefulness among organizational members. But I am
not convinced we are yet willing to give full recognition to the constraints im-
posed by the external environments of individuals and organizations. The issue
is one of Individual Purposefulness versus Environmental Determinism. This
issue recently has been taken up by Salancik and Pfeffer (1977; 1978) in the
micro-organization behavior context.
To identify criteria one has to buy into a paradigm. Within the rational-
causal subparadigm which I prefer, some possible criteria are:
How well does the science describe and explain niche attributes, evolution,
and development?
How well does it explain organizational behavior in terms of niche attributes?
How useful is it to practitioners?
How great is the incidence of practical improvements emanating from scien-
tific theory and fmdings?
How successful is it in terms of Schwab's (1960) four criteria of a good
theory (i.e., interconnectivity, adequacy, feasibility, and continUity)?
Is Purpose a Necessary Concept?
Despite the sad record of structural-functionalism (Merton, 1957), as a theoret-

ical basis for empirical inquiry, the authors still advocate attention to function
and purpose. Taking the open systems idea seriously (decidedly not the case in
the past) suggests one view organizations as part of an ecological system, as
Hannen and Freeman (1977) did. On the one hand, it is possible that all ele-
ments of an ecosystem could be seen as having a purpose/function in maintain-
ing the system. Thus, the purpose of rodents is to dig holes to provide homes
for snakes and to aid water percolation into the ground. The purposes of hippo-
potamuses and beavers are to create mudwallows and ponds to store water
for other animal and plant life. The purpose of schools is pattern-maintenance
(after Parsons, 1956). But, on the other hand, elements of an ecosystem could
also be seen as simply being competitive in a niche that is interdependent with
other niches. Needless to say, the latter is the ecological view. (See Hannen and
Freeman, 1977) for a full discussion of niche theory in particular and popula-
tion ecology in general.) Thus, an organization is not for any purpose. It is just
an entity having certain technical skills that mayor may not allow it to survive,
grow, adapt, and so on, to a niche.
Focusing on purpose diminishes the importance of interdependency, the
ultimate systems concept. In disagreement with a statement the authors attri-
bute to Ashby (1956), which is that systems thinking focuses on what an orga-
nization does, I would argue that systems thinking emphasizes links among
entities rather than only describing the entities and what they do, though of
course the latter are also necessary. Goals and purpose are best seen as tempo-
rary enactments (after Weick, 1969) that come and go in individual members of
a specie during the long-term competitive struggle of a specie for resources in its
interdependent ecological network. I conclude purpose is an unnecessary concept

for the explanation of macro-organizational behavior. I think this is one place
where the biological hippo-mudwallow metaphor allows a much more insightful
view of organizations than the human metaphor.
Is Systems Thinking Necessary for Taxonomic Development?
The authors argue that: (a) meta-theory is necessary to guide taxonomic formu-
lation through the tangle of the many attributes presently seen to characterize
organizations, and (b) systems thinking can provide a unifying conceptual frame-
work. As stated, the authors are correct but not very helpful.
What kind of meta-theory? Presumably one rooted in holism. In my paper
presented at this conference (McKelvey, 1978a) I noted that holism is the only
principle of enquiry not heretofore used in biological or organizational taxon-
omy. For a variety of reasons I argue for an evolutionary theory rooted in the
rational principle of enquiry. The causal complexity of holism has, as I have
previously noted, prevented it from success elsewhere in organization science.
I see little reason why that would not be true for taxonomic inquiry, too. The
numerical taxonomists (Sneath and Sokal, 1973) have argued that no a priori
theory is necessary. In light of these arguments the authors' bland call for meta-
theory doesn't wash. I would like to see the authors carry forward explicit
arguments for holism in taxonomy.
In the search for taxonomic characters, systems theory seems to provide an
overall, very broad framework. But, taking the view that systems theory is to
organizations what water is to fish, meaning an all-pervasive medium in which to
exist and through which to view the world, successful search for taxonomic
characters now calls for much more precision than possible with systems con-
cepts as presently developed. Of course, if systems theory is further explicated
into a fmer set of elements, as the authors call for, it may be of more obvious
help to taxonomists.
"HAVING ONE'S CAKE AND EATING IT TOO: MIDDLE RANGE

CONTENT AND GENERALIZED PROCESS AS WAYS OF
UNDERSTANDING ORGANIZATION" BY PETER J. FROST
AND DAVID C. HAYES
No Such Thing as THE Organization????
The authors say, "Consequently, there is no such thing as the organization,

in a 'concrete social entity' sense, but rather there are coexisting and frequently
competing images of what the substance of organization is at any point in time."

I wonder about that. Seems to me it takes considerable imagination to think that
just because one can identify individual cause maps, what Schutz (1953) referred
to as conceptions of reality, that no objective reality exists. We are all well aware
of the fact that every time there is an airplane or automobile wreck there are as
many different accounts of how it happened as there are eyewitnesses. Yet,
surely you would not deny the objective reality of what transpired to create the
wreck. So why do this for organizations?
Part of the difficulty arises because organizations are made up of component
entities that range from concrete to abstract nature. And as Schutz pointed out,
the common sense conceptions of the members intervene between the concrete-
abstract reality of an organization and the scientific conceptions organization
scientists are trying to develop. Thus, there is no question in my mind that it is
more difficult for members and scientists alike to validate their images of an
organization. But that is not sufficient reason to conclude there is no such thing
as the organization -not yet anyway. What happened to objective things such as
long-linked assembly lines, rank hierarchies, closed doors, spatial relations, and
the chief executive officer's behavior?
I can see by the references cited by the authors and by the list of those at-
tending this conference that those exploring the subjective side of organizational
behavior probably outnumber those, if any, who follow Peter Blau's lead in
sticking close to the objective side. I do not want to be placed in either camp,
nor forced into an either-or type of debate. Instead, following the lead of Diesing
(1966), I took the view some time ago (McKelvey, 1973) that the matter was
really a methodological/empirical question. Methodologically one can translate
many subjective interpretations of reality into language concepts, as Osgood,
et al. (1957) did, and then study the language usage via objective methods. And
rather than argue about it, I would prefer to see investigators establish empiri-
cally just which aspects of organizations are objective and which have no basis
other than individual or social construction.
How Particularistic?
I am confused as to just what kind of middle range theory the authors advocate
for handling organizational content. They take great pains to state that cause
maps vary by individual. Presumably, one could think the authors were arguing
for a separate theory for each organizational member-an impractical idea, I am
sure they would agree. Probably they are arguing for theory compartmentalized
by kinds of negotiated cause maps. But they cite data showing there could be
more than one map per organization. This means, since they don't say otherwise,
that they anticipate more than one theory per organization. How many theories
for a large organization like General Motors, Inc.? How many for the United
States?
I don't see much scientific payoff (i.e., good science) from a super-particu-
laristic middle range theory. While the goal of satisfying curiosity (and perhaps
getting things published and beating the university reward system) is met by
super-particularistic theory (I suspect this has already happened in leadership
and motivation theory, which is to say it is great for the professors and useless
to students and managers), such theory may offer little aid in helping people gain
leverage on their environment. What possible good is it to anyone for some scien-
tist to spend his or her time and somebody else's money studying a situation
and developing theory that is not expected to be generalizable to even one other
setting? And yet that is what the authors, and many other people at this confer-
ence, would appear to want us to do! There are some exceptions. I could see the
tremendous importance of a study of former President Nixon's cause map and
the negotiated cause map of the Watergate crew because of the power position
they were in and the great national attention. But what scientific use is there for
a similar study of the CEO of Amalgamated Pea Co. or Boy Scout Troop N?
Enough already!
Of course the reason Merton (I957) picked middle range theory in the first
place was to steer a course between the useless super-abstractions of Parsonian
general theory and the super-particularism of mindless empiricism -subjective
or objective data about some fly-by-night social grouping. He knew that scien-
tific payoff was the product of generalizability times precision. Too parsimoni-
ous a compartmentalization lowers precision and one too complex lowers
generalizability. My personal solution to this problem should be obvious-
better classification!
Why are Organizing Processes Expected to be Nonparticularistic?
I think the paper is aptly titled, but it is the authors having their cake and eating
it, not organization science. Given the arguments made in support of individual
and negotiated cause maps and consequently very particularistic middle range
theory, I do not see how the authors could have it the other way around and
conclude that such particularism did not also apply to the process of organizing.
Why do the authors, " ... perceive the mechanisms of organizing to be content
free in that they are likely to occur regardless of the characteristics of organiza-
tional members of (sic, it should be or, I think) the particular situation." Why
don't cause maps affect how organizing takes place? I am confused by the fact
that at one point the authors say that organizing is independent of content
particularities and yet earlier in their paper they say, in using the figure-ground
analogy, that content gives substance to organizing process? Given that most
theories are hierarchical, with very abstract general statements at the top and
substantive particularistic statements at the bottom, it seems to me that the
authors have chosen to focus on the bottom of the hierarchy in dealing with
content but stayed at the top of the hierarchy when dealing with process. Why
not the other way around? If it doesn't work the other way around, I would
like to see some argument to that effect. Otherwise, what's good for the goose
is good for the gander!
The authors (following Weick, 1969) focus on uncertainty or equivocality
reduction. Surely there is more to organizing than that. What happened to other
constraints imposed by the external environment (the objective constraints the
authors choose to ignore) such as threat or hostility, heterogeneity, dynamism,
or what Khandwalla (1977) termed restrictiveness? And there are probably more.
(One of my pet peeves is that the contingency theorists and others have tended
to treat environments as having one attribute dimension, certainty-uncertainty,
assuming that everything else is absorbed into that. Khandwalla's study shows
that is not true.
It is possible that there are individual, collective (Le., socially constructed),
and objective organizational (management layers and horizontal buffers, etc.)
uncertainty reduction approaches, and further, that some of these are subjective
and some objective. I suspect that theory about organizing process needs to
be elaborated along these dimensions, whether or not it ends up being general
or particularistic.
In conclusion, I feel this paper is instructive in pointing our attention to
how best to handle theories about content and process, but I feel some further
elaboration is necessary and some areas of confusion and inconsistency need to
be cleared up. I also feel that, in organizational behavior at least, the bandwagon
toward subjectivity is about to leave the station. The scientific implications of
this can of worms need to be given more thought than I see happening to date.
At this point, I see little real payoff in the more extreme renditions. If this paper
gets us to discuss some of these problems, we will be well served indeed.
REFERENCES
Ackoff, R. L. Towards a system of systems concepts. Management Science.

1971,17,661-671.
Aldrich, H. Organizations and Environments. Englewood Cliffs, N.J.: Prentice-
Hall, 1979.
Ashby, W.R. An Introduction to Cybernetics. London: Chapman and Hall,

1956.
Campbell, D. T. Variation and selective retention in socio-cultural evolution.
General Systems, 1969,14, 69-85.
Diesing, P. Objectivism vs. subjectivism in the social sciences. Philosophy of Sci-
ence, 1966,33, 124-133.
Gouldner, A.W. Organizational analysis. In R.K. Merton, L. Brown, and L.W.
Cottrell, Jr. (Eds.), Sociology Today. New York: Basic Books, 1959,400-
428.
Hannen, M., and Freeman, J. The population ecology of organizations. Ameri-
can Journal of Sociology, 1977,82, 929-964.
Hempel, C.G. Aspects of Scientific Explanation. New York: Free Press, 1965.
Kaplan, A. The Conduct of Inquiry. San Francisco: Chandler, 1964.
Wiley, 1966.
Khandwalla, P.N. The Design of Organizations. New York: Harcourt, Brace,
Jovanovich, 1977.
March, J.G., and Simon, H.A. Organizations. New York: Wiley, 1958.
McKelvey, B. Dimensions of Organization Phenomena. Working Paper, Graduate
School of Management, UCLA, 1973.
- - . Organizational Speciation. Paper presented at the Middle Range Theory
Conference, University of British Columbia, Vancouver, Canada, 1978a.
- - . Organizational Systematics: Taxonomic Lessons from Biology. Working
Paper, Graduate School of Management, UCLA, 1978b (forthcoming in
Management Science).
- - . The Evolution of Organizational Form in Ancient Mesopotamia. Paper pre-
sented to the Management History Division, Academy of Management Annual
Meeting, San Francisco, 1978c.
Merton, R.K. Social Theory and Social Structure (2nd ed.). Glencoe, lll.: Free
Press, 1957.
Osgood, C.E., Suci, G.J., and Tannenbaum, P.H. The Measurement of Meaning.
Urbana: University of lllinois Press, 1957.
Parsons, T. A sociological approach to the theory of organizations. Administra-
tive Science Quarterly, 1956,1,63-85, and 2,225-239.
Popper, K. The Logic of Scientific Discovery. New York: Harper and Row,
1959.
Salancik, G.R. and Pfeffer, J. A social information processing approach. Admin-
istrative Science Quarterly, 1978,23, 224-253.
- - . An examination of need-satisfaction models of job attitudes. Administrative
Schutz, A. Common-sense and scientific interpretation of human action. Philos-
ophy and Phenomenological Research, 1953,14, 1-37.
Schwab, J.J. What do scientists do? Behavioral Science, 1960,5, 1-27.

Journal, 1974,17,242-254.
Sneath, P.H.A. and Soka1, R.R.Numerical Taxonomy. San Francisco: Freeman,
1973.
Wallace, W.L. The Logic of Science in Sociology. Chicago: Aldine and Atherton,
1971.
Weick, K. E. The Social Psychology of Organizing. Reading, Mass.: Addison-
Wesley, 1969.
Woodward, J. Management and Technology. London: HMSO, 1958.
32 A PIECE OF CAKE
A Response to McKelvey
David C. Hayes and Peter J. Frost,
McKelvey's comment on our paper uses cliches similar to those we used in our
title, including ganders and geese, bandwagons leaving stations, cans of worms,
and so on. In responding to McKelvey's remarks, we will resist as much as possi-
ble continuing in the same vein, but if he's the gander and we're the geese,
then-or should that be the other way around?
There appears to be a single "antisubjectivism" thread running through his
comments, despite an avowal of neutrality. Thus, McKelvey questions the sub-
jectivist premise as his first point and then uses that in an attempt to build an
anti particularism argument in almost a reductio ad absurdum style.
However, we would argue that the citation from which he builds his argu-
ments is the basis for his undoing. We repeat it here: Early in the paper we say,
"Consequently, there is no such thing as the organization in a concrete social
entity sense, but rather there are coexisting and frequently competing images of
what the substance of organization is at any point in time." McKelvey has
jumped on the first phrase, skipped over the qualifier, and leaped to the re-
mainder of that sentence. But, the qualifier employed here is critical to interpre-
tation of what we were trying to argue. By arguing that organizations are not
concrete social entities, it is meant that social and behavioral interactions and
relationships as the essence of organization are not observable in the sense that
379
one can actually touch them. Thus, the examples of observed reality given by
McKelvey as analogies (plane and car wrecks) suffer badly. It is the social inter-
action aspect of organization to which we were referring, and perhaps because
we did not make the point clearly enough in the paper, McKelvey has been led
to some inferences we believe are not valid.
For example, such concepts as long-linked assembly lines and organization
charts, which McKelvey refers to as objective elements or organization, are not
social interactions but are touchable, physical things. Such "objects" neverthe-
less are concepts or labels given to phenomena observed, and in the context of
understanding organizations we are left with a "so what" feeling. Why they are
important seems to be the relevant question. There undoubtedly is organization
that can be studied in assembly-line settings, and organization charts and rank
hierarchies may become pertinent to understanding organization as and when
they exist as meaningful concepts in the cause maps of organizational partici-
pants, or at least in the maps of those studying organization. We see little point
in pursuing an understanding of organization as if there were such a thing as the
organization, or as if organization were some objective reality. Consequently,
we reaffirm our assertion in the paper that the path to an understanding of
organization lies in exploring the sense which individuals and coalitions give
to organization (including concepts which McKelvey suggested comprise the
organization) and in the behaviors and interactions that characterize the orga-
nizing process.
With respect to the antiparticularism arguments advanced by McKelvey, we
are not sure whether he is joking or serious in his comments about studying
negotiated cause maps of Boy Scout Troops or hinting at large numbers of maps
for General Motors. He is using the reductio ad absurdum again, and perhaps
the reason for that is that we failed to make clear in our paper what the limits
and intents of our argument were. Let us try to explain once again what we are
getting at.
The argument advanced was that, as individual perceptions of organization
vary, then for organizational action, individuals would need to negotiate com-
mon grounds of beliefs and actions, so forming coalitions. Such competing
coalitions would then seek common grounds for getting things done through a
process of negotiation. We attempted to argue that this negotiating process is a
general phenomenon of organizing and might even follow similar patterns, irre-
spective of the particular groups and individuals involved. We conjectured
that such individuals and groups would behave in ways that are classifiable as
to the degree of routineness and of consensus or agreement involved in these
negotiations.
We argued that the elements or stuff of the cause maps are individual percep-
tions of the particular organization setting -and, the individual's position within
that setting. This we referred to as the content of organization. Thus, we would
A PIECE OF CAKE 381
expect individuals to make differential sense (and thus draw different cause-
effect linkages) about such things as the technology, environment, formal proce-
dures, and so on, dependent upon their perceptions. Again, take the assembly-line
example of McKelvey: Perceptions of the tasks, speed, and so on would differ,
we would argue, between workers and managers in an assembly-line setting. The
former see it as a pain in the butt, controlling their working lives; the latter, as
an efficient means of producing marketable products. (The organizing process
reflects in part the interactions of groups or coalitions attempting to negotiate
shared cause maps.)
Our argument about content, then, is simply that given differential percep-
tions, if you wish to study content, then it can only be studied in a somewhat
particularistic fashion: To generalize via sweeping questionnaire studies across
different organization settings and treat the data as comparable seems to beg the
question of whether or not they are comparable in the first place. Thus, the
argument was made for a midrange approach: Site-by-site study of content to
determine what, if any, the similarities are in sense-making perceptions of the
individuals and groups or coalitions involved. This is what we understand mid-
range theorizing to be about.
On the other hand, as we indicated before, we attempt to make the argument
that if it is the ongoing, dynamic process of organizing that is of concern and
interest, then the types of behaviors involved as people attempt to negotiate
the removal of equivocality (i.e., negotiate the common grounds or cause maps)
are generalizable. Thus an explicit, broad approach across settings is appropriate,
as groups engaging in such processes are studied.
One other potentially misleading aspect of our paper has caused McKelvey to
infer that organizing is just equivocality reduction. That was not our intent.
Rather, we chose to focus on this aspect of the organizing process as a way of
building from Weick's work and illustrating the argument of a general approach
to the process. There may well be a number of other significant aspects to it-
that is, significant in terms of generalizability across negotiating environments;
we simply focused on one as a way to explore and illustrate the argument.
A final point we think worth making is to quote from McKelvey'S comments
on the Cooper and Wolfe paper (Theory Development in Organization Behavior).
It seems to us that he is after all agreeing with our essential argument. Midway
through his remarks, he says, "Goals and purposes are best seen as temporary
enactments that come and go in individual members of a specie during the long-
term competitive struggle of a specie for resources in its interdependent ecologi-
cal network." He seems hoist by his own petard.
Finally, we could not resist the temptation to pursue our cake analogy one
more time and decided to call our response "A Piece of Cake" rather than just
"Marie Antoinette."
VI CONTEMPLATIVE
PANEL DISCUSSION
33 EXPLORING BIG FIERCE
THEORIES
Leslie L. Roos, Jr., University of Manitoba
I came up with a tentative title for these comments: "Exploring Big Fierce
Theories." I plan to relate some of the notions about evolutionary theory that
emerged during the conference both to others' concerns and to my own, which
in Lou Pondy's terms are closer to "exploring" rather than to "map mapping."
I have adapted to my own needs a set of phrases in Cooper and Wolfs paper.
"Everyone else in the room may well be guilty of theorizing by analogy," but
I "sensitively apply insights from all branches of science to the study of organi-
zational behavior." That statement stuck in my mind.
I ftrst wanted to discuss some ofthe factors aiding the construction of better
theory. I was very impressed with the fact that, when Darwin and Wallace
thought up their evolutionary theory, they were in rather strange places. Wallace
was out in the jungle and Darwin was cruising along on the Beagle. Although he
did his best-known writing at home, Darwin's voyages at least served as his
inspiration. Colinvaux's (1978) very readable survey of modern ecology cites
example after example of theoretical advance inspired by work in an unusual
environment.
This suggests some advantages of data-rich environments. We must ask, "What
are the characteristics of an interesting environment?" Much good organizational
work seems to have come about through studying environments that were either
385
very tough for the organization or very easy. Let's look at an organization oper-
ating in a very rich environment and see the fun that its members have, not
holding conferences certainly, but spending money, enjoying fringes, and so on.
A great deal more about the workings of organizational slack could be learned
by such studies. Similarly, growth situations or crisis situations can be fruitful
for investigation.
If we're interested in history or tradition or evolution, the organizational
fields are in a data-poor situation. We have some of Chandler's (1962, 1977)
studies, and Hall's (1976) longitudinal study of the death of the Saturday Even-
ing Post, but such case studies with an explicitly theoretical orientation are
comparatively few. Studying the birth of a new organization, as was done by
Sproull, Weiner, and Wolf (1978), can provide significant insights. As Bigelow
has noted, a chronological stage within a single organization may be an impor-
tant factor.
Things outside our normal personal or professional experiences may aid
theory building. Clearly, super novae and black holes have stimulated astron-
omers to rethink certain aspects of their science. Weick has pioneered in the
use of insights garnered from unconventional organizations-from Apollo
Skylabs (Weick, 1977) to jazz orchestras (Bougon, Weick, and Binkhorst, 1977).
Better theories might result from concentrating our efforts in such situations;
much interesting research in the natural sciences has been stimulated by unusual
experiences.
How might some of the developments in evolutionary theory stimulate or-
ganizational research? Biologists are now studying the many ways in which the
highlighting of certain characteristics can act to minimize competition by help-
ing to establish different ecological niches for each species. For example, Colin-
vaux notes that, when different bird species are geographically separated, the
intra-species variation may be considerable. Many individuals belonging to dif-
ferent bird species of the same genus may look quite similar when they are far
apart geographically. This is true for two species of nuthatches-one in Greece
and Asia Minor, the other in central Asia. The range of the two species overlaps
in Iran, but in this zone of overlap the species can be easily distinguished by
their markings. Colinvaux reports a number of studies emphasizing that certain
behavioral or physical characteristics of a species become clear in an environ-
ment that would otherwise be highly competitive.
Some hypotheses for organizational research are suggested, even though
defining niches and identifying species is difficult for both biologists and or-
ganization theorists. McKelvey makes a laudable first effort at organizational
taxonomy. If different species of organizations might look or act differently in
order to avoid head-on competition, this has implications for organizational
strategy. Successful companies tend to have the dominant market share in one
EXPLORING BIG FIERCE THEORIES 387
or more fields (Buzzell, Gale, and Sultan, 1975). If the market share is being
equally divided among four or five firms, anyone company finds it more dif-
ficult to be terribly successful (Hamermesh, Anderson, and Harris, 1978). Thus,
evolutionary theory may be helpful beyond taxonomy in suggesting some
dynamic propositions to help us understand the way public and private organi-
zations behave.
We might also build theory by looking at unoccupied niches; such niches may
occur because of a natural or manmade catastrophe. In fact, the biologists have
actually run some experiments here. Colinvaux cites research by the "Maine
Gunners," scientists hired to go into the woods to study a caterpillar pest, the
spruce budworm. Because spruce warblers raise their broods almost entirely
on budworms, the Maine Gunners wanted to determine the extent to which
the birds were restraining the pests. They chose two stretches of forest to have
one experimental and one control area. They decided to count, then remove, all
the birds in the experimental area. When they started shooting, it became neces-
sary to kill almost twice as many birds as were originally nesting in the selected
area. Under normal conditions a surplus floating population of nonbreeding
birds was unable to get niches and territories.
Analogs in the organizational world include such dramatic, catastrophic
events as complete defeats in war. For example, theory-relevant longitudinal
case studies of the reemergence of German and Japanese industry after World
War II would be most useful. Can we identify particular species of organizations
able to take advantage of the situation? Which kinds of organizations are partic-
ularly adaptive for this unoccupied niche situation?
Questions of specialization are relevant here. Within biology specialization
exists in such ways as one-to-one relationships between predator and prey
species. What happens when the environment changes such that the prey is no
longer there? The predator'S niche is altered markedly. The biological question
is: Which species can handle this and change their source of food? Which species
can't? Can we unravel how and why? Organizational parallels are clear. Firms,
such as defense contractors, may get so specialized that when the environment
changes, they don't know what to do. Some organizations can deal with this flux
in the environment, some can not. Bigelow has noted how military organizations
exist in one of two niches-peacetime or wartime. Interesting organizational
research on specialization and adaptation could be done.
Another evolutionary theme concerns how animal species handle problems
of variation, succession, and reproduction. A long lead time is needed to produce
a great white shark or a tiger. In the same way the kind of high technology nec-
essary to develop a large, new computer leads to a long incubation period. Other
biological strategies to facilitate reproduction are based on numbers and wide
dispersal, as gardeners fighting dandelions well know. Perhaps a better under-
standing of these biological approaches would help marketers working with

consumer products being sold across several different sorts of geographical areas
or market segments.
What are the problems with these kinds of formulations? I know that not
everyone has been taken with evolutionary theory. Very little organizational
data has been collected with any sort of evolutionary theory in mind. History of
science descriptions, such as that of Gillespie (1960), highlight how biology has
benefitted from its data base; comparative anatomy, biochemistry, embryology,
and paleontology all provide valuable clues. Biology is really data-rich now,
although it was not always that way. Much of the theory has stimulated the
search for data. But I also think biologists had more relevant data to start with
than do the organizational sciences.
Convergent evolution poses difficulties for taxonomizing because it raises
such questions as: What features do we study? Which are really critical? Which
have come about later and are not fundamental to the development process?
These questions present obvious problems for biology and have been mentioned
in the papers here. This question of what organizational researchers should be
studying is not easily resolved.
How should we react when something does not support our theory? If we
adopt McKelvey's core technology concept as the characteristic of interest, and
things don't seem to go correctly, several questions about the role of discon-
firming evidence arise. If the concepts are developed enough to permit falsi-
fiable predictions, and the predictions are wrong, what then? Is evolutionary
theory still useful, but the key characteristic chosen incorrectly or operation-
alized inappropriately? Or is this really some kind of fantasyland and the whole
approach should be abandoned?
Scholars are likely to disagree about this, just as we've seen arguments about
the utility of exchange theory, the utility of organizing processes, and so forth.
Gilfillan's paper suggests that, when operationalizations are difficult or when
the results are fuzzy, relating back to the larger theoretical scheme will be diffi-
cult. This is just a basic dilemma that we're going to have to live with, at least for
the time being.
At this point, I wanted to mention some data-rich situations with relatively
unexploited implications for organizational theory. At least two substantive
social science research areas are presenting a rather disheartening picture of the
world. In both areas a lot of intellectual work and data collection has been ac-
complished. People think they understand what the inputs are, what the outputs
are, and what is going on. Yet discouraging results have been obtained.
One area is that of modern portfolio theory, and here I think that we should
talk to our colleagues in finance. Modem portfolio theory suggests that for a
portfolio manager meeting some criteria of diversification, consistently beating
the averages, in both up and down markets, is very hard. Students of fmance
(Sharpe, 1978) say there is a great deal of information in the system; the market
is nearly efficient.
One of the relevant research questions is, "Are there any organizations that
manage money, are there any people, who are doing better than the others? Can
they articulate the reasons why they are superior?" Some are doing worse, but
doing better seems to be very difficult. If a few organizations do do better, we
should be studying these. If you can find them, you can make some money
yourself. Moreover, the study of such organizations should be tremendously
stimulating for the development of better organizational theory. The policy
relevance and academic interest in understanding excellence should be clear.
The same parallel with other variables, a few different concepts, exists in
what one might call "the dilemma of modern education" -education theory.
Coleman et al. (1966) and Jencks (1972) suggest that, within the "normal"
range of variables related to resource allocation and controlling for recruit-
ment, differences among schools are associated with only small differences
in student outcomes. Although more recent research (Wiley, 1976) has for-
warded such promising new variables as "quantity of schooling," additional
work must be done to establish a more optimistic approach.
One question to be asked is: "Can we identify those who consistently do
better or who consistently do poorer?" In a data-rich environment these de-
viant cases can be spotted more easily than would be possible without exten-
sive, well-organized data. Some progress has been made in developing and
applying techniques for isolating such cases (Klitgaard and Hall, 1977).
We might focus on the specific factors, both subjective and objective, which
might influence why a given organization is more or less effective. Often, leader-
ship is going to be critical; saying this is not terribly useful theoretically. How-
ever, other factors in the organizational environment undoubtedly interact with
the leaders' abilities and would permit more interesting formulations. Thus,
Sowell's (l976) research on excellence in black schools highlighted community
characteristics that supported the academic efforts of successful principals.
Often, we will be able to elaborate on particular aspects of the process, on
some dynamics of resource allocation that make a given firm more successful.
Perhaps the organization has particular strategies for getting resources from
the environment; some of the resource acquisition devices used by public sector
organizations are particularly interesting. Such research might help on the pro-
cess side, moving beyond mere taxonomy. This would ask: ''What are the
characteristics of effective organizations in different environments? What is
generalizable and what is not?"
In clOSing, I would like to tell about something that links to evolutionary
theory. Not only did I enjoy reading this in Colinvaux, but the theme relates to
teaching in a business school and having an applied, managerial focus. Some of

you may know about the Banff School of Advanced Management which involves
U.B.C. and the other western Canadian universities. The Banff School has taken
as its insignia, Tyrannosaurus Rex, partly because a number of fossils have
been found in Alberta, not too far from where the school is located. This huge
beast is looking out, and the motto is "Adapt or Perish." You see the classic
version of Tyrannosaurus Rex and you know: He's obviously aggressive; he's on
the fast track; he's a sort of hard-charging individual of considerable bulk.
However, both recent fossil evidence and new reconstructions suggest that
Tyrannosaurus Rex lay down for about twenty-three hours out of twenty-
four, conserving its energy and preventing heat loss from this huge bulk. The
dinosaur resembled a very slow-moving truck. Occasionally it staggered to its
feed to waddle around, looking for a sick or dying, large, slow, planteater that
it could pounce upon. Given the new interest in healthy lifestyles, I'm not sure
we should feel superior. Tyrannosaurus Rex was a well-adapted individual who
avoided cardiac problems and had considerable leisure time. Tyrannosaurus Rex
lasted several million years; the evidence is not yet in as to whether we, and our
theories, will do as well.
REFERENCES
Bougon, M., Weick, K., and Binkhorst, D. Cognition in organizations: An analy-

sis of the Utrecht Jazz Orchestra. Administrative Science Quarterly, 1977,
22 (4), 606-639.
Buzzell, R.D., Gale, B.T., and Sultan, R.G.M. Market share-a key to profitabil-
ity. Harvard Business Review, 1975,530),97-107.
Chandler, A.D. Strategy and Structure. Cambridge, Mass.: M.LT. Press, 1962.
- - . The Visible Hand. Cambridge, Mass.: Belknap Press of Harvard University
Press, 1977.
Coleman, J.S., Campbell, E.Q., Hobson, C.J., McPartland, J., Mood, A.M.,
Weinfeld, F.D., and York, R.L. Equality of Educational Opportunity. Wash-
ington, D.C.: Office of Education, U.S. Department of Health, Education,
and Welfare, U.S. Government Printing Office, 1966.
Colinvaux, P. Why Big Fierce Animals are Rare-An Ecologist's Perspective.
Princeton, New Jersey: Princeton University Press, 1978.
Gillespie, C.C. The Edge of Objectivity. Princeton, New Jersey: Princeton Uni-
versity Press, 1960.
Hall, R. I. A system pathology of an organization: the rise and fall of the old
Saturday Evening Post. Administrative Science Quarterly, 1976, 21 (2),
185-211.
Hamermesh, R.G., Anderson, M.J., Jr., and Harris, J.E. Strategies for low
market share businesses. Harvard Business Review, 1978,56 (3), 95-102.
Jencks, C. Inequality: A Reassessment of the Effect of Family and Schooling in

America. New York: Basic Books, 1972.
Klitgaard, R.E., and Hall, G.R. A statistical search for unusually effective
schools. In William B. Fairley and Frederick Mosteller (Eds.), Statistics and
Public Policy. Reading, Mass.: Addison-Wesley, 1977, pp. 51-86.
Sharpe, W. F. Investments. Englewood Cliffs, New Jersey: Prentice-Hall, 1978.
Sowell, T. Patterns of black excellence. The Public Interest, 1976,43, 26-58.
Sproull, L., Weiner, S., and Wolf, D. Organizing an Anarchy: Belief, Bureaucracy,
and Politics in a New Federal Agency. Chicago, Ill.: University of Chicago
Press, 1978.
Weick, K.E. Organization design: Organizations as self-designing systems. Organi-
zational Dynamics, 1977, 6 (2) 30-46.
Wiley, D.E. Another hour, another day: Quantity of schooling, a potent path for
policy. In W.H. Sewell, R.M. Hauser, and D.L. Featherman (Eds.), Schooling
and Achievement in American Society. New York: Academic Press, 1976, pp.
225-265.
34 MIDDLE RANGE THEMES
IN ORGANIZATIONAL THEORIZING*
Karl E. Weick, Cornell University
The common cold continues to defy understanding. Attempts to theorize about

it falter in much the same way that theorizing about organizations falters. It is
instructive to examine an exhibit of theorizing about the common cold to learn
something about theorizing about the common organization.
'What about these injections that are supposed to prevent colds?' I asked. 'I
see by this medical paper that some chimpanzees that have been given the
injections have colds just the same and some don't.' 'They have not been
entirely successful,' the doctor said. 'They are designed to attack the causes
of secondary and tertiary infections that follow the initial infection or irrita-
tion known as the common cold. In many cases they seem to make the cold
milder for this reason. The idea is that, whatever causes the initial inflamma-
tion, it weakens the general resistance of the mucous membrane, and the
pneumococcus, various kinds of streptococcus, and other known germs or
viruses rush right in. The injections cause a purely chemical reaction, which
for some reason, tends to counteract the general irritation or infection. Like
codeine. Codeine really seems to have some specific effect on a cold, but it
*Support for this work was provided in part by the National Science Foundation through
Grant BNS 75-09864.
392
MIDDLE RANGE THEMES IN ORGANIZATIONAL THEORIZING 393
doesn't always have it, and when it does have it, we don't know why it does.
There's another preparation made from a cold vaccine that is taken internally,
like a pill. 1 think it was forty percent of a group of people who reported a
great improvement after taking this, but forty percent of another group who
were given pills containing nothing at all also reported great improvement.
One reason we are beginning to think a cold isn't caused by a germ is that a
cold, unlike most infectious diseases, seems to develop no immunity. You
can have one cold on top of another, as you know. We're up a tree, as 1
said before.'
'This business of the common cold not being caused by a germ at all sort
of strikes at the foundations of American civilization, doesn't it?'
'Don't get me wrong,' the doctor said. 'The no-germ theory hasn't been
proved, any more than the germ theory. The thing is we just don't know.
There's a great deal to be said for the germ theory, although it seems right
now that there is more to be said for the no-germ theory. And anyway, influ-
enza is a contagious disease definitely caused by a germ, so the precautions
people usually take to keep from giving other people their colds are just as
well. The funny thing about influenza is that the turbinates don't seem to be
particularly affected by it. You don't get the same acute sniffling or stuffed-
up feeling with influenza that you do with the common, grippy cold, yet with
influenza you are a whole lot sicker than you are with a common cold.'
'I guess 1 ought to get out of here tomorrow,' 1 said.
'No reason why you shouldn't,' he said. 'You've had no fever for two days
and there doesn't seem to be any serious bronchial infection. You'll cough a
little for a day or two, and then you ought to be O.K.' The doctor sneezed
suddenly.
'I hope 1 haven't given you my cold,' 1 said.
'Y ou may have,' he said. 'If there are common cold germs, 1 have doubtless
picked some of them up from you. On the other hand, it is probably just the
turbinates seeing ghosts. Of course, a piece of dust or something may have got
up my nose just then and made me sneeze, the sneeze being a reflex action
designed by nature to clear the nose of any irritating substance. But this re-
flex action isn't absolutely dependable either, any more than the turbinates
are. 1 suspect that what happened just now was that the turbinates, for some
reason known only to themselves, got excited and worked up an overabun-
dant amount of moisture; this moisture, trickling through my nose, fooled
my sensory nervous system into thinking that there was some irritating sub-
stance in there and caused me to sneeze. Sneezing, incidentally, isn't such a
good method of ridding the nose of an irritating substance because sneezing
itself irritates the nose.'
'And perhaps causes or contributes to an inflammation of the mucous mem-
brane of the upper respiratory tract?'
'Exactly,' he said (St. Clair McKelway, in Siu, 1968, pp. 198-200).
Understanding of the common cold takes some unusual twists in this excerpt
and these twists mirror themes of the middle range in organizational theory. At
the very end of the explanation, the possibility arises that sneezing causes colds
rather than colds cause sneezing. Thus, the direction of causal arrows in the com-
mon cold is elusive just as it is in common organizations. People become dog-
matic about the direction of causation and about the abundance of unilateral
causation, at their own peril. Middle range theory can serve as corrective that
redirects causal arrows, adds causal arrows where none were thought to exist,
and questions the presence of arrows presumed to exist.
A supposedly robust theory of the common cold, the germ theory, plays to
very mixed reviews in this excerpt. It both looks as if germs are the cause of the
common cold and as if they're not. Similarly, most theories in organizational
research such as resource dependence look at times as if they're major themes
and at other times as if they're not.
Much of the agonized theorizing about colds and their origin may boil down
to something like a random origin to the process. Just as is true with organiza-
tions, full-blown colds may develop from inconsequential beginnings. The modest
scope of middle range theories may be temporary and hold only for the initial
stages of a process. Given a modest phenomenon the components of which are
tied together in a deviation amplifying causal loop, the result of several cyclings
through this circuit may produce a dominant phenomenon that encompasses the
field and drives out other possibilities.
The germ theory directs much of the physician's seeing as a patient is ex-
amined, and the same holds true for organizational theory. Believing is seeing
and the belief that the world is made up of middle range phenomena that are
loosely coupled to one another can produce the measurements and observations
that support the existence of such a world.
The large variety of possible causes for the common cold that are described
in the preceding incident bears a close resemblance to the large variety of possible
causes for outcomes such as organizational effectiveness. It is the very abun-
dance of these variables and the possibility that in various combinations they are
crucial, that generates middle range theories that highlight subsets of them.
Theorizing about the common cold has a decidedly piecemeal and contradic-
tory quality. It's not much different with organizations. The value of this con-
ference has been that it provides a pretext for us to reflect on both the content
and the form of our arguing. The papers provide a splendid opportunity to re-
examine how we feel about theorizing and about the content of the theories.
The papers as a set provide considerable new information and do much of
Merton's homework for him. I have no doubt that we know more about middle
range theories and some of the traps in that concept now than was true at the
beginning of this conference.
In these remarks I want to suggest a variety of images that can be used to
think through the implications of middle range theory. My intentions here are
several. As much as possible I want to link the topic of middle range theory with
other themes that have been present in organizational theory for some time. I
want to suggest a set of images that do not simply duplicate those already avail-
able in the papers in this volume.
IMAGES OF MIDDLE RANGE THEORY
The Explorer Image
During this conference Lou Pondy invited us to sort ourselves into explorers and
map makers. Various other people have suggested that the participants at this
conference sort themselves into idealists and realists, essentialists and evolution-
ists, and those who think teleologically and those who think nonteleologically.
At first, it may seem surprising that issues of labelling and identification should
surface at a conference that supposedly is about as innocuous a topic as middle
range theory. Considering, however, that the topic of middle range theory is
ambiguous, then you would expect that people would try to develop a social
definition of reality and in doing so, would seek out similar others. In answer
to the question, what's going on here, potential idealists spot and talk to other
potential idealists, and the same holds true for the realists. As a consequence,
novel definitions develop as to what this conference "really" is about.
That's innocent enough as long as the labels aren't taken too seriously. Gonos
(1977), for example, argues that what looks like an idealist perspective in fact
isn't. McKelvey argues that people are either essentialists or evolutionists, but
doesn't leave room for the possibility that both perspectives can be accomodated
in a single formulation. It can be asserted, for example, that interlocked be-
haviors are a fundamental unit of analysis out of which all organizing grows,
and that these essences change over time. It could be argued that loosely coupled
interlockings are going to survive for a longer period of time than tightly coupled
interlockings, because loosely coupled interlockings can adapt to a greater vari-
ety of circumstances. Similarly, you could argue that the prototypic action in
organizations is enactment but make the further assertion that some forms of
enactment survive longer than other ones.
When Pondy asks us to sort ourselves into explorers or map makers, he for-
gets that man is that animal who draws lines that he then stumbles over. Pondy
is interested in circles of inquiry. When forcing people into either explorers or
map makers he momentarily loses sight of the fact that most people are both.
Exploration affects map making, but the maps in the hands of an individual
affect subsequent exploring, a circle of inquiry that has been described recently
by Neisser (1976). Furthermore, it's not obvious that exploring and map making
exhaust the possibilities. There are map readers, people who invent the idea of a
map, people who create what explorers explore, there are people who discover
they are explorers a posteriori, and in short, the dichotomy loses much of the
richness that is preserved in several of the other distinctions raised by Pondy.
The Leavitt Image
One of the major summaries of the field of organizational theory is Leavitt's

(1965) diagram summarized below.
/ Structure
~
Task
· 1 / Technology
"- People (Actors)
Middle range theory could be operationalized as the number of arrows present

in the Leavitt diagram that are activated by an explanation. Small numbers of
arrows represent empirical generalizations, two or more arrows represent middle
range theories, and any theory that simultaneously talks about the relationships
among four or more arrows tends toward a grand theory. The numbers in this
illustration are purely arbitrary, the principle is not. Much of what these papers
describe are potential interrelations among different portions of the Leavitt
diagram. If we simply treat the Leavitt diagram as being explored more or less
fully, we can use this description of content to describe theorizing that is of
smaller or greater scope.
The Link Pin Image
Organizational theorists operate at a variety of levels that are variously described

as macro-micro, individual-group, psychological-sociological, cell-society. Despite
pleas that more than one level be incorporated into a single study, this seldom
occurs. Part of the optimism associated with a resurgence of interest in middle
range theory is that such theory could serve to provide the means by which
individual and structural or societal variables might be incorporated into a
single study.
For example, suicide is an individual activity that is variously coupled with

societal differences. Recently Phillips (1977) has extended this linkage by sug-
gesting that there appears to be a sharp rise in motor vehicle fatalities in the
week immediately following publicity given to a suicide. He proposed that this
link between fluctuations in motor vehicle fatalities and publicized suicides
occurred because publicity stimulates a wave of imitative suicides, and some of
these imitative suicides are disguised and recorded as motor vehicle accidents.
Phillips' explanation is a middle range theory that links individual phenomena
(suicide) with social phenomena (imitation, pUblicity). While readers may take
issue with the particulars of this example, the important point is those explana-
tions which couple two or more levels of analysis appear to be middle range
theories. They assume value because they relate different units of analysis or
different preferences for form and scope of explanation.
The Fish-Scale Image
Campbell (1969) has developed the fish-scale model of omniscience. He argues

that present disciplines such as psychology and sociology consist of clusters of
specialties with large gaps between these clusters that are unfilled by any kind
of work. He invokes the image of fish scales as a possible way to structure
science and override the ethnocentrism of disciplines. The pictorial image of fish
scales depicts a situation where there is collective comprehensiveness across large
areas, which is accomplished through overlapping patterns of unique narrowness.
Each narrow specialty is a fish scale. It is the continuous texture of narrow spe-
cialties overlapping with other narrow specialties that produces comprehensive
understanding and fills the gaps between the somewhat arbitrary clusters of
specialties that now are thought to comprise disciplines.
A novel, narrow, deep specialty such as is associated with a fish scale seems to
be virtually an ideal approximation of the kind of image implied by middle range
theories. The emphasis on novel, narrow, and deep specialties that lie adjacent to
other unique combinations suggests the potential benefits that can accrue when
people actively pursue middle range theories. The trick seems to be that people
ensure that there is modest redundancy among the topics addressed by these
theories. The greater the redundancy, the greater the likelihood that we simply
duplicate the arbitrary clusters and nonsensical gaps in understanding now asso-
ciated with social science.
People worry that intentional pursuit of fish scales heightens the fragmenta-
tion of inquiry rather than helping to reduce it. The futility of this worry lies
in the fact that fish scales produce collective comprehensiveness. It will never
be the case that any single individual will represent and encompass a field. In-
stead, it is the combination of variable ignorance and knowledge among scholars
that collectively comprises what is known about an area. The understanding is
collective and the full scope of the understanding is found only by quizzing all
of the components.
Ironically, fields in social science fragment in the interest of unity. Increased
fragmentation fills gaps and links specialties that previously had been uncon-
nected. It is this fIlling of gaps that promises to unify a field even as the immedi-
ate activity involves fragmentation. Middle range theories, as they have been
described in the preceding papers, hold some promise for fragmenting the field
in the way that will induce fish-scale unity.
The Commensurate Complexity Image
Warren Thorngate (1976) has argued that it is impossible for a scientific explana-
tion to be simultaneously general, accurate, and simple. Any explanation that
meets two of these criteria will violate the third. I have found it useful to exam-
ine the nuances of this assertion by representing its three components in terms
of three positions on a clock face with General represented at twelve o'clock,
Accurate at four o'clock, and Simple at eight o'clock. An explanation that
is simultaneously general and accurate (it would fall at the two o'clock position)
is as far from simplicity as it is possible to get. In accomplishing generality
and accuracy in an explanation an investigator has paid the price of simplicity
and runs the risk that his or her work will be criticized as impenetrable, inac-
cessible, and obscure. Other tradeoffs represented by strategies of inquiry
at six o'clock or at ten o'clock create similar patterns of assets and liabilities.
General
12
Simple Accurate
With the aid of this representation we can describe middle range theory in
two different ways. First, it can be argued that any movement of an explanation
away from twelve o'clock is a movement toward middle range theory. Any
explanation that falls at six o'clock, a combination of simplicity and accuracy, is
the empirical generalization referred to by Evered and Bluedorn. Viewed in this
way, middle range theories will be criticized either because they are excessively
complex (the investigator has moved away from the twelve o'clock position in a
clockwise direction) or because they are inaccurate as explanations of specific
problems (an indictment that is produced by counterclockwise movement away
from twelve o'clock). A tacit prescription implied by this is that the investigator
who moves away from twelve o'clock simultaneously toward both four and eight
o'clock is in a stronger position than an investigator who moves in only one of
the two directions. It might also be argued that an investigator who moves com-
pletely around the clockface will have produced a more valuable contribution to
the field than a person who covers a smaller portion of the clockface.
If we modify thee two-dimensional character of the clockface and make
it three-dimensional, we can see a second way to pose the issue of middle
range theory.
Grand Theory
Middle range theory
Empirical
Generalizations
399
Viewed in this manner, empirical generalizations consist of explanations that

have either relatively pure generality, accuracy, or simplicity, or compromise
any two of these characteristics while ignoring the third. As we move in the third
dimension, however, we see that it becomes increasingly possible to accommo-
date all three characteristics in an explanation. Partial accommodation comprises
middle range theory and Grand theory involves explanations that fully accommo-
date all three. Viewed two-dimensionally, grand theory is a logical impossibility.
Viewed three-dimensionally, it remains a significant undertaking, but it can
be accomplished.
The Stable Subassembly Image
One of the irresistible parables in organization theory is the parable of Hora

and Tempus and one of the irresistible temptations is to link the issue of middle
range theory with these two classic actors. Tempus and Hora are watch makers
who build thousand-piece watches. Tempus' watches have no subassemblies,
meaning that all one thousand parts have to be assembled for the watch to hold
together. If Tempus is interrupted before all thousand pieces are in place, the
watch falls apart and he has to start over. Hora's watches contain subassemblies
of ten parts each and ten sets of ten subassemblies each in a nested arrangement.
Every time Hora puts together ten pieces he has a stable unit and the effects of
interruptions on his work are much more localized.
If we suppose that each watch maker is interrupted randomly with a fre-
quency of once every one hundred parts, and that work in progress is lost at
each interruption, the astounding outcome is that Hora will complete four
thousand times as many watches as will Tempus. The importance of inter-
mediate, stable stages in any evolutionary sequence is pointedly illustrated by
these calculations.
Pondy (l977) has produced a provocative extension of this parable that sets
the stage for middle range theory. Pondy places these watch makers into two
different environments, a placid low interruption environment and a turbulent
high interruption environment. Having done this he argues as follows. "One
would expect evolutionary pressures of the turbulent environment to produce
watch makers like Rora. But in the placid environment, Tempus would continue
to produce undifferentiated watches without internal structure, unpressed
by environmental interruption to begin producing hierarchically structured
watches. . . . Low power managers in high interruption environments will nec-
essarily produce highly differentiated work-small projects of short duration.
High power managers with the power to buffer interruptions or managers in
placid environments will do work characterized by larger, longer, less differen-
tiated projects. Conversely, when it is necessary to produce large, richly inter-

connected, undifferentiated products, for example, a social movement, a novel,
or a painting, the utility of power is great and one would expect managers
or other actors to seek power to control interruptions. Perhaps that is why
prophets, novelists, and artists seem to demand so much control over their
environments.
As an interesting twist on this theory of interruptions, consider private
secretaries. The facu1ty at the New York State School of Industrial and Labor
Relations, Cornell University, have private or semiprivate secretaries to screen
their incoming calls and visitors, but the faculty at the University of lllinois
Business School do not; telephones ring directly in their offices and there is
nothing between them and hall but a door. The ... analysis would predict
that papers written at Cornell had fewer section headings and longer paragraphs
than those written at lllinois. That is an eminently testable conjecture" (1977,
pp.67-68).
Middle range theories such as exchange theory are the stable subassemblies
of social science. They prevent the discipline from disappearing when data dis-
confirm more venturesome, risky theories. It is important to cu1tivate these
building blocks because, as Simon's example shows, they hasten the speed with
which evolution occurs.
Pondy's specu1ations suggest that outright development of grand theories
without stable subassemblies requires a situation of low interruption and high
power. Grand theories shou1d be developed only by tenured people who are
able to buffer themselves against their environment. Middle range theories
should be produced by those who experience higher interruptions and do so
from a position of lesser power. One of the valuable features of Pondy's analysis
is that it lets us think about the alternation between grand theory and empirical
generalizations over time in terms of the capabilities of investigators to ward off
interruptions and in terms of the discontinuities present in the objects of the
theorizing (Moore, 1964).
The Falsifiablility Image
My own encounter with middle range theorizing is confined to a paper in 1974

showing a form of confounding that can occur in these theories. This confound-
ing is shown below.
My concern was with the sweeping claims made for general systems theories,
cybernetics, and information theory (Handy and Harwood, 1973). Many of
these theories were presented as if they were fmdings and conclusions rather
than possibilities. There seemed to be no way in which the theories could be
Quality of Propositions
Scope of Theory Falsifiable Not-falsifiable
Grand theory Miller, 1978 Jantsch, 1975

(Weick, 1974)
Middle range theory Mulder, 1977 Klapp, 1978
modified, improved, amended, tied to specific circumstances, examined, or

treated as anything but philosophical exercises. The attempt in that paper was to
take system notions such as equilibrium and transform them into more modest
sets of propositions that could be both tested individually and applied selectively
to limited sets of organizations. My concern was with falsifiability, not scope. In
terms of the preceding table, it can be seen that my concern with middle range
theories ignored three crucial combinations (e.g., grand theories that are falsifi-
able, middle range theories that are falsifiable, etc.).
If a theory of any scope can be shown to be incorrect or limited, then we
are in a stronger position to improve organizational scholarship than if many
of our propositions are refractory to disproof. In the papers presented at this
conference, those that dealt with issues of falsifiability and nonfalsifiability
seemed more effective than those that argued solely for greater or lesser scope
of content.
Understanding however, is not just an exercise in falsification. Phil Bobko
has reminded us of some other options that we have available. Much of my own
writing has proposed alternatives to falsification. The issue of falsification, how-
ever, takes on added importance because the rarified quality of grand theories
and the abundance of abstractions in them makes it difficult to tie them mean-
ingfully to particular problems.
My naive bet is that it is harder to produce a nonfalsifiable middle range
theory than it is to produce a non falsifiable grand theory. If we confme our
attention, therefore, to theories of the middle range, then it becomes more
possible that we routinely will work on a series of self-correcting, feedback-
modified hypotheses that synthesize types of data. Middle range theories, be-
cause of their propensity toward falsifiability, improve the likelihood that
disciplines will be self-correcting. And that's one ofthe benefits associated with
these formulations.
The Amplifying Loop Image
Phenomena that occur on a modest scale can have massive consequences if their
components are tied together by means of deviation amplifying causal loops.
The importance of this point is that the phenomena addressed by middle range
theory are not necessarily phenomena of middling importance. In fact, I'd argue
that middle range phenomena and deviation amplifying feedback loops may be
all you need to produce many of the major phenomena that we worry about.
Any chance event such as a small crack in a rock that gathers water or a far-
mer who builds his home on a homogeneous agricultural plain can become
amplified into a tree that grows out of a rock or people congregating around the
original farmer to form a city. It is the process that generates the complex out-
come, not the placement of the initial crack or the initial settlement (Maruyama,
1963). With sufficient cycling, small deviations can be amplified either into
complex homogeneous events or into a virtual disappearance of the relevant
elements. It all depends on the initial kick that triggers the deviation amplify-
ing process.
Consider the simple deviation amplifying loop shown in the following figure.
self-evaluation
coping ability +---depression

\-
This is a middle range theory of depression that says basically, depression is a
function of self-evaluation which, in turn, is a function of one's ability to cope
with the world. When this modest set of determinants is tied together in the
form of a deviation amplifying loop (an even number of negative causal relation-
ships is the criterion for a deviation amplifying feedback loop, and in this case,
there are two negative relationships), it becomes possible for the individual
either to develop a highly invigorated and energetic lifestyle or a lifestyle that
basically grinds to a halt. It all depends on the way in which these variables
start to move.
Assume that there is a slight increase in coping ability. This increases self-
evaluation, which decreases depression, which increases coping ability, which
increases self-evaluation, which further decreases depression, which further in-
creases coping ability, and so forth. Greater and greater efficacy in dealing with
the world is the outcome of these small beginnings that are amplified because
of the causal circuit. Assume, however, that coping ability drops suddenly, for
whatever reason be it fatigue, inexperience, bad luck, or whatever. A decrease
in coping ability decreases self-evaluation, which raises depression, which further
lowers coping ability and self-evaluation, and the individual is bound into a
vicious circle of despair.
The scope of these concepts remains modest until they become coupled in
a deviation amplifying manner. This coupling provides the mechanism whereby a
small portion of the world comes to dominate. My argument is that there is
a certain relativity in the phenomena encompassed by middle range theory for

just this reason. Middle range phenomena can swamp other phenomena and
account for most of the variance because of the way their components are
coupled together. When one talks about the generality of middle range theories
it's important to take account of just this capability.
To see how a middle range phenomenon can remain a middle range phe-
nomenon suppose that the small theory of depression that we are working with
undergoes a minor change as in the following diagram.
self-evaluation
-!
coping ability - depression
We now see that coping ability is tied negatively to self-evaluation, which says
that the greater my coping ability the lower is my self-evaluation. I may like to
see myself as someone who is unable to cope with the world and who needs to
be taken care of. In such a world, characterized by an odd number of negative
signs, the variables associated with depression now constitute a deviation coun-
teracting loop. There is no chance whatsoever that a phenomenon tied together
in this manner will dominate the world.
Suppose again that my coping ability goes up for some reason. This decreases
my self-evaluation, which increases depression, which decreases coping ability,
which now increases my self-evaluation, which now decreases my self-evaluation
again. With an odd number of negative signs self-evaluation fluctuates around a
central value as do depression and coping ability. There is no chance that from
these small beginnings a massive depression or a massive manic condition will
emerge. Instead, the changes are controlled.
CONCERNS ABOUT MIDDLE RANGE THEORIZING
Middle range theorizing is not without problems and several of these are evident
in the preceding papers. I want to focus briefly on three issues that supplement
those raised in preceding papers; disconfumed assumptions, playfulness, and
auxiliary theories.
Davis (1971) has argued that theories that violate weakly held assumptions
are typically perceived as interesting. A tacit quality of middle range theories
is that they are often eclectic and often affirm rather than violate assumptions.
That being the case, theories that affirm assumptions are likely to be regarded
as obvious. If a theory affirms rather than violates assumptions, then it is likely

to be neglected and this carries the danger that middle range theories will not
receive the degree of refmement and improvement that will be true of theories
found at other ranges. A middle range theory that violates no assumptions may
be dismissed, and this reduces the likelihood that it can be joined with other
middle range theories to form cumulative understanding.
Bobko's paper, more than any other paper presented, invited people to be
playful and diverse in the images they entertain. That's a stimulating recom-
mendation, but when people are urged to take things lightly they have to be
careful how they ope rationalize that. One of the things that's true about com-
plex organizations is that almost anything asserted about them can contain
some truth. For that reason, Bobko's invitation to be playful and to think dif-
ferently makes sense. Once one admits the world is complex and difficult to
understand and that most explanations are true for someone somewhere, that
realization can lead either to a feeling of humility or to a feeling of arrogance
because anything goes and any idea is just as good as any other idea. The latter
reaction to complexity seems much more destructive than the former.
A further complication is that people may adopt the "let a million flowers
bloom" philosophy for one of two reasons. They may advocate that position
because they believe there's no way you can tell the good theories from the bad
theories or they may advocate that position in the belief that the proliferation of
theories aids the emergence of good ones. Needless to say, those two positions
are contradictory and those who urge proliferation should be explicit about
which assumption they make.
Finally, advising scholars to be playful and to generate many theories may
seem ludicrous given the fact that most people already have more ideas than
they know what to do with. What that observation misses is the fact that as
newer ideas are developed and articulated, older ones change. They may disap-
pear, become absorbed into newer ones, differentiate, form into new combina-
tions, or seem even more important than was originally thought. Each new idea
is not simply an addition to an already lengthy queue of problems awaiting
attention. Chronic idea generation chronically reshuffles the content and mean-
ing of ideas already generated. Newer ideas edit prior ideas and this is one of the
strongest arguments in favor of rugged playfulness, rugged in the sense that it is
enduring and done with dogged determinism.
Morrow and Muchinsky's observation that middle range theories work only if
they are supplemented by auxiliary theories, is an important one to consider. If
middle range theorizing is a workable strategy only when it is supplemented,
then it could well be the case that the supplements rather than the middle range
theory, are largely responsible for any explanatory power. It might also be the
case that a figure ground reversal becomes necessary in which the potent auxil-
iary theory becomes the middle range theory and the middle range theory now
merely supplements. Finally, if middle range theory works only if there is
auxiliary theory, then this could mean that middle range theorizing is middle
range only in name. If one combines both the middle and auxiliary theory then
what one is really talking about is yet another grand theory. Thus, it may turn
out to be the case that theorizing does not partake of quite as many gradations
as we have presumed during this conference.
CONCLUSION
These papers have provided a superb pretext for all of us to reflect on our
theorizing skills and the place of theory in our work. The spirit in which we
should ponder the advice implicit in these papers is captured by George Romans:
Naturally I believe that what I have said here is true. But the most important
advice I can give contemporary sociologists has nothing to do with the valid-
ity of my arguments. It is this: you do not have to believe anything about
theory or methodology that is told you pretentiously and sanctimoniously
by other sociologists-including myself. So much guff has gotten mixed with
truth that, if you cannot tell which is which, you had better reject it all. It
will only get in your way. No one will go far wrong theoretically who remains
in close touch with and seeks to understand a body of concrete phenomena
(1964, pp. 975-976).
REFERENCES
Campbell, D. T. Ethnocentrism of disciplines and the fish-scale model of omnisci-

ence. In M. Sherif and C.W. Sherif (Eds.), InterdiSCiplinary Relationships in
the Social Sciences. Chicago: Aldine, 1969,328-348.
Davis, M. S. That's interesting: Towards a phenomenology of sociology and a
sociology of phenomenology. Philosophy of Social Science, 1971, 1, 309-
344.
Gonos, G. "Situation" versus "frame": The "interactionist" and the "structural-
ist" analyses of everyday life. American Sociological Reveiw, 1977,42, 854-
867.
Handy, R., and Harwood, E.C. A Current Appraisal of the Behavioral Sciences.
Great Barrington, Mass.: Behavioral Research Council, 1973.
Homans, G.C. Contemporary theory in sociology. In R.E.L. Faris (Ed.), Hand-
book of Modern Sociology. Chicago: Rand McNallY, 1964,951-977.
Jantsch, E. DeSign for Evolution. New York: Braziller, 1975.
Klapp, O. E. Opening and Closing. Cambridge: Cambridge University Press, 1978.
Leavitt, H.J. Applied organizational change in industry: structural, technological

and humanistic approaches. In J.G. March (Ed.), Handbook of Organizations.
Chicago: Rand McNally, 1965, 1144-1170.
Maruyama, M. The second cybernetics: Deviation-amplifying mutual causal pro-
cesses. American Scientist, 1963,51, 164-179.
Miller, J.G. Living Systems. New York: McGraw-Hill, 1978.
Moore, W.E. Predicting discontinuities in social change. American Sociological
Review, 1964,29, 331-338.
Mulder, M. The Daily Power Game. Leiden: Martinus Nijhoff, 1977.
Neisser, U. Cognition and Reality. San Francisco: W.H. Freeman, 1976.
Phillips, D. P. Motor vehicle fatalities increase just after publicized suicide
stories. SCience, 1977,196, 1464-1465.
Pondy, L. R. The other hand clapping: An information processing approach to
organizational power. In T.H. Hammer and S.B. Bacharach (Eds.), Reward
Systems and Power Distribution. Ithaca, NY: NYSSILR, Cornell University,
1977,56-91.
Siu, R.G.H. The Man of Many Qualities. Cambridge: MIT Press, 1968.
Thorngate, W. "In general" vs. "it depends": Some comments on the Gergen-
Schlenker debate. Personality and Social Psychology Bulletin, 1976, 2,
404-410.
19,357-367.
35 SCIENCE AS A SOCIAL
REALITY
Gerald R. Salancik, University of Illinois
Lou Pondy said that he wanted me to speak last because he wanted the con-
ference to end on a cynical note. However, after listening to Karl, I don't think
I can reach the expected pitch, but I can summarize some of the content of the
conference: Briefly, we went through pebbles and popes and circles and now
clocks; without straining imagination, it would seem that our conference was on
the rock of the Middle Ages.
Beyond that, I'd like not to talk about the contents of the papers. You can
read them. They've been discussed already. I agree with Karl Weick that they
were extremely well summarized by the discussants.
Instead I'd like to discuss the social process of the conference.
During the conference, I've been curious about two things: the enormous
interest in middle range theories and the disparaging things people said about
them. From these, I'm trying to figure out why a smart fellow like Robert
Merton would say a really stupid thing like: Go ye into the world and grow
theories of the middle kind and gather them altogether into a grand theory?
The very first paper by Bluedorn and Evered, I think accurately suggested
why Merton's charge was not the way science gets done or the way grand images
of the world are developed. It is just cognitively very difficult to take several dis-
parate and independently created ideas and unify them into one grand idea.
408
SCIENCE AS A SOCIAL REALITY 409
What more likely happens is a third independent idea develops that balances
or contains the prior two.
So the question is: Why would Merton suggest middle range theory develop-
ment as a model for science? And one answer to the question may be he was
being foolhardly and didn't use his sense. But that is not likely and doesn't
make sense from what else we know about Merton's work. This answer also
doesn't answer another question, which is, why would so many people orient
and rally around Merton's notion? That is, why would all of the people who
have read Merton and have spoken his truths from their mouths also suggest
middle levels for science?
I think the answer is that Merton, rather than being foolish, really was a
cagey capitalist. What he was doing was solving the problem that he started with,
which was to stimulate scientists to be more productive. And like any grand old
man or woman of any discipline or any enterprise, Merton was in the business of
creating a myth. As a figurehead for science, he was inventing myths for scien-
tists to orient around.
The myth Merton created was that of a very orderly process for the doing of
science. You take comprehensible sets of observed phonomena, and you develop
a limited set of ideas that account for those sets of data. And then you take
these bits and pieces of limited ideas and put them together into a grand theory.
But, in order for the grand theory to be comprehensible, the scientist must begin
with small segments of the world's phenomena, because they can be compre-
hended. For science to proceed with such order gives scientists a very good sense
of what to do and how to do it. It gives them direction. This myth is like the one
used in the organization of Ph.D. training. Nobody really knows how to train a
Ph.D., but what mentors do know is how to organize the process in an orderly
and knowledgeable manner. This is done by saying there are certain stages a
Ph.D. candidate must go through. By this device, both the professors-the
trainers or masters-and the students know what to do at anyone point in time.
The procedures inform each about what to do. No one is confused.
Thus, we have this neat myth about Ph.D. education that allows us all to be-
have and engage in the business of training brilliance. And I think that was what
Merton really was doing in developing his myth about science proceedings in
stages from middle ranges to grand theory. The problem he was addressing, de-
ducible from his own words, was no less than the problem of shaping brilliance.
The object of Merton's myth, however, was not Ph.D. education but science
and the motivation of scientists to be productive. At the time Merton began his
writings there were a lot of grand theorists, about whom Merton perhaps didn't
think kindly because they were writing elaborate and weighty statements about
the nature of the social world. The theories were grand because they were large;
they required large books to present them, and large words to capture their
essence. And like most large things, they awed people and frightened people.
They were difficult to comprehend. They were obtuse, that is, they did not
shed much insight into the nature of the workings of the world or the workings
of society even though they were glowing and grand and exciting. They excited
poeple, but when you got down to brass tacks, nobody could figure out what
the hell was going on.
At the time Merton wrote, there were people like Talcott·Parsons unraveling
the conceptual mess of social systems and making sense of it. And maybe grand
theories made sense to such people. But the rest of science's poor working stiffs
may not have been able to figure it out. And at this time of the history of the
world, the production of Ph.D.'s, and the production of scholars itself was an
expanding phenomenon. The problem then, became: How to motivate such
large numbers of less grand scholars to do science.
It wouldn't make sense to have everybody sit around and write grand theories
because most of us are going to be incompetent on a pure probability basis. Most
of us are not going to be able to do that, and so the rest of us need something
that we can do and which, in fact, will get us on to the business of doing science.
So Merton hit upon the notion of working on a little piece of the puzzle at a
time; which essentially is what is meant by middle range theorizing. What Merton
meant by doing a science in a middle range was to take limited sets of phenom·
ena, understand those, and develop ideas about them. He said little more than
that. And I think his idea was a brilliant solution to the problem of motivating
the growing numbers of everyday scholars because, indeed, it gives them a sense
of what they can do. By presenting science as a middle range activity, Merton
suggested that everyone, the middling and the grand, can participate in this busi·
ness of science because everyone can do something. One of the things that was
pointed out about the nature of middle range theory during this conference was
that the middle ranges all the way from some endpoint called dust·bowl em·
piricism to some endpoint at the grandest of the grand. Scholars of all types
could fit somewhere along this mythical continuum.
One could fit somewhere at a level one can comprehend. Merton's myth
allowed you to practice your art or your science or the art which is your science.
You did not have to sit idly by awing the grand theories of a Parson or a Marx.
Well, I think that this is the value of middle range theory. As evidence I can
point to an observation about this conference. The conference organizers put
this rather ambiguous stimulus level "middle range theory" on the title of a flyer
calling for papers, and amazingly, forty-one papers came flying back. The label
attracted and inspired some eighty people. Being able to attract eighty people
from a field that has maybe only five hundred at maximum, even counting the
fringes, is doing a hell of a lot. Truly, the myth Merton created motivates.
In addition to the intellectual attraction of this flypaper, I'm also amazed
about the social process thereby created. A paper call is a kind of stimulus to
which people respond on their own volition; and everyone does so relatively
independent of everyone else. Nobody knows who else is submitting ideas, or
what ideas are submitted. So what you end up with is a fairly random mix of
flies. In this case, the concept "middle range theory" is a very neat label. It's
like "loose coupling." You can attach anything you want to it and sort of have
a sentence to say. And thus would I offer this conjecture as the middle range
theory of the coarsely coupled ideas presented here.
I said I wasn't going to be cynical. And so to continue: People came with
ideas that were of immediate concern to them. Some people came with the idea
that truth could be found in traditional lore. Other people came with the very
new idea of applying topological theory to interpersonal relationships. It was
an idea randomly caught by the flypaper of the conference. But, it is also a very
interesting idea and should be pursued. The conference is a success for catching
such flies.
But what amazes one is not only that so many different theoretical and em-
pirical projects came in through this process, but also that such diverse people
came here to discuss their ideas and to hear each other's thoughts. The result is
essentially a mingling of strange ideas. And mixing has its hazards. You, as par-
ticipants, expose to yourself and to others the reality that there are a hell of a
lot of ideas in social science. Everyone is so different. There were the idealists,
the realists, the map makers; the map breakers; the sags and the gases. There was
diversity and confusion.
I think on the first day of the conference, I observed a bit of testiness in the
meeting. People were a bit anxious. That was very understandable. People were
aware that other people around them behaved slightly differently. Some were
amazed or distressed. Others were looking for compatriots who shared ideas.
There was a lot of social communication. Extraordinary affmities were developing.
For instance, at one point, Bobko used this word ''multilectic.'' Now, there
aren't as many as a thousand people in the world who would use a word like
''multilectic.'' Yet he blasts the room with this word, and it turns out on a pure
chance basis there's another person in the room who understood. Ann Huffs
eyes lit up. And while there may be alternative interpretations of this reaction,
I happen to know that Huff had also used this word "multilectic" in a paper she
wrote. An affinity was found. I later learned that this affmity went beyond lit
eyes. Professor Huff gave to Professor Bobko her paper with its strange "multi-
lectic" words.
So in the end, this flypaper conference, despite its random attractions, makes
sense. It makes sense because people learn that there are others like them, de-
spite their differences and their other ways of looking at things. The differences
are slightly frightening, slightly confusing. But in the end, they are sorted out.
We are forced, flies that we are, to come together randomly and stuck by the
propriety of attending, we learn that there are others in this community, with
whom we can talk, with whom we can feel comfortable, and indeed, in the
future, with whom we will work and collaborate. From such as this may come
grand theories eventually.
This, in short, is why I think Merton was brilliant. His was the clever strategy
that all organizations use to get people to work their noses off for nothing. They
create a myth that gives purpose to work. Merton invented "middle range"
mythologies to get scientists to work at their emerging craft. And his idea that
middle range theories will become integrated into grand theories, I think is an
inherently necessary part of the overall purpose I'm reading into Merton. He has
created a myth that works. He glorified the middle range with a nirvana. It sells
better. If anybody is selling you soap and they say to you that, here is a bar of
soap, it's a little different from other bars of soap, but then stops at that point,
it wouldn't sell very well. When you sell soap, you've got to say it's the best
soap in town and leads to the cleanest skin.
And when you sell science, you do so with the notion that it will lead to
truth and to grand theory. It feeds on an implicit assumption which we accept,
which is that we all want to end up to a grand understanding of the world, a
grand theory. Of course, some of us don't want to do that. Some of just like to
flick pebbles around, like me.
But Merton addresses the series scholars. To them he offers grand theory as
an end. And, in saying this, he's saying that we all know what we're doing, that
in fact, this science, is a serious business. He is saying that we are getting some-
where. And by suggesting middle range theories as a model of science, he gives
us a specific task to do to get to our goal.
And in this way, Merton has given us something important. Everybody needs
to have a sense that what they are doing is important. And thus Merton's ques-
tionable notions make sense, in the same way that gatherings of this nature do.
Both are exactly what they should be. But because we participate in them, we
shall walk away feeling: "Bejeezes, I really know what it's all about." It's a bit of
a mess, but it's getting us somewhere. Where? Nobody really knows, but some-
where nonetheless. One thing we do know is that, if ain't nobody doing anything
to figure out how this world works, we ain't gonna ever figure it out.
In the process, however, we need to accept our distress about diversity.
There's a paradox that I will call "the pondeviant paradox," after Louis Pondy
because he exemplified it to some degree at this conference.
If you observed him carefully during this conference, you would have noticed
the loud excited tones he used to describe the diversity in paradigms, the many
ways of doing things. Even the circle image is characteristic. The idea that he
would stand anywhere and still have somewhere else to be is pure pondeviant. It
communicates to us that he embraces diversity. But the paradox comes when he

then says that this embace of diverse paradigms is itself the only real paradigm.
He's got this great tolerance for intolerance but he's intolerable about it. In this
way, Pondy embodies the paradox of doing science and living science in a social
reality. We have a norm that allows us to come together and listen to all the
craziness of other people while we're putting out our own crazinesses. But from
time to time this get us very angry and distressed. During this conference, I saw
people distressed because they couldn't get others to believe what they wanted
them to believe. Or they were distressed because they couldn't understand how
some of the nice people they met here could believe some of the crazy things
being said. But that's okay. The distress is okay because everyone will go off
with their tensions and anxieties and they'll talk to other people who agree with
them; and they'll go and they'll try to sort out the differences. And through this
social process, new paradigms may get developed. In fact Kenneth Benson is
right: paradigms aren't the beginning point: They're the endpoint of some other
process. People trying to beat down one another with their versions of wisdom is
science in its social reality. Anything more is just words and since I've run out of
them-

Middle Range Theory and The Study of Organizations Middle Range Theory and The Study of Organizations

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Middle Range Theory and The Study of Organizations Middle Range Theory and The Study of Organizations

Uploaded by

Copyright:

Available Formats

MIDDLE RANGE THEORY

AND THE STUDY

GMartinus GJVijhoff Publishing

1. Toward Middle Range Theory: An Overview and Perspective 1

I. WHAT IS MIDDLE RANGE THEORY? 17

5. Commentary on Gilfillan, Morrow-Muchinsky, and Bluedorn-

II. WHY AND HOW DOES MIDDLE RANGE THEORY DEVELOP? 85

III. GENERAL FRAMEWORKS FOR MIDDLE RANGE

16. The Resurrection of Taxonomy to Aid the Development of

IV. EXAMPLES OF MIDDLE RANGE THEORY 255

28. A Rationale for the Limited Domain Approach to the

V. COUNTERPOINTS AND ALTERNATIVES 337

VI. CONTEMPLATIVE PANEL DISCUSSION 383

2. They tend to be (ultimately) linkable to each other.

HISTORY OF THEORY AND RESEARCH IN

1. A period of grand theory, ranging approximately from 1920 to 1950;

The Grand Theory Period

Until around 1950, thinking about organizations was dominated by classical

The third characteristic of the classical and human relations theories is

The Search for Empirical Generalizations

busy generating observations, searching for generalizations in these observations,

Empirical Shotgunning. There was a tendency to correlate a large number of

The Assumption of Consequences. Researchers often made assumptions about

Neglect of Moderators. Study after study correlated measures of job satisfac-

Recent Trends in Theory Development

The Complex Organization Approach. A second major research thrust was

CURRENT PROBLEMS IN ORGANIZATIONAL BEHAVIOR

Overemphasis on Hypothesis Testing at the Expense of

Worship of Moderator Variables at the Expense of

Disappointment over the Results of Present Research Efforts

The Loss of Parsimony

The increasing tendency of O.B. researchers to make use of multivariate methods

It is our view that the dominant mode of theory-building currently charac-

OVERVIEW OF THE CONFERENCE

What is Middle Range Theory?

Although this paper began with a summary of Merton's (1968) conception of

Why Does Middle Range Theory Develop?

General Frameworks for Middle Range Theorizing

Continuing with the notion of a multiple-paradigms perspective suggested by

network or method of fostering integration of theories that may have developed

Examples of Middle Range Theory

Counterpoints and Alternatives

Hage, J., and Aiken, M. Relationship of centralization to other structural prop-

THE MEANING OF MIDDLE RANGE THEORY (MRT)

MIDDLE RANGE THEORY AS PROCESS

THE STRATEGIES OF THEORY CONSTRUCTION

EMPIRICAL . MIDDLE RANGE

Table 1. Theory Construction Strategies.

Strategy Number Strategy Description

TESTING THE MIDDLE RANGE IMPERATIVE

An eighth disconfirmation can be found in the physical sciences: Newtonian

TOWARDS A MULTIPLICITY OF THEORY-BUILDING

The foregoing criticism of Merton's formulation of MRT should not be taken as

In Merton's world view, MRTs act as a kind of buffer to separate empirical

Alternative 1. Theorizing as Blending. It can simply be assumed that all scien-

Alternative 3. Theorizing as Personal Style. Psychological research on individ-

Alternative 4. Theorizing as Socially Detennined. From the perspective of the

Asimov, I. Asimov's Biographical Encyclopedia of Science and Technology. Gar-

Tshetverikov, S.S. On certain aspects of evolutionary process from the stand-

constituent propositions are couched at a middling level of generality. They are

THE MIDDLE RANGE LEGACY

LIMITATIONS OF THE MIDDLE RANGE APPROACH

ISSUES IN MIDDLE RANGE THEORY CONSTRUCTION: