Barley y Tolbert Institutionalization
Barley y Tolbert Institutionalization
Barley y Tolbert Institutionalization
Abstract
Stephen R. Institutional theory and structuration theory both contend that institutions and
Barley actions are inextricably linked and that institutionalization is best understood
Department of as a dynamic, ongoing process. Institutionalists, however, have pursued an
Industrial
empirical agenda that has largely ignored how institutions are created, altered,
Engineering and and reproduced, in part, because their models of institutionalization as a pro-
Engineering cess are underdeveloped. Structuration theory, on the other hand, largely
Management, remains a process theory of such abstraction that it has generated few empirical
Stanford
University, studies. This paper discusses the similarities between the two theories,
U.S.A. develops an argument for why a fusion of the two would enable institutional
theory to significantly advance, develops a model of institutionalization as a
Pamela S. structuration process, and proposes methodological guidelines for investigating
Tolbert the process empirically.
School of
Industrial and
Labour Relations, Descriptors: structuration, institutionalization, organizational change, social
Cornell action, methodology, longitudinal analysis
University,
Ithaca, U.S.A.
Introduction - ~ . ~ ~ - ~ . , -
.
’~
constrain the very actions that produce them ( ‘emergence of ... struc-
tures of domination’). Thus, in their early work, institutionalists expli-
citly postulated that institutions exhibit an inherent duality: they both
arise from and constrain social action. More often than not, however,
institutionalists have concentrated on an institution’s capacity to con-
strain (Zucker 1987: 444-447; Whittington 1992).
With the passage of time, Meyer and his colleagues have come to treat
institutions primarily as exogenous to organizational action (Meyer and
Scott 1983; Scott and Meyer 1987; Sutton et al. 1994; Scott and Meyer
1994). Their more recent work associates institutional pressures with
the demands of centralized authorities or regulatory agencies and, only
secondarily, with widespread beliefs, practices, and norms. Con-
sequently, their research programme has focused on the causes and
consequences of conformity and on the manner in which the environ-
-
I
96
f’ . ~.: j ..
n :j .’.. ~ . ...
,,:> ... ,
t í’
~
Defining Institution --
relationships (see also Burns and Flam 1987). This definition bears a
,
strong resemblance to Giddens’ (1984:2 377) notion of ’structure’ and
..
Sewell’s (1992) idea of a ’schema’. Neither Giddens nor Sewell, how-
ever, emphasize the degree to which institutions vary in their normative
.
point out, practices and behavioural patterns are not equally institution-
,
.
Institutions in Action ,
_~
Figure 1
Giddens’ Model
of Structuration
98
.
.
analysts have treated scripts primarily as cognitive phenomena (Schank
and Ableson 1977), we believe it is empirically more fruitful to view
scripts as behavioural regularities instead of mental models or plans.
I i From this perspective, scripts are ohservable, recurrent activities and
..
patterns of interaction characteristic of a particular setting. Scripts
encode the social logic of what Goffman (1983) called an ’interaction
order’. Our contention is that the institutions relevant to a particular
setting will manifest themselves in behaviours characteristic of that set-
ting and, hence, will appear as local variants of more general
principles.
The notion of a script usefully substitutes for Giddens’ more abstract
. notion of modalities because scripts can be empirically identified,
.
Because Giddens argues that institutions exist only insofar as they are
instantiated in everyday activity, critics have charged that he ’conflates’
- .
structure with action (Archer 1982, 1989; Layder 1987; Callinicos
1985). Conflation concerns the problem of reducing structure to action
(or vice versa) and the difficulty of documenting the existence of an
.
institution apart from activity. Unless institutions and actions are ana-
lytically as well as phenomenologically distinct, it is difficult to under-
stand how one can be said to affect the other. Although the critics of
,
structuration theory have aimed their critique at problems they believe
, to be inherent to the theory’s logic and, for this reason, have sometimes
argued for re-establishing the separation between structure and action
.
: . ;
&dquo; , ..- . l’
, ’; I . «« ’; ~, . I. I.: .. , . :&dquo; iI
,
.
Institutions and Actions: A Recursive Model
.
< .
.
new institution is, to say the least, a formidable task. An enormous
amount of luck or prescience are required to recognize an emerging
.
institution and then gather data on relevant, ongoing actions and interac-
tions. Moreover, historical or archival material will rarely contain the
detailed data necessary for documenting the link between everyday acts
and the creation of an institution. Thus, it is likely that most investi-
gations seeking to link actions and institutions will focus on the pro-
cesses by which existing institutions are maintained and modified. To
process whose operation can be observed only through time. The bold
’
horizontal arrows that define the figure’s upper and lower boundaries
,
signify the temporal extensions of Giddens’ two realms of social struc-
ture : institution and action. The vertical and diagonal arrows linking
the two realms denote the duality of social systems. Vertical arrows
represent institutional constraints on action, while diagonal arrows rep-
resent maintenance or modification of the institution through action.
Thus, social behaviours constitute institutions diachronically, while
institutions constrain action synchronically.
The first moment (arrow a) entails the encoding of institutional prin-
.
izational rules and procedures often define scripts that embody institu-
101
102
laws, major economic shifts, etc. are most likely to occasion institu-
-
tional change. Structuring a study around such events also has the desir-
able property of allowing the researcher to use a ’before and after’
design. This approach, however, carries important implications for
where and when one collects data.
For instance, it implies that researchers will need to identify, in advance,
events with system-disturbing potential and begin collecting baseline
.
data on the system before an event occurs. Baseline data are crucial
for assessing whether an event brings about changes in scripted action.
The ability to foresee occasions for potentially significant institutional
change is less difficult than it might seem. Organizations usually
anticipate and even announce mergers and acquisitions, the adoption of
new technologies, and changes in personnel before they occur. Sim-
ilarly, governments typically publicize new laws before they are enacted
and routinely stipulate when they will be enforced.
To evaluate the impact of such events on institutional outcomes,
. researchers will need to choose sites for observation that vary along
; dimensions that might be expected to mediate the impact of such events.
.
Because one can normally collect detailed observational data over an
.
since these characteristics should affect how market control will affect
interaction orders. Barley (1990) chose sites that permitted him to com-
pare ongoing interactions surrounding the use of new and old technolo-
gies because he was interested in how computerized medical imaging
devices, in general, and CT scanners, in particular, affected the institu-
tion of professional dominance in radiology departments. He also
required sites that would acquire a scanner sometime during the course
of the research since this was the only imaging technology that was
diffusing into community hospitals at the time. Finally, he chose one
hospital which had close ties to a medical school and one that did not,
..
to determine whether a teaching mandate would influence role relations
and the structuring process. Studies examining the impact of declines
in college enrollment on accreditation systems might similarly target
schools and/or academic departments of varying prestige levels or those
,
emphasizing different degree programmes. In short, to study structur-
.
ation, scholars need to select sites based on careful consideration of
factors apt to affect an institution’s boundaries or that may impede or
.
enhance institutional change (see Scheid-Cook 1992).
,,~. _&dquo;i’ I I ..
-, .
>
105
and hypotheses while there is time to acquire additional data. Four pro-
cesses are crucial to identifying and analyzing scripts: ( 1 ) grouping the
data by category or unit of observation, (2) identifying behavioural pat-
terns (scripts) within categories, (3) identifying commonalities across
scripts, and (4) comparing scripts over time.
Grouping Data
.
A system of categories for collecting and sorting data greatly facilitates
identifying scripts. Although the types of categories used depend on the
research project, experienced field researchers typically employ several
.
broad categorization schemes for collecting observational data (see
Lofland 1976). One such scheme targets types of events or aotivities
that have discrete temporal and spatial boundaries and that are ecolo-
.
/~ ( .~ ’
I ’.,
ging. Close observation and careful recording increase the odds of iden-
tifying scripts, even though one may not recognize at the time that the
events one is observing are either important or scripted. Meticulous
.
sought directions from mid-level managers. One might not want to con-
clude that the institution of hierarchical authority had been seriously
altered by the shift to teams.
An additional point is worth noting. Up to now, we have discussed the
collection and analysis of data in terms of a simple comparison of social
action and interaction before and after some exogenous change. In actu-
ality, the data following an exogenous event may evince multiple
periods. As discussed earlier, researchers are unlikely to anticipate such
additional periods of structuring and, even after extensive observations,
the existence of later phases may not be immediately obvious to the
researcher. Consequently, researchers need to remain sensitive to the
possibility of multiple phases defined primarily by observed shifts in
, the patterned properties of ongoing action. For example, Barley (1986)
was able to estimate the approximate timing of the CT scanners’ arrival
in the radiology departments he studied. However, he was unable to
foresee later events, such as personnel changes, that proved to be
significant junctures in the history of scanner operations and that sub-
sequently affected the evolution of the interaction order between techni-
cians and radiologists (and, ultimately, the maintenance and change of
institutional parameters ). Unless researchers remain acutely sensitive to
the possibility of additional, unanticipated partitioning points, they may
misrepresent the dynamics and even the outcomes of the structuring
process.
, 4 .. - ~ . ~
can also be useful. Such data may track changes in institutional para-
meters and help researchers assess the boundaries of such change. Inso-
far as they can be shown to vary with the same type of events that
triggered changes in the interaction order, they may also provide sup-
port for the process-based conclusions drawn from the observational
.
data.
Indices of the concern within the field of radiology for the proper role
.
nologies once again raised the specter of technicians usurping the core
of the radiologists’ role (Barley 1984). Fluctuating interest in the topic
can be interpreted as evidence for the relative stability and change of
’
the institution of professional dominance in radiology which pivots on
restrictions regarding who can interpret medical images (Larkin
1983 ).3
Interestingly enough, after the advent of ultrasound, the first computer-
ized and cybernetic imaging device, the number of published discus-
sions on the rightful role of the technologist in a radiology department
increased. Moreover, new journals targeted at sonographers published
more articles on the interpretation of images than do journals read by
1984).
Several organizational indicators also support the thesis that computer-
ized imaging devices have altered the interaction order of radiology
departments and, consequently, the institution of professional domin-
ance in radiology. During the early 1980s, sonographers established
their own professional society and their own registry, both of which
represented moves toward an occupational identity separate from that
of radiological technologists in general. In sharp contrast to the tests
that x-ray technicians take to become certified, the examination for son-
ographers includes questions on the interpretation of images. Pay scales
for sonographers, CT technicians and special technologists are typically
.
higher than those for x-ray technologists. Hospitals are increasingly
’
treating CT, ultrasound, and special procedures as separate organiza-
111
tional units, distinct from the x-ray department. Finally, and perhaps
most importantly, as the number of computerized imaging devices has
grown, more and more radiology departments have begun to change
their name to ’departments of medical imaging’. By tying such
artifactual data to observational data gathered before and after the intro-
duction of new technologies into radiology departments, Barley ( 1984,
1986, 1990) was able to articulate the link between technological
change, patterned interactions among physicians and technicians, and
changes in the institution of professional dominance in radiology
departments.
·
~ ,
, ~
ways. This is most likely to occur when common conditions affect many
actors more or less simultaneously in much the same way (as occurs
with a shift in an infrastructural technology) and when the social net-
works among actors are relatively dense. The degree to which changes
in the interaction orders of particular settings can affect widespread
institutions and the paths by which such changes unfold are empirical
questions that still require considerable research.
It is also possible for an institution to seem to change without concomit-
ant changes in interaction orders. For example, in theory, a significant
Summary .
.
Structuration theory and institutional theory provide complementary
insights. Both share the premise that action is largely organized by
.
although there are notable exceptions (see, for example, Stimpert et al.
1995; Orlikowski 1992; Orlikowski and Yates 1995; Poole and
DeSanctis 1993; DeSanctis and Poole 1994). Thus, while both institu-
tional and structuration theory promise important insights into the pro-
cess of organizing, our ability to apply and test those insights has been
limited. By offering some preliminary guidelines for studying institu-
tionalization as structuration, we hope to spark further interest in and
debate on the issue. We view our contribution as but an opening remark
in a long-running conversation, and far from the final word.
In fact, by choosing to focus on the identification and analysis of scripts
in our approach, we have consciously emphasized the behavioural and
the structural rather than the cognitive and the cultural. Although we
believe such an approach has value because it enables systematic
empiricism, it does so at the cost of relegating interpretations to the
background. Consequently, our approach is far less sensitive to the role
that cognitions and interpretive frames play in the institutionalization
process than are the methodologies employed by Pettigrew ( 1987 ),
Willmott ( 1987) and other European students of structuration in organ-
izations. What is required is a systematic exploration of the relative
importance of behavioural and interpretive phenomena in the institu-
tionalization process and, on the basis of such exploration, the fash-
ioning of a set of methods that are sensitive to and systematic about
documenting both cultural and structural dynamics.
4otes 1. Scott ( 1988) has identified four separate variants of institutional theory. In addition
to the approach taken by Zucker and her colleagues and that developed by Meyer and
his colleagues, Scott distinguishes Selznick’s approach, which associates institutions
with valued organizations, and a functionalist or Parsonian approach which identifies
institutions with sectors of society that are characterized by distinct systems of values
and beliefs. While the latter two clearly represent important approaches to the study
of institutions, their influence in contemporary organization theory has been slight in
comparison to that of the more recent camps. We, therefore, focus explicitly on more
recent conceptualizations.
2. While socialization often connotes a specific period of time during which neophytes
enter a group and leam its ways, we contend that such a connotation is too limiting.
Socialization is a process that never ceases, because individuals are always being
assigned new roles and adapting to changes in existing roles. To understand the role
of socialization in the structuring process, we must not make the mistake of limiting
socialization to a particular phase of a person’s career in a social collective.
3. Although proscriptions against technologists interpreting films are central to the
system of professional dominance in radiology, this particular proscription is relatively
meaningless in other medical specialties. Yet, professional dominance is an institution
that encompasses all medical specialties (Friedson 1970). The difference underscores
a point we made at the outset: broad institutions may be represented by different scripts
in different settings.
Hinings, C. R., J. L. Brown, and Royston Meyer, John W., and Brian Rowan
Greenwood 1977 ’Institutionalized organizations:
1991 ’Change in an autonomous pro- formal structure as myth and
fessional organization’. Journal ceremony’. American Journal of
of Management Studies 28/4: Sociology 83: 340-363.
375-393.
Meyer, John W., and W. Richard Scott
Hughes, Everett C. 1983 Organizational environments:
1936 ’The ecological aspects of institu- Ritual and rationality. Beverley
tions’. American Sociological Hills, CA: Sage.
Review 1: 180-189.
Mintz, Beth, and Michael Schwartz
Hughes, Everett C. 1985 The power structure of American
1939 ’Institutions’ in An outlineof the husiness. Chicago: University of
principles of sociology.
Robert E. Chicago Press.
Park (ed.), 283-346. New York:
Bames and Noble. Noble, David
1984 Forces of production: A social
Jepperson, Ronald historyof industrial automation.
1991 ’Institutions, institutional effects New York: Knopf.
and institutionalism’ in New insti-
tutionalism in organizational Oliver, Christine
analysis. W. Powell and P. 1991 ’Strategic responses to institu-
DiMaggio (eds.), 143-153. Chicago: tional processes’. Academy of
University of Chicago Press. Management Review 16: 145-
179.
Kline, Ronald
1987 ’The origins of industrial research Orlikowski, Wanda J.
at the Westinghouse Electric 1992 ’The duality of technology:
Company,1886-1922’. Unpub- Rethinking the concept of techno-
lished manuscript. Program in logy in organizations’. Organiza-
History and Philosophy of Sci- tion Science 3: 398-427.
ence and Technology, Cornell
Sutton, John, Frank Dobbin, John Meyer, Yates, JoAnne, and Wanda J. Orlikowski
and W. Richard Scott 1992 ’Genres of organizational com-
1994 ’The legalization of the work- munication : structurational
A
place’. American Journal of Soci- approach to studying communica-
ology 99: 944-971. tion and media’. Academy of
Management Review 17: 299-
Tolbert, Pamela S. 326.
1985 ’Resource dependence and insti-
tutional environments: sources of Zucker, Lynne G.
administrative structure in institu- 1977 ’The role of institutionalization in
tions of higher education’. cultural persistence’. American
Administrative Science Quarterly Sociological Review 42: 726-743.
30: 1-13.
Zucker, Lynne G.
Tolbert, Pamela S., and Lynne G. Zucker 1983 ’Organizations as institutions’ in
1983 ’Institutional sources of change in Research the sociology of
in
the formal structure of organiza- organizations. Sam Bacharach
tions : the diffusion of civil ser- (ed.), 1-47. Greenwich, CN: JAI
vice reform, 1880-1930’. Admin- Press.
istrative Science Quarterly 30:
1-13. Zucker, Lynne G.
1986 ’Production of trust: institutional
Tolbert, Pamela S., and Lynne G. Zucker sources of economic structure:
1996 ’Institutionalization of institu- 1840 to 1920’ in Research in
tional theory’ in Handbook of organizational behavior. L. L.
organizational studies. S. Clegg, Cummings and Barry Staw (eds.),
C. Hardy and W. Nord (eds.). 55-111. Greenwich, CN: JAI
London: Sage. Press.
Woodward, Joan
1958 Management and technology.
London: HMSO.