Barley 1990
Barley 1990
Barley 1990
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
INFORMS is collaborating with JSTOR to digitize, preserve and extend access to Organization Science.
http://www.jstor.org
This paper discusses the problems and processes involved in conducting longitudinal
ethnographic research. The author's field study of technological change in radiology provides
the context for the discussion. Specific attention is paid to how researchers can design a
qualitative study and then collect data in a systematic and explicit manner. Consequently, the
paper seeks to dispel the notion that participant observation and quantitative data analysis
are inimical. Finally, the social and human problems of gaining entry into a research site,
constructing a research role, and managing relationships with informants are illustrated.
(LONGITUDINAL ETHNOGRAPHY)
ContextualInfluences. The story of any research program rightfully begins with the
decision to investigate a specific topic. Although some social research is undoubtedly
formulated, as textbooks advise, to answer nomothetic questions or to expand on
previous research, my work on radiology began as an attempt to fill what I thought
was an interesting empirical void. To be sure, theoretical concerns played a back-
ground role in so far as they provided a frame of reference and a set of sensitizing
concepts. The study was, in fact, guided by two tenets integral to a symbolic
interactionist's notion of social structure: (a) that social structures emerge gradually
out of an ongoing process of negotiation and interaction and (b) that technology
should occasion structural change by altering roles and relationships. However, like
most ethnographers. I embarked on the project with little more than a handful of
*Accepted by George P. Huber; received February 15, 1989.
220
1047-7039/90/0103/0220/$01.25
Copyright e 1990, The Institute of Management Sciences
general questions and no strongly articulated hypotheses. Far more important for
determining the study's focus were personal interests and the circumstances of my
location in time and space.
Few students at MIT in late 1970s could avoid the perception that computers and
other advanced technologies would slowly, but surely, transform the infrastructure of
society and that this transformation would stem from changes in the workplace. Even
in the Institute's School of Management, conversations frequently turned to talk
about computer-aided design and manufacturing, computer networks, fiber optics,
genetic engineering, artificial intelligence, parallel processing, decision support sys-
tems, and a host of other technologies far too esoteric to mention. Most of these
machines and techniques were designed for use by professionals and semi-profes-
sionals; the so-called "knowledge workers" of Daniel Bell's (1973) post-industrial
society. Yet, then as now, most social research on technical change concerned either
mid-level managers or blue-collar and lower-white collar workers.
Most previous research had also failed to examine closely the dynamics by which
technologies might occasion social change. For although many studies of technology
were wittingly or unwittingly premised on Marx's notion that shifts in the technical
infrastructure transform societies by altering modes and relations of production, few
researchers had actually traced organizational and occupational changes to shifts in
the nature of work or the manner in which work is performed. Instead, most had
settled for auguring such broad trends as decentralization, deskilling, and reskilling
without much data and with little first-hand knowledge of how machines were
actually used. This state of affairs was particularly true for the managerial and
organizational literature on new technology where it seemed that only the early
socio-technical theorists (Trist and Bamforth 1951, Rice 1958, Herbst 1962,
Fensham and Hooper 1964) and a handful of others (Mann and Williams 1961,
Meissner 1969) had ever bothered to observe the actual turning of a machine's cam.
Largely because I had a keen interest in occupational dynamics and because I had
been steeped in Everett C. Hughes' (1958, 1972) brand of occupational sociology, the
time seemed ripe for closely studying the process by which computerized technologies
were altering the work of professionals and semi-professionals. Such a study would
chronicle the actions, interactions, and interpretations occasioned by specific ma-
chines to explain how technically induced changes in an interaction order (Goffman
1983) might lead to organizational and occupational change. Given this rather broad
mandate, the specific decision to examine radiological technologies was largely
opportunistic.
Medicine also seemed appropriate for personal reasons. As a teenager, I had been
deeply interested in the life sciences and was certain that I would pursue a medical
career. Not only had I studied molecular biology and human physiology as an
undergraduate, but during high school I had worked in a hospital's pathology lab.
Consequently, the physician's task was less foreign to me than the work of an
engineer or lawyer, and since I was interested in understanding how technologies
alter the details of professional practice, it seemed pragmatic to look for a setting
where I had some understanding of the science on which the practice was based. In
retrospect, these affective and personal motives proved to be extremely critical. For
without an intrinsic interest in medicine, I would have probably jettisoned the project
in face of the frustrations and difficulties I encountered while negotiating entry.
In the early fall of 1981, I began shopping for a suitable medical technology. I
initially investigated several expert programs and a number of computerized devices
used in surgery or intensive care. However, all proved to be either uninteresting or
else sophisticated prototypes located in medical centers where physicians were
themselves primarily researchers. I eventually stumbled across radiology and the
medical imaging revolution quite by accident during a conversation with Dr. Octo
Barnett, a member of the staff at Massachusetts General Hospital and an expert in
the use of computers in medicine. After explaining that I wanted to study a
technology that was already affecting medical practice, but which was still in the early
phases of its diffusion curve, Dr. Barnett counseled, "You ought to study radiology. If
there's any area in medicine where computers are transforming medical practice,
radiology is it."
After spending several weeks reading the radiological literature, I concluded that
radiology indeed offered an opportune setting. Since the late 1960s radiology had
been jolted out of its technical calm by a series of innovations, most of which involved
computers and each of which represented a new way of looking inside the human
body. Three of these technologies-special procedures, ultrasound, and computer-
ized tomography (CT) scanning-had become prevalent in community hospitals.
Common wisdom was that the new technologies had occasioned a revolution in
medical diagnosis (Banta and McNeil 1978, OTA 1978, 1981, Stocking and Morrison
1978, Wiener 1979). Hoping that the influx of sophisticated technologies had also
occasioned social change, I began to design a study for determining whether and how
new imaging devices had altered the social organization of radiological work.
Research Design
Three broad and interrelated, but distinct, questions guided the research design.
Did the social organization of special procedures, ultrasound, and CT scanning differ
from the social order surrounding radiology's traditional technologies, radiography
and fluoroscopy? If so, why did such differences exist? Finally, were the differences
peculiar to specific hospitals or were they to be found in any radiology department
that had adopted the technologies? To answer each question, data were collected
from three distinct vantage points which might be called, respectively, the synchronic,
the diachronic, and the parallel.
mere shadows of some long forgotten substance. What existed and why it existed can
be culled only from the memories of people who lived during the period or from
records that survived their time. In other situations, rare by comparison, a prior social
order continues to exist side by side with what has come since and with what is likely
to come in the near future. When living traces of a former order reside along side
vestiges of a new, it is possible to compare the two simultaneously.
The sociology of technology and work is perhaps blessed for this reason. In periods
of technical change, workplaces often become mixtures of old and new technologies
operated concurrently. Associated with each technology is the social order that grew
up around its use. Because the mixture occurs in the same setting, one can essentially
hold constant variations brought about by the organization's cultural, historical,
environmental, and social idiosyncracies in order to see more clearly the ramifications
of the technologies themselves.
Radiology departments were (and remain) living, Janus-like documents of their
social and technical evolution. In most hospitals, radiographic and fluoroscopic
equipment stand juxtaposed to the newer devices and techniques of ultrasound, CT,
and special procedures. This temporal and spatial concurrence was especially propi-
tious for analyzing old and new imaging technologies synchronically. Because of the
average department's technical mix, one could treat each technology's context as if it
were fixed in time. Although clearly fictional, the assumption of stasis enabled me to
compare simultaneously each technology's operation with an eye to determining
whether the new devices' social contexts were similar to each other and yet different
from the social organization of radiography and fluoroscopy. In particular, I antici-
pated that a synchronic analysis would highlight differences and similarities in the
tasks, roles, and role relationships of the radiologists and technologists who worked
with each device.' If the new technologies had indeed engendered social change, then
the data should show that the roles of radiologists and technologists in new areas
were different from the roles of those who worked with older technologies.
Diachronic Analysis. Should the synchronic analysis show that the new technologies
had occasioned social orders that were similarly different from the social organization
of radiography and fluoroscopy, then one might reasonably claim that the new devices
had pointed the practice of radiology in a more or less consistent direction. However,
from synchronic data alone one could not determine how or why a new social order
had evolved. To understand how the new technologies had breached radiology's
structural and cultural traditions required charting changes as they occurred. In other
words, a diachronic perspective was also necessary. Whereas a synchronic analysis
would freeze time and look across a radiology department as a whole, a diachronic
analysis would seize time and examine the developmental path of a specific technol-
ogy's use. Said differently, a synchronic analysis would compare technologies with
each other, whereas a diachronic analysis would contrast earlier and later periods of a
single technology's use.
Diachronic analysis was also most compatible with a symbolic interactionist's
notion of a negotiated order: the idea that social structures sediment out of a stream
of ongoing actions, interactions, and interpretations that gradually define the con-
tours of tasks, roles, and relationships as well as a technology's identity as a social
object (Strauss et al. 1963, 1964, Berger and Luckmann 1967, Strauss 1978, Giddens
1984, Barley 1988c). Since the events of day-to-day life fade quickly into the flow of
1The occupational groups most central to the work of a radiology department are radiologists and
technologists. Radiologists are physicians who specialize in the interpretation of medical images. Technolo-
gists are individuals with an associate's degree in radiological technology. Technologists produce the films
that radiologists interpret.
experience, participants in a work setting may not notice their cumulative import. To
be sure, participants may realize retrospectively that changes have occurred, but they
are unlikely to remember how or why the changes arose. Thus, if researchers are to
understand how a technology becomes embedded in a particular social order, they
need to observe the actions and interactions that actually generate the order. The
most accurate way to collect such data is to follow the use of a new technology from
the moment it goes on-line.
2The terms "diachronic" and "synchronic" are adopted from linguistics. "Parallel" is, to the best of my
knowledge, my own creation.
clarify such ambiguities. Finally, parallel studies allow one to generalize synchronic
and diachronic findings across similar social settings. Only by parallel analysis can a
researcher conclude that an organization's idiosyncracies do not account for the
outcomes associated with a set of computerized devices.
More importantly, conclusions become problematic when research questions are
paired with an inappropriate comparative stance. For example, synchronic data may
seem to suggest that similar outcomes are rooted in similar processes. However,
similar outcomes may arise from different processes and different outcomes may arise
from similar dynamics (Barley 1986a). Only diachronic data can disentangle such
possibilities. By itself, a parallel study of a class of events, objects, or activities may
also lead to wrongful conclusions. Suppose, for instance, that one were to investigate
the effects of new technologies by studying CT scanning in a number of hospitals.
Even if one found that all CT scanners occasion similar phenomena, one could not be
sure whether the findings would apply to all computationally based imaging devices or
only to CT scanners. A synchronic analysis of several technologies conducted in
tandem could resolve the issue. In other words, the synchronic, the diachronic, and
the parallel represent three distinct axes of comparison which, when used in combina-
tion, allow researchers to examine explicitly the spatial and temporal boundaries of
their claims.
For these reasons, I chose to pursue the triple comparative focus summarized in
Figure 1. The triple focus, in turn, strongly constrained the universe of potential
research sites. Since synchronic differences can be assessed by merely comparing the
daily use of different technologies, any of a large number of radiology departments
could have served as sites for the study's synchronic thrust. However, the diachronic
agenda severely restricted freedom of selection. At the time of the study, only
whole-body CT scanners were being widely adopted by community hospitals in
Massachusetts. Most radiology departments had already implemented special proce-
dures and ultrasound, and only the largest medical centers were using magnetic
resonance imaging. The goal of conducting a diachronic comparison, therefore, left
little choice but to target hospitals that were about to receive their first body scanner.
The diachronic agenda also required that I begin the research before the scanners
went on-line. Finally, to perform a parallel comparison required that I study at least
two sites, each of which satisfied the diachronic criteria.
Execution
Entry
Like most states, Massachusetts limits the number of hospitals allowed to purchase
costly medical devices through a "Determination of Need" process (OTA 1978, Banta
and Behney 1980). Because CT scanners cost between 0.5 and 1.5 million dollars, only
four Massachusetts hospitals were permitted to purchase body scanners during 1982,
the year the study was slated to begin. In January 1982, I obtained the four hospitals'
identities from the Massachusetts Public Health Council. Of the four, one was too
distant to afford frequent access. In February, I contacted the three remaining
hospitals in hope of persuading at least two to participate in the study.
I presented the study as basic social science aimed at better understanding the
social implications of computerization. Although I played on the widespread percep-
tion that radiology was in a period of transformation, I purposely distanced myself
from economists and other critics who were at the time writing articles on radiology's
contribution to the high cost of medical care. Instead, I emphasized my affiliation
with MIT, a well-known haven for technophiles and, as would-therefore be expected,
expressed great interest in the technologies themselves. I also marketed myself as an
Hospital 1 SYNCHRONIC
1
Radiography
Fluoroscop
y DIACHRONIC-
|Ultrasound|
Specials
T T+ 1 T+ 2 . . T+ N
PARALLEL
1
SYNCHRONIC
Radiography
Fluoroscopy|1 DIACHRONIC-
Ultrasound|
Specials
Hospital2 T T+ 1 T+2 .T+N
0 TI MF
3In my experience, the general population has greater esteem for anthropology than for sociology.
Anthropology probably has a better name for a variety of reasons. Most people not only know less about
anthropology than sociology, but because of the popularity of Indiana Jones, National Geographic Specials,
and other PBS programming, anthropology has acquired a certain mystique. In contrast, since the 1960s
sociologists have been given high visibility in the media as society's professional muckrakers. Because of
difficulties encountered in my earlier research on funeral directing, I learned that ethnographers run less
risk of premature rejection when presenting themselves as anthropologists rather than sociologists.
administrator and acknowledged our conversation with a brief letter and research
proposal. Within the first five minutes of our meeting, the administrator firmly
informed me that the radiologists had discussed the proposal and concluded they had
no desire to participate. Although officially rejected because of the project's potential
disruption to the department, I later learned from a prominent and well-connected
radiologist that the hospital had been reprimanded a year earlier for attempting to
skirt the Determination of Need procedure. My research proposal had apparently
aroused the radiologists' suspicions of investigative surveillance, an interpretation
whose viability seems less fanciful given events at the second hospital.
There, I began by negotiating directly with the radiologists as well as the depart-
ment's administrator. The proposal was relatively well received within the depart-
ment, in part, because one of the more influential radiologists had once worked with
a team of anthropologists in New Guinea and was therefore kindly disposed towards
ethnographers. However, the proposal encountered snags at higher levels of the
hospital's administration.
Unknown to me, during the previous year a journalist had entered the hospital
posing as a researcher. He later published an article critical of the cost of medical
care in which he cited physician's names and annual salaries. The incident left the
hospital's administration hesitant to open doors to other persons claiming to be social
scientists. Moreover, two former members of the radiology department's staff had
recently been indicted for selling silver recovered from x-ray film on the black market.
The hospital's higher-ups apparently suspected the purity of my investigative motives
for this reason as well.
At the third radiology department, I was referred to the hospital's second in
command. Although the hospital's higher administration proved enthusiastic about
the research, their enthusiasm did not filter down to the radiology department. I was
eventually able to present my case to the department's Chief Administrator, but I was
unable to meet with the radiologists. In fact, over the next four months I called the
third hospital on a biweekly basis hoping to arrange a meeting with the Chief
Radiologist or to learn from the Chief Administrator whether the department had
made a decision regarding the study. The receptionists always told me that neither
person was in his office, and although the receptionists promised to forward my
messages, neither the administrator nor the Chief Radiologist returned my calls. I
had greater success contacting the administrator at the second hospital. However, on
each call I was politely informed that the hospital had not yet acted on my proposal.
As the weeks dragged on without word from either hospital. I began to fear that I
might never gain access. By May, the thought of wasting a year of preparation became
so acute that I decided to observe scanner operations at a local medical school as a
fallback position. The radiologist in charge of CT operations at the medical school
eventually spoke on my behalf with the chief radiologist at the third hospital.4 Within
three days of the radiologist's intervention, I was granted access to the third hospital.
Later that same week, the second hospital cleared the study. I do not know whether
the second hospital's timing was coincidental or whether the second hospital had also
been contacted by my newly found champion. At any rate, only after successfully, but
blindly, working my way into the radiologist's professional network did "Urban" and
"Suburban" finally grant access. Aside from a promise that I would not reveal the
identify of either hospital, no strings were placed on my participation. In particular, I
4However, he did so only after learning that Edgar Schein was a member of my committee. Several years
earlier, the chief radiologist had been a Sloan Fellow at MIT and had done a thesis with Schein. Although
I make no pretense of understanding the radiologist's motives and interpretations, for whatever reason he
apparently felt that my ties to Schein were a sufficient reason for him to champion my cause.
was not obligated to provide feedback at the end of the study. The experience
underscored a crucial lesson about gaining entry which I have since relearned on
several occasions: despite an academic's proclivity to think otherwise, whom one
knows is often far more practical that what one knows.
Methods
Sustained observation. Negotiated order theorists assert that social order consists
largely of patterned activities and shared interpretations that smooth the vagaries of
daily life by establishing the bedrock of consistency and coherence necessary for
sustained action. Without such patterns, persons would often not know how to
behave or what to think. As a result, organized activity would crumble against the
wake of ongoing experience and be washed aside by complexity and novelty. From
this perspective, it follows that if new technologies are to change a social order, they
must do so by altering old patterns and creating new ones.
If social orders are patterns and if technical change alters those patterns, then
investigations of a technology's implications for the organization of work would seem
to require sustained observation: for unless one observes, one is unlikely to uncover
the behaviors and interpretations that compose the patterns. Moreover, unless
observation is sustained, one can document neither recurrence nor consistency.
Sustained observation is also crucial for tracing the evolution of social institutions.
Institutions do not arise fully formed. Instead, habits and definitions of reality accrue
over time as problematic incidents demand interpretation and action (Berger and
Luckmann 1967). Once an interpretation is posed, an act executed, and the result
deemed sufficient, all three fade into the past to become taken-for-granted aspects of
the present (Schein 1985). Studies of technical change based solely on interviews or
archival materials are, therefore, quite likely to depict the dynamics of social change
inaccurately. It was to avoid the difficulties associated with retrospective interpreta-
tion that I chose sustained observation as my primary methodology.
In the second week of June 1982, nearly a year after the project's inception, I
began a year long stint as a participant observer in the radiology departments at
Urban and Suburban. With the exception of a seven-week hiatus from the field in the
middle of the year, I spent an average of four days a week in observation. By
alternating sites on a daily basis, I logged 84 days of observation at Urban and 87 at
Suburban. My typical routine was to arrive on site between 8 am and 9 am, spend six
to seven hours accompanying technologists or radiologists as they went about their
day's work, and then leave by 3:30 pm when the day shift ended. In the evening after
each day's observation, I expanded my fieldnotes at a typewriter. On average, 20
pages of single-spaced typed fieldnotes were compiled each day, yielding a total
corpus of slightly over 2,500 pages for the entire project.
Systematization. Like most participant observers, I cite the volume of fieldnotes as
evidence of sustained observation and as a not-so-subtle indicator that real work was
done. But while sustained observation and copious notes are necessary for adequate
fieldwork, in my opinion, they are not sufficient. In addition, observation needs to be
both systematic and explicit. Tactics for making observations systematic and explicit
are notoriously difficult to specify before fieldwork begins. Not only do the structures
of social settings vary, but in most cases researchers will be initially ignorant of how
the setting is organized. Nevertheless, as one gains familiarity with a setting, its
routines suggest avenues for structuring inquiry.
Because the synchronic thrust of the research aimed at discovering how the social
organization of work varied across different technologies, I required an observational
strategy that would allow a systematic charting of the tasks, roles, and role relations
associated with each modality. Preferably the strategy would also be relatively
unobtrusive and attuned to the rhythm of the each department's work. Shortly after
entering the field, I discovered that radiologists and technologists parsed their work
into what were variously called "procedures," "studies," or "examinations." A
procedure entailed using a particular device in a certain way to produce pictures of a
specific anatomical region. The flow of work through each technology, therefore,
consisted of a stream of examinations, each with a discrete beginning and end. The
work of technologists and radiologists revolved around an unfolding series of such
studies, and since most interactions between radiologists and technologists occurred
during examinations, procedures were the stage on which role relations were enacted.
For these reasons, procedures seemed to be an ecologically natural unit of observa-
tion for studying the social organization of each imaging device. During the third
week of the study, I began to use the occurrence of procedures to structure daily
observations.5
To systematize the data collection, I created the observational matrix displayed in
Figure 2 by cross-classifying roles and role relations by procedures. My strategy was
to observe multiple instances of the most frequent exams conducted with each
technology in order to fill the cells of the matrix with redundant descriptive data.
Once the matrix was complete, I could compare any procedure synchronically with
another. More importantly, by combining all procedures subsumed under each
technology, I could compare the use of different technologies. Because a scanner's
later use could be contrasted with its earlier use by examining how the particulars of
the data in CT columns evolved with time, the matrix also proved useful for collecting
diachronic information.
Since technologists and radiologists were usually assigned to one (or at most two)
technologies each day, I habitually attached myself to a radiologist or technologist in
the morning and then followed that person over the course of the day. Whom I
observed on any given day depended on two considerations: (1) which of the matrix's
cells required more data and (2) whether by choosing an informant I could observe a
procedure being performed by someone whom I had not repeatedly observed in that
context. As the second criterion implies, I attempted to observe all radiologists and
technologists who operated each technology in order to more easily distinguish social
regularities from individual habits and idiosyncracies.
Guided by the observational matrix and the study's triple comparative design, the
substantive focus of the investigation changed over the course of the year so that
discrete phases of observation emerged. The phases reflected the CT scanners'
arrival, decisions to alternate between collecting synchronic and diachronic data, and
variation in attention to the work of technologists and radiologists. Figure 3 arrays
the phases schematically along a time-line representing the study's duration. From
mid-June to late-September, I collected data for a synchronic comparison of tech-
nologies. During that time, observation shifted on a daily basis between radiography,
fluoroscopy, ultrasound, and special procedures. I attached myself exclusively to
technologists and observed the department from their perspective.
By late-September the scanners had been assembled, and by the first of October
both departments had brought their scanners on-line. From the end of September to
the end of December, I observed, almost exclusively, the body-scanner operations and
5Note that choosing procedures as a unit of analysis focused my attention in one direction rather than
the others that were possible. The choice seemed reasonable given my concern with behavior and
interactions. However, had I been primarily interested in mapping insider's interpretations, I would have
probably chosen another tactic for collecting data. For instance, I might have relied more heavily on
ethnosemantic interviews to construct linguistic taxonomies and semantic contrast sets.
*C
CD)
L)
CD 9,~~~~~~~~~~~~~~~~~~~~~~~~~~c
0~~~~~~~~~~~~~~~~~
L)
C,)
C [I_1__
o 9,
O- _ _
..
0.
(I)~~~~~~~l
Y(0 - . - - O -
~L9,. @l
C.)~ ~ ~ ~ ~ ~ ~~~~~~~~~~~C
cn~~~~~~~
o~ ~o. ~...C_
FOCI OF
OBSERVATION
Analytic Sjnchronic I
Hiatus
Primary Technologists from Radiologists Both
Attachment field
Technologies I
Observed All CT Scanners All
June Jul Aug Sept Oct Nov Dec Jan Feb Mar Apr May June
1982 1983
TIME
FIGURE 3. Phases of Observation.
their day-to-day development. During this period the analytic focus shifted from
synchronic to diachronic. The physical layout of the two CT departments and the
nature of the radiologists' duties when assigned to CT allowed me to observe the flow
of radiologists' and technologists' work more or less simultaneously. However, during
this period I continued to attach myself primarily to the technologists.
From Christmas to Valentine's Day I withdrew from the field to spend time with
the data I had collected. During that period, I spent a day at each site at two-week
intervals to maintain connections, to provide assurances that I would return, and to
keep abreast of major developments. In mid-February, daily observation of the
scanners resumed, but now primarily from the point of view of the radiologists. By
the end of March, I shifted from exclusively observing CT work to observing the
radiologists in other areas of the department.
From the first of April to the end of May, I once again adopted a synchronic
orientation and observed the entire range of procedures performed in the depart-
ments. However, rather than attach myself primarily to either technologists or
radiologists, I alternated back and forth between their company. While most of my
time was spent outside the CT areas, I logged at least two hours a day with the
scanners' personnel so that I could continue to chart developments. In general, this
rather complicated schedule of observations enabled me to collect data for a syn-
chronic analysis of all technologies, as well as a diachronic analysis of the two CT
operations, while capturing the perspectives of both radiologists and technologists. It
is critical to note that the observational schedule was devised during the course of the
study in light of issues of timing as well as imbalances in the data, both of which
would have been difficult to anticipate in advance.
Explicitness. Social scientists often draw rigid distinctions between qualitative and
quantitative studies. Methods texts ritualistically extol the former for their richness,
while advocating the latter for their explicitness and presumed replicability. Although
explicitness is perhaps easier to attain in survey or experimental research, there is no
reason to conclude, a priori, that explicitness is inimical to qualitative research. The
value of sustained observation is not simply its suitability for revealing patterned
complexities. Sustained observation also increases the odds that researchers will
uncover meaningful indicators of those patterns. After all, if participant observers
succeed in identifying patterned activities and interpretations, they must do so by
attending to repetitive cues.
Although it is impossible to document all facets of a setting that lead to ethno-
graphic insight, some can be isolated more or less precisely. In fact, one value of
longitudinal fieldwork is that it increases the likelihood of identifying setting-specific
cues. The reason is simple: repeated observation of repetitious actions or events
becomes boring. Not only is boredom a sure sign that repetition is taking place, but
the events that occasion boredom may represent useful setting-specific indicators.
The power of a setting-specific indicator is its capacity to make more explicit the
grounds on which one has drawn one's conclusions. By carefully documenting the
frequency of setting-specific indicators, researchers cannot only describe what hap-
pened, they can also prove that it did, in fact, occur.
The value of interplay between description and documentation in qualitative
research is often discussed, but rarely demonstrated.6 To be sure, researchers cannot
usually know ahead of time what will count as an indicator of the patterns they
discover while in the field. Nevertheless, one can devise indicators as the study
develops, presuming, of course, that the researcher stays in the field long enough to
assemble the indicators once they have been discovered.7 Two general tactics for
augmenting an ethnographic account with explicit, even quantitative, indicators
proved particularly useful in the course of my work: the keeping of behavioral records
and the use of nonobservational data.
Radiological work is fortuitously composed of visible physical and verbal behaviors.
The execution of procedures involves human action as well as the mechanical or
electronic responses of machines. Even the interpretation of films by radiologists
usually entails conversations among radiologists or between radiologists and referring
physicians. Given this wealth of observable behavior, after several weeks in the field I
settled on a general stance toward note-taking: to capture the chronological stream of
events, actions, and interactions that occurred over the course of an exam. Metaphor-
ically, I resolved, to the best of my ability, to become a human camera.8
From the beginning of the study I made no secret of the fact that I would be
constantly taking notes on what was going on around me. However, to reduce
intrusion, I worked with small spiral note pads that fit easily in palm of my hand.
During the course of a procedure, I jotted down a running record of as many actions
6
A number of researchers talk about "triangulation," the use of several methods to make findings more
believable (Webb et al. 1966, Jick 1979). But fewer researchers actively attempt to triangulate. Meyer's
(1982) study of hospitals facing an anesthesiologist's strike and Faulkner's (1983) work on the careers of
Hollywood composers are masterful examples. On tactics for making qualitative and ethnographic observa-
tions explicit, I have found the work of Michael Agar (1980, 1981), Miles and Huberman (1984), and Roger
Barker (1963, 1978) to be extremely useful. Barker's studies of the "stream of behavior" and his methods
for documenting it are classics of qualitative empiricism too often overlooked by modern fieldworkers.
7At this point it seems appropriate to make clear one of my pet peeves. The reason that so much of the
so-called "qualitative" research in organizational behavior is thin and of questionable worth, has little to
do with the fact that the research is qualitative in nature. Rather, the difficulty arises from the fact that
researchers have shied away from sustained observation. Three weeks worth of unstructured interviews
with a handful of respondents (usually top management) does not make for ethnographic
understanding-much less allow for the possibility of collecting data on setting-specific indicators. From
my perspective, when an ethnography does not involve extended periods of field work, it does not deserve
the name.
8Although my behavioral focus differs from that of most ethnographers who have been primarily
concerned with insider's interpretations, I certainly did not ignore interpretations. Not only did I record
interpretive comments as part of the note taking, but if participants did not verbalize their interpretations
of events spontaneously, I elicited them between procedures or when action was otherwise in a lull.
and bits of conversation as I could record. As part of the regimen, I noted the times
at which the procedure began and ended, as well as the times at which actors entered
and left the scene. Since all procedures took place in examination rooms, entrances
and exits were easily distinguished. When the timing of beginnings, ends, entrances,
and exits became habitual, my time keeping expanded to the duration of each
procedure's major phases. Once a procedure's repetitive aspects began to appear
routine, I devised a shorthand to document more parsimoniously the actions and
idioms that recurred across all instances of a particular exam.
Because my notes took the form of behavioral records, they provided a sufficiently
detailed database for supporting a number of explicit analyses. For example, it was
possible to calculate from fieldnotes the proportion of a procedure's duration during
which radiologists and technologists were in face-to-face contact. This particular
measure of interaction was used to substantiate an ethnographic account of role
relations among radiologists and technologists operating different technologies
(Barley 1990). Similarly, because the notes recorded who had made what decisions
during the course of each CT scan, it was possible to plot when, and at what rate,
decisions shifted from the domain of radiologist's work to the domain of technologist's
work (Barley 1986a). As a final example, because interactions were chronicled in
considerable detail, it was possible to identify the scripts that guided interactions and
to analyze the distribution of those scripts not only across technologies but, in the
case of the CT scanners, over time (Barley 1984, 1986a, 1990). One value of such
setting-specific indicators is that they can be analyzed using standard statistical
techniques. However, unlike most measures which are formulated prior to the study,
indicators captured in a chronology of fieldnotes enhance a study's ecological validity
because they remain close to the phenomenon under investigation.
In addition to recording countable behaviors, I also sought greater explicitness by
augmenting observational data with a number of observationally independent sources
of information. For example, even though I became adept at recording snippets of
conversation, it was physically impossible to capture on paper the complexity of
longer exchanges. To address this problem, I secured permission to tape record
exams during the last three months of the study. The taping was done with a small
micro-cassette recorder kept in the pocket of my lab coat or held beneath my memo
pad. The tapes were later transcribed and the transcripts were used to document
verbal parameters of role relations among radiologists and technologists.
After three months of observation, I compiled a set of sociometric questionnaires
based on topics of technical and social conversation common among radiologists and
technologists. The questionnaires were distributed to all members of the departments
in August, and again in May. The sociometric data served two purposes. First, they
were useful for verifying qualitative observations regarding role relations among
radiologists and technologists. Specifically, the sociograms confirmed that role rela-
tions in CT had, in fact, come to resemble those of ultrasound and special proce-
dures. Second, the sociometric questionnaires allowed a graphic representation of the
larger social network for each department and, hence, an examination of the global
structure that emerged from the various role dyads (Barley 1990).
Finally, both departments granted free access to the records they routinely com-
piled. Whenever a particular set of records seemed as if it might shed light on some
pattern, I arranged to raid the departments' archives. For example, duty assignment
sheets were particularly helpful for verifying how the new technologies had effected
the demographic distribution of radiologists' work (Barley 1990). Daily CT schedules
with their cancellations and "add-on's" offered measures of workflow and workload.
Other records allowed me to test hunches regarding the scanner's ability to substitute
for older radiographic procedures such as the myelogram. As a final example, data
field data more systematic, explicit, and hence, valid, replicable, and convincing.9 To
impart a flavor of how fieldnotes can be systematically analyzed, consider how the
scripts that characterized daily life in the two radiology departments were identified
and analyzed.10 The analytic activity progressed through four phases: (1) developing
categories, (2) grouping data, (3) identifying patterns or scripts, and (4) comparing
and contrasting scripts both synchronically and diachronically.
GroupingData. With a single category system in mind, I read and reread the entire
corpus of fieldnotes. Passages relevant for each category were noted by placing an
appropriate code in the margins. The notes could then be sorted by category. For
instance, all fieldnotes that described barium enemas were isolated and grouped
together in a file according to their order of occurrence. Similar files were compiled
for other procedures as well as for types of interactions and typical days. Because
grouping involved sorting the fieldnotes into separate files and because I worked
without the aid of a computer, I used a separate set of xeroxed fieldnotes for each
typology. The duplicates allowed me to cut and rearrange relevant sections of a day's
notes as the analysis required.
9I have found the following works particularly useful guides for conceptualizing the task of analyzing
field data: Glaser and Strauss 1967, Schatzman and Strauss 1973, Lofland 1971, 1976, Agar 1980, Spradley
1980, Burgess 1982, Miles and Huberman 1984.
l?As noted in the text, a script is fundamentally a structuralist notion. The reader should, therefore, note
that the analytic techniques reflect a decidedly formalist orientation to ethnographic research.
SUBURBAN URBAN
Phase I Phase 2 Phase I Phase 2 Phase 4
HOSPITAL PHASE UV AQ PS CT RR BT DG CM UC DS UCrit AcQ TC ME
Suburban 1 6 9 0 1 0 0 0 10 00 51 0 01 0 0
2 2 3 2 113 14 7 11 0 0 51 0 11 9 7
Urban i 0 0 t0I 0 0 1 47 12 12 21 10 21 0 0
2 0 0 1 1 0 0 2 14 1 7 13 111 61 0 ?
3 1 1
0 0 1 1 0 0 50 4 9 33 4 71 0 0
4 0 0 0 1 0 1 0 13 0 3 11 1 1 01 11 10
FIGURE 4. Distribution of Scripts in Urban's and Suburban's CT Operation During Specific Phases of
Each Scanner's Evolution.
Across the top of the table are arrayed the scripts characteristic of each hospital's
interaction order. Suburban's scripts are to the left of the solid vertical line drawn
through the body of the table, Urban's scripts are to the right. Each hospital's scripts
are further grouped by the phase of structuring where they appeared to be most
heavily concentrated in the fieldnotes. The rows of the table are ordered by hospital
and, within hospital, by the phases of structuring hypothesized for that hospital. A
solid horizontal line separates Suburban's and Urban's phases. Accordingly, each cell
of the table reports the number of times a specific script occurred in the fieldnotes
from a particular phase of structuring at a particular hospital.
Consider the two by two division of table formed the solid vertical and solid
horizontal lines. Cells in the upper left and lower right-hand quadrants of this two by
two (i.e., the main diagonal) report the occurrence of scripts in the hospital where
they were thought to be most characteristic. Cells contained in the off-diagonal
quadrants of the two by two represent the occurrence of one hospital's scripts in the
fieldnotes from the other. Examination of the table shows that Suburban and Urban
were characterized by different scripts since Suburban's scripts are more frequent in
the upper left-hand quadrant, while Urban's scripts are most frequent in the lower
right-hand quadrant. Moreover, it is also clear that within each hospital, different sets
of scripts did indeed characterize different phases of structuring. For example,
unsought validation (UV), anticipatory questions (AQ), and preference stating (PS)
were more common at Suburban during phase 1 than during phase 2. Similarly,
clandestine teaching (CT), role reversals (RR), and blaming the technologists (BT)
did not appear at Suburban until phase 2. By comparing the structural commonalities
of the various scripts at each hospital, it was then possible to indicate in what general
direction and at what rate the social order of the two CT scanners had evolved (see
Barley 1986a).
Critics often claim that the pitfalls of participant observation are primarily episte-
mological. The foregoing sections have shown that epistemic difficulties can be
managed. In sharp contrast to the rhetoric of most methods texts, I submit that
participant observation is far more likely to be tainted by emotional, social, and moral
difficulties. To be sure, such problems accompany all forms of research, but they
acquire special poignancy whenever researchers decide to use themselves as research
tools. Longitudinal observation only serves to intensity the complications, for the
longer one is in the field, the more the research task takes on the interpersonal
trappings of daily life.
Periods of intense pleasure and elation certainly punctuated my year in the
hospitals. However, I recall most vividly the darker emotions and the interpersonal
complexities that haunted my daily routine. I was acutely aware, from the very outset,
that the project's continuation rested largely on being tolerated, accepted, and even
liked by others. Each day, therefore, carried with it the nagging threat that I could
always be expelled from one or both sites for committing a faux pas. As a result, I
became more concerned with how others perceived me than I had been at any other
time in my life since junior high school. Such self-consciousness consumed consider-
able energy. Like an actor in Goffman's essays or a freshman at a college mixer, I
schemed, plotted, and worried constantly about how to present myself. Did that
comment sound insufferably academic? Would asking this question go beyond the
bounds of propriety? What effect would being friendly with this person have on my
relations with others? Should I laugh at racist or sexist jokes?
I resolved to manage such difficulties by taking cues from others and by engaging in
behaviors commonly valued in other walks of life. For instance, I attempted to be
congenial at all times. I avoided discussions of emotionally charged topics and
belittled my education lest I be perceived as self-important. When possible, I went
out of my way to do favors. I bought donuts, made coffee, and refrained from
speaking badly of anyone, including those who were widely disliked. Even though I
could not bring myself to laugh at racist or sexist jokes, I also did not confront their
tellers.
As I gained familiarity with the people at each site and as I learned more about the
technologies, I gradually began to take a more active part in the departments'
day-to-day routines. Although I was legally constrained from running the machines
and although I was wary of harming patients while trying to be helpful, I nevertheless
devised small measures of usefulness by assisting technologists with their "grunt
work." I routinely helped lift patients off gurneys and onto tables, held urinals for
feeble males, carried messages back and forth between technologists and radiologists,
and lugged countless film cassettes for those who needed them. After several months
of observing the CT scanners, I also found that I had learned a great deal about the
software. When technologists and radiologists discovered that they could use me as
"walking-talking" documentation, they did not hesitate to do so. My ability to be of
assistance reflected the fact that I had worked with computers, whereas most of the
technologists and radiologists had not. Moreover, I had observed problems encoun-
tered at both sites. In contrast, most technologists and radiologists knew how to cope
only with those glitches that had plagued their own scanner.
The Emergence of Trust
Field workers often wax poetic on the necessity of gaining the trust of one's
informants. Many suggest that, as in everyday life, trust and acceptance are largely
based on what the researcher is able to do honestly for others. While give and take is
undoubtedly important, in my experience trust cannot be easily managed and is only
partially contingent on familiarity and congeniality. In retrospect, any acceptance I
gained appears to have been largely determined by events and perceptions over which
I had little control.
For instance, building trust and friendship at Urban was exceptionally difficult
because of a long history of conflict within the department. Urban's x-ray technolo-
gists, at least initially, regarded me with intense suspicion and some hypothesized that
the administration had sent me to spy on them, a perception never voiced, even
indirectly, at Suburban. In fact, Urban's administrator had made overtures about my
consulting with him on managerial issues. During the first several months I, therefore,
had considerable difficulty engaging technologists in conversation unless I asked them
direct questions to which I usually received curt, and not always polite, replies.
Although I cannot pinpoint exactly when relations began to unfreeze, I suspect the
x-ray technologists gradually lowered their guard for four reasons. First, I studiously
avoided the department's administrator and made certain that we spoke only on
public turf so that witnesses could overhear our conversations. Second, in the first
days of the study I developed good relations with the technologists who ran Urban's
head scanner.1" Because of a staff shortage in early July, the CT technologists were
forced to work briefly in the x-ray department. Their demonstrations of acceptance
during this period may have worked to my advantage. Third, I had several opportuni-
ties to express verbal solidarity with technologists after witnessing particularly nasty
encounters involving technologists and radiologists. However, the most important
factor was undoubtedly my relationship with Urban's sonographer, who was tied to
the dominant clique of x-ray technologists and who went out of his way to vouch for
me with his friends. Once the members of the dominant clique became more open,
most of the other x-ray technologists followed suit.
Gaining the support of key insiders also proved critical at Suburban. Although far
more hospitable than their counterparts at Urban, Suburban's technologists also
remained aloof for several months. Relations with the technologists were transformed
almost overnight by a completely serendipitous event. A patient from a neighboring
hospital was to have a nephrostomy at Suburban. The procedure had been performed
twice before, but each time the catheter had become dislodged. The third nephros-
tomy would require a catheter of unusually large diameter to prevent slippage.
Suburban had only one such catheter on hand.
A specials tech, who was also the most central person in the department's network,
had prepared the catheter the night before the procedure was scheduled. However,
an hour before the nephrostomy was to begin, he discovered that a janitor had
thrown the catheter in the trash. After a number of phone calls the technologist
located a replacement at a hospital in Boston, but because he was unfamiliar with the
city, he feared he would not be able to get to the hospital and back within an hour.
Because I knew where the hospital was located and because I also owned a relatively
fast car, I offered to drive him. We succeeded in making the round trip in 45 minutes
without encountering the highway patrol and without the radiologists ever learning
that the catheter had been missing. Thereafter, I had no difficulty getting any of
Suburban's technologists to cooperate.
11Although neither Urban nor Suburban had previously used a body scanner, Urban had been operating
a head scanner since 1977.
ogists to be distrustful, distant, cynical, and at times, openly hostile. In fact, the
atmosphere at Urban was so distasteful that I eventually had to bolster my resolve to
return on a daily basis. On many a morning I found myself driving aimlessly around
the streets of Boston for as long as an hour before I succeeded in talking myself into
the hospital's parking lot.
The longitudinal component of the study brought with it another set of unantici-
pated anxieties and predicaments. I was constantly plagued by the fear of overlooking
important social dynamics. The desperate and irrational thought that I might com-
plete the study empty-handed led me to spend far too many hours questioning the
"real" meaning of what I had seen or heard. Halfway through the project, boredom
reared its head. I began to think that I had seen all there was to see. From time to
time, I even found myself nodding off as I watched exams or listened to conversa-
tions. On the heals of boredom, however, came guilt which renewed my anxiety about
overlooking important details. Thus, a spiral of complacency and doubt arose. The
discipline of writing analytic memos proved crucial for managing these difficulties as
did periodic rump sessions with my colleagues and fellow ethnographers: Edgar
Schein, John Van Maanen, and Gideon Kunda.
A final difficulty, partially associated with the experience of boredom, was a
phenomenon that anthropologists term "going native." The primary intellectual risk
of long term participant observation is that one may eventually come to view the
setting entirely from an insider's perspective. As a result, situated behaviors and
interpretations begin to appear normal, unsurprising, and commonsensical. When
going native, ethnographers forget that they must remain, above all else, a "profes-
sional stranger" (Agar 1981).
By the end of the year I had certainly gone native on several fronts. Like the
radiologists and technologists, I had long become numb to patients and their
emotional outbursts. I not only regarded patients almost entirely as biological
organisms, but I even began to appreciate how such emotional armor could allow one
to execute tasks that might otherwise be too disturbing to perform. For instance, like
my informants, I ceased to question how I could watch a barium enema performed on
an elderly, groaning patient, while talking nonchalantly about what I had eaten for
lunch. The upshot of my conversion was that I now have little of sociological
importance to say about the patients I observed.
Finding a Niche
Thus, the underbelly of long-term participant observation partially consists of the
effects of the research act on the researcher. The remainder involves the researcher's
influence on the setting. With the possible exception of archival research, all studies
probably affect the people who are studied to some extent, even though researchers
often pretend otherwise. Nevertheless, the probability seems particularly acute during
an ethnography. By becoming a participant observer, one creates for the ecology of
an organization a role that did not exist before: namely, the role of the ever-present
and ever-curious observer. Aside from allowing one's research to be used by manage-
ment as part of an organizational intervention, it is difficult to imagine how a research
act could be more intrusive for a longer period of time.
During the last week of the study, I brought a camera to both departments intent
on capturing mementos of the previous year. Taking pictures instantly became a
farewell rite between myself and my informants. A number of technologists and
radiologists even offered to take my picture after I took theirs. I accepted on the
condition that they pose me in a stance typical of my observing. As my informants
directed me on where to stand and how to hold my body, I found their attention to
detail alarming. For instance, several of the technologists refused to take my picture
until I had taken the cap off the top of my pen and placed it in my mouth. They
realized, as I did not, that when observing I had often used my teeth to pull the cap
off my felt-tip and that I habitually held the cap in my mouth. All of these
photographs show me standing in a corner of an examination room (or directly
behind technologists seated at a CT-scanner) holding a pen in my right hand and a
notepad in the left. For every picture, I was instructed to write. What, then, can be
said about a role which appears, as I look at these photographs, so incongruent with
its surroundings?
Although I made no secret of my intentions, my notetaking was certainly more than
a little suspicious, especially at first. Early on, a few of the bolder radiologists and
technologists ventured to ask what I was writing. Not wishing to appear secretive, I
decided to offer my notebook to whoever expressed interest in seeing what I had
written. Since most of my notes pertained to the flow of action, those who accepted
my offer expressed surprise that I had recorded the obvious and that I could find such
pedestrian matters interesting. Nevertheless, most technologists and radiologists were
initially quite self-conscious whenever something even slightly embarrassing occurred
in my presence. On these occasions, they made jokes ("Steve, did you get all that
down?") accompanied by nervous laughter, of which the most nervous was probably
my own. I resolved early to take no notes when events turned to the untoward or
when I sensed that a compromising situation was about to unfold. Although I might
later jot down a summary of these events while secluded in a bathroom stall, in doing
so I remained acutely aware that I should not be seen as someone after the
department's "dirt."12
I prefer to think that with time, most technologists and radiologists made peace
with my notebook, even if they did not completely forget its presence. On more than
one occasion, technologists took the pad from my hand and wrote down events that I
had not myself noted. In all cases, this behavior followed a stream of events that I
had been studiously attempting to ignore. Similarly, after several months, radiologists
routinely informed me of events that I had not witnessed, but which they thought I
would find interesting. It therefore seems reasonable to conclude that over the course
of a year my presence as observer became more or less routine for the technologists
and radiologists with whom I interacted on a daily basis. In fact, contrary to popular
opinion, one reason for pursuing fieldwork longitudinally is that it actually enables
researchers to get beyond presentational rhetorics and behavioral shows. People
simply find it difficult to monitor their behavior or to dissemble for an entire year.
Consequently, there is little reason to believe that my presence had a deleterious
effect on my informant's lives or that my observations skewed their behaviors. But
what of the patients whom I rarely encountered twice? How did they understand my
presence?
Both radiology departments accepted student technologists from local training
programs. Interns were common at Urban and from the perspective of a patient,
interns would not have been out of place at Suburban. Without planning it to be so,
my implicit cover became that of "student," although neither I nor anyone else
mentioned what kind of student I might be. In other words, without solicitation,
technologists and radiologists gave me a situational identity so that I could blend into
121 was also conscious that my fieldnotes could be subpoenaed should a patient decide to sue a
radiologist for malpractice. For this reason, I avoided recording any information on the interpretive
mistakes that radiologists made. I also took no notes on procedures performed on lawyers, an occupation
of whose members physicians appear to be justifiably wary. In fact, the radiologists were themselves
especially careful whenever they encountered a patient who was known to be a lawyer or a lawyer's wife.
the setting as a plausible anomaly. At Suburban, I was given a white lab coat identical
to those worn by doctors and male technologists. The technologists always introduced
me to patients in a manner that suggested my right to be in the situation. If patients
directly questioned my identity, technologists would usually respond before I had the
chance: "He's trying to learn what we do," "That's just Steve, he's learning the
scanner with us," and so on. In point of fact, however, patients rarely questioned my
identity (or for that matter, anything else). Since the technologists and radiologists
treated me as if I belonged, so did most of the patients. Being a pragmatist, I was
more than willing to take advantage of any definition of my role that might felicitously
arise.
In general, then, I do not believe that my presence skewed the social dynamics of
the departments or that the act of observation led to filtered information. At the
same time, one should not conclude that my presence had no implications or that my
role was merely that of a fly on the wall. Instead, the social ramifications of long-term
participant observation may simply be more subtle than most critics suggest. In fact, it
seems that the most significant effects of my presence were fashioned less by the
research act, than by where I became situated in each department's social ecology.
For like all other roles, the role of participant observer is fashioned by how the
researcher's presence meshes with an ongoing stream of interaction to create a niche
more or less unique to the context in which the observer is embedded.
Political Alignments
The phased nature of my observational strategy highlighted the nature of my role
as well as what may be the most significant hazard associated with all forms of field
research. Because I had initially approached the study from the technologist's vantage
point, I found switching to the radiologist's perspective difficult. However, the
difficulty was not simply a matter of my inability to change frames. Instead, it resulted
largely from how my presence had been defined by others. When settings contain
multiple groups with potentially conflicting interests, one is inevitably perceived as
being more aligned with one group than another. At the time of the switch, I had not
only yet to establish the type of rapport that I had developed with the technologists,
but more importantly, I had not yet overcome the radiologist's perception that I was
primarily interested in and aligned with the technologists.
A similar transition occurred when I returned to the synchronic strategy at the end
of the year. The x-ray technologists were suspicious of the fact that I had spent so
much time with the CT technologists and the radiologists, whom they perceived as
having interests different than their own. Fortunately, by this time I knew the
technologists well enough to confront the situation head-on and to explain my
position both as a researcher and as a person. The stance appeared to work in that
during the last two months I easily moved from one group to another without
apparent ill feelings. However, had I not been engaged in longitudinal research and
had I not consciously chosen to view the department from the perspective of different
groups, I would probably not have become cognizant of the political nature of the
research act.
Given the salience of political analysis in recent organizational theory, it is
surprising that most methods courses completely overlook the political implications of
doing research. My experience suggests that in stratified organizations informants are
quick to ascertain with whom the researcher appears to side. Moreover, it is
extremely difficult to alter the definition of one's role once an alignment has been
drawn. Such definitions undoubtedly influence the nature of the cooperation and
information that researchers receive. The unfortunate fact is that most organizational
researchers align themselves with one faction or the other unwittingly and fail to
appreciate how the alignment may affect their data.13 One of the most important
advantages of longitudinal field work may be that extensive emersion in an organiza-
tion enables one to better appreciate, and perhaps counteract, political definitions of
the researcher's role.
From the very beginning I sensed that technologists and radiologists were watching
to see if I would pass potentially sensitive information to other persons in the
department. Whether or not my perception was accurate, I cannot say for sure.
Nevertheless, I was careful never to repeat what technologists or radiologists told me
unless specifically asked to carry a message or unless the information was obviously
meant to be public (i.e., "Don't use room two because the film changer is jammed"). I
can remember only one occasion on which I broke my self-imposed vow of confiden-
tiality.
In late spring, the cardiologists at Suburban attempted to usurp cardiac ultrasound.
I had been informed of the cardiologists' intentions by sonographers and radiologists
alike. However, several radiologists did not know that I knew about the turf battle
even though I thought that my knowing was common knowledge. Under this assump-
tion, I asked two radiologists a rather pointed question regarding the progress of the
conflict. One immediately demanded to know how I knew about the situation. Taken
aback, I named another radiologist as the person who had informed me. At first I
thought I had betrayed my informant. However, as it turned out, the radiologist was
glad to hear that it had not been a cardiologist who had told me. Afterward the
radiologist became more open with me than he had been up to that point in time. His
openness led to the discovery that he had known my father-in-law while he was a
resident. My faux pas, thus, ironically led our relationship onto firmer ground.
With time, I believe I came to be seen as a harmless, but perhaps eccentric,
individual without factional loyalties. In fact, judging by the actions of the technolo-
gists and radiologists themselves, I would venture that my situated identity became
that of a trusted neutral. For instance, after Christmas several of Suburban's technol-
ogists came to speak to me about difficulties they were having with one of their
colleagues. They claimed that they had decided to talk with me because they were
certain I "wouldn't talk." A more far dramatic incident occurred at Urban.
Near the end of the study, I witnessed an instance of extreme hostility between a
technologist and a radiologist which resulted in the radiologist calling the technologist
a number of unflattering names. The technologist, who felt that she had quietly borne
a long history of affront, was quick to recognize that my presence meant she could
successfully lodge a formal grievance, if I were willing to testify on her behalf. The
technologist lost no time in asking for my aid. After a moment of agony, I decided
that I could not risk compromising my position as researcher. I therefore lamely told
the technologist that I could not assist her because I could not attribute blame or
malicious intent when I had no access to the radiologist's thought processes. The
technologist became justifiably indignant, and I feared that my refusal would prove
costly.
To my surprise and great relief, on returning to the department four days later, I
was told that in my absence the technologists had discussed the issue of how I should
13Because of my experience, I now have serious reservations about the accuracy of any piece of research
conducted by researchers who even remotely engage in what might be construed as a consulting role or
who align themselves with the management of an organization. Perhaps it is time organizational re-
searchers gave as much attention to the biases of their alignments as they do to the biases of their methods.
I fear that we may actually find the former to be more epistemologically damaging than the latter.
have responded. While one or two felt I should have sided with the technologist, the
consensus had been that I had all along been a neutral party and that I should not be
expected to sacrifice that status now. My informant told me that to do so, the
technologists had decided, would jeopardize my work. God bless them!
ConcludingComments
Although longitudinal field research seems particularly well suited for studies of
the social ramifications of new technologies and perhaps even most organizational
change, like all research methods, mine suffer from biases and limitations that should
be made explicit. First and foremost, my approach to ethnographic field research is
decidedly structuralist in orientation and realist in tone (for an elaboration of these
genres, see Van Maanen 1988). Nowhere have I sought to. offer an ethnographic
account of the "native's point of view." Instead, my methods and tales are weighted
toward what an anthropologist would call an "etic" or formal analysis. Had I been
interested in writing a description of native viewpoints, I would have focused on other
types of data and would have chosen other analytic and descriptive techniques.14 It
would, therefore, be a mistake for readers to treat my work as paradigmatic of all
ethnographic research. Ethnography has many faces. I have described but one.
My approach to fieldwork is also rather costly, at least in terms of time. From
beginning to end, the project took two and a half years to complete. I am still
analyzing some of the data that I collected in 1982. Long-term observational field-
work is not something that can be done on an occasional afternoon. It requires a
considerable commitment of energy, a high tolerance for ambiguity, and in some
cases, a strong stomach. Moreover, the ethnographic stance is, in general, incompati-
ble with the intent to prescribe or intervene, a stance that so frequently characterizes
research in organizational behavior. As on the voyages of the Starship Enterprise, the
watchword of careful ethnographic research is the "Prime Directive."
At the same time, however, organizational studies might profit both theoretically
and empirically, if more researchers were to embark on observational travels framed
by an ethnographic sensitivity. For this to occur, we will surely need to change the
way we reward our students and ourselves. The study that I have described was
conducted as a doctoral dissertation. As a graduate student, I had few organizational
constraints on my time and I experienced no pressure to crank out articles for tenure
(or for that matter, to complete my degree by an arbitrary date). At first glance,
conducting a longitudinal ethnography would seem to be more difficult for academics
carrying a full time teaching load in institutions where the sheer number of annual
publications is used as a blunt measure of reward. Nevertheless, such obstacles can be
overcome by juggling one's schedule creatively to allow solid blocks of time for
research. Far more limiting are geographical constraints. Suburban and Urban were
located within half an hour's drive of my residence. Had I been located in a less
urban area, the study would have been far less tractable.
Finally, if I had it to do over again, I would probably do the study differently. For
example, I can imagine (1) employing coders to help analyze data, (2) using Abbott's
(this volume) techniques for analyzing sequences to study temporal structures, or (3)
taking on a partner and doing a team ethnography. I would have also begun my tape
recording much earlier. If I were undertaking the study today, I would also undoubt-
edly use a microcomputer and a word processing program instead of a typewriter to
14Those interested in learning more about emic ethnographies might consult the work of Dalton (1951),
Geertz (1973, 1983); Douglas (1966, 1975), Spradley (1980), Gregory (1983), and Kunda, (forthcoming).
compile my fieldnotes. There now exist numerous software packages that enable field
researchers to code portions of a text file and then reassemble those coded sections in
a separate database. Several of these packages are commercially available (e.g., Note
Bene, Go Ask Sam). Others have been explicitly written for and by ethnographers
(Shelly 1988). The ability to sort, search, merge, and scan text files electronically
seems far more efficient than the manual methods on which I relied. In fact, such
programs should enable a researcher to rapidly compile counts and distributions and
then subject these numeric data directly to statistical analysis.
Note, however, that each of these "improvements" aim at making the observation
and the analysis even more explicit, systematic, and credible. As such, they bow to
empiricist, though not necessarily positivist, canons. While I am personally most
comfortable with an empiricist's stance, there are insights to be gained from ap-
proaches more interpretive, and more lyrical than my own. In fact, measured against
such standards, my own work is little more than a hack. Caveat emptor.
Acknowledgements
The research discussed in this paper was supported, in part, by a doctoral
dissertation grant from the National Center for Health Services Research (HS 05004).
Collegial readings by Gideon Kunda, Pam Tolbert, Andrew Van de Ven, and several
anonymous reviewers were valuable in checking my excesses.
References
AGAR,MICHAEL (1980), "Getting Better Quality Stuff: Methodological Competition in an Interdisciplinary
Niche," Urban Life, 9, 34-50.
(1981), The Professional Stranger, New York: Academic Press.
BANTA, DAVID AND CLYDE J. BEHNEY (1980), "Medical Technology Policies and Problems," Health Care
Management Review, 5, 45-52.
AND BARBARA J. McNEIL (1978), "Evaluation of the CAT Scanner and Other Diagnostic
Technologies,," Health Care Management Review, 3, 7-19.
BARKER, ROGER G. (1963), The Stream of Behavior, New York: Appleton Century Crofts.
(1978), Habitats, Environments, and Human Behavior, San Francisco: Jossey Bass.
BARLEY, STEPHEN R. (1984), "The Professional, the Semi-Professional and the Machines: The Social
Ramifications of Computer Based Imaging in Radiology," Doctoral Dissertation. Sloan School of
Management, MIT.
(1986a), "Technology As an Occasion for Structuring: Observations on CT Scanners and the
Social Order of Radiology Departments," AdministrativeScience Quarterly,31, 78-108.
(1986b), "Changing Roles in Radiology," AdministrativeRadiology, 5, 32-41.
(1988a), "The Social Construction of a Machine: Ritual, Superstition, Magical Thinking and
Other Pragmatic Responses to Running a CT Scanner, in M. Lock and D. Gordon (Eds.),
Biomedicine Examined, Dordrecht, The Netherlands: Kluwer Academic Publishers, 497-539.
(1988b), "On Technology, Time, and Social Order: Technically Induced Change in the
Temporal Organization of Radiological Work, in F. A. Dubinskas (Ed.), Making Time: Ethnogra-
phies of High TechnologyOrganizations,Philadelphia, PA: Temple University Press, 123-169.
(1988c), "Technology, Power, and the Social Organization of Work: Towards a Pragmatic
Theory of Deskilling, in N. DiTomaso and S. Bacharach (Eds.), Research in the Sociology of
Organizations,Volume 6. Greenwich, CT: JAI Press, 33-80.
(1990), "The Alignment of Technology and Structure through Roles and Networks," Adminis-
trative Science Quarterly,35, 61-103.
BECKER, HOWARD S., BLANCHE GEER, EVERETT C. HUGHES, AND ANSELM L. STRAUSS (1961), Boys in
White: Student Culturein Medical School, Chicago: University of Chicago Press.
BELL, DANIEL (1973), The Coming of the Post-IndustrialSociety, New York: Basic.
BERGER, PETER L. AND THOMAS LUCKMANN (1967), The Social Construction of Reality, New York:
Doubleday.
BURGESS, ROBERT G. (ED.) (1982), Field Research: A Sourcebook and Field Manual, London: George Allen
& Unwin.
DALTON, MELVILLE (1951), "Informal Factors in Career Achievement," American Journal of Sociology, 61,
407-415.
DOUGLAS, MARY (1966), "Purity and Danger: An Analysis of the Concepts of Pollution and Taboo,"
London: Routledge Kegan Paul.
(1975), ImplicitMeanings:Essaysin Anthropology,
London: Routledge Kegan Paul.
FAULKNER, ROBERT R. (1983), Music on Demand: Composers and Careers in the Hollywood Film Industry,
New Brunswick: Transaction Books.
FENSHAM, PETER J. AND DOUGLAS HOOPER (1964), The Dynamics of a Changing Technology, London:
Tavistock.
FREIDSON, ELIOT (1970), Professional Dominance: The Social Structure of Medical Care, New York:
Atherton.
_ AND JUDITH LORBER (1972), Medical Men and Their Work, Chicago: Aldine.
GEERTZ, CLIFFORD (1973), The Interpretationof Cultures, New York: Basic.
(1983), Local Knowledge: Anthropology,
FurtherEssaysin Interpretive New York: Basic.
GIDDENS, ANTHONY (1984), The Constitutionof Society, Berkeley, CA: University of California Press.
GLASER, BARNEY AND ANSELM STRAUSS (1967), The Discovery of Grounded Theory: Strategiesfor Qualitative
Research, Chicago: Aldine.
GOFFMAN, ERVING (1959), Presentation of Self in EverydayLife, New York: Anchor.
(1967), InteractionRitual,New York: Anchor.
(1983),The Interaction Order, AmericanSociologicalReview,48, 1-17.
GREGORY, KATHLEEN L. (1983), "Native-View Paradigms: Multiple Cultures and Culture Conflicts in
Organizations," AdministrativeScience Quarterly,28, 259-377.
HALL, EDWARD T. (1969), The Hidden Dimension. New York: Anchor.
HERBST, P. G. (1962), Autonomous Group Functioning, London: Tavistock.
HUGHES, EVERErr C. (1958), Men and the Work, Glencoe: Free Press.
(1972), TheSociologicalEye, Chicago: Aldine.
JICK, TODD D. (1979), "Mixing Qualitative and Quantitative Methods: Triangulation in Action," Adminis-
trative Science Quarterly,24, 602-611.
KUNDA, GIDEON (Forthcoming), "Engineering Culture: Culture and Control in a High Tech Organization,"
Philadelphia: Temple University.
LARKIN, GERALD (1978), "Medical Dominance and Control: Radiographers in the Division of Labor,"
Sociological Review, 26, 843-858.
(1983), Occupational Monopolyand ModemMedicine,London, Tavistock.
LOFLAND, JOHN (1971), Analyzing Social Settings: A Guide to Qualitative Observationand Analysis, Belmont,
CA: Wadsworth.
(1976), Doing Social Life: The Qualitative Study of Human Interaction in Natural Settings, New
York: Wiley.
MANN, FLOYD C. AND LAWRENCE K. WILLIAMS (1961), "Observations on the Dynamics of a Change to
Electronic Data-processing Equipment," Administrative Science Quarterly,5, 217-256.
MECHANIC, DAVID (1968), Medical Sociology: A Selective View, New York: Free Press.
MEISSNER, MARTIN (1969), Technologyand the Worker,San Francisco: Chandler.
MEYER, ALAN (1982), "Adapting to Environmental Jolts," AdministrativeScience Quarterly,27, 515-532.
MILES, MATTHEW B. AND A. MICHAEL HUBERMAN (1984), Qualitative Data Analysis: A Sourcebook of New
Methods, Beverly Hills, CA: Sage.
OFFICE OF TECHNOLOGY ASSESSMENT (1978), Policy Implications of the Computed Tomography (CT)
Scanner. Washington, D.C.: Congress of the United States.
OFFICE OF TECHNOLOGY ASSESSMENT (1981), Policy Implications of the Computed TomographyScanner: An
Update, Washington, D.C.: Congress of the United States.
RICE, A. K. (1958), Productivityand Social Organization: The Ahmebabad Experiment, London: Tavistock.
SCHEIN, EDGAR H. (1985), Organizational Cultureand Leadership, San Francisco, CA: Jossey Bass.
SCHATZMAN, LEONARD AND ANSELM L. STRAUSS (1973), Field Research: Strategiesfor a Natural Sociology,
Englewood Cliffs: Prentice Hall.
SHELLY, ANNE L. (1988), "Analytic Strategies and Computer Assistance as Tools of Qualitative Re-
searchers," unpublished paper, School of Computer and Information Science, Syracuse University.
SPRADLEY, JAMES P. (1980), Participant Observation, New York: Holt Rinehart and Winston.
STOCKING, BARBARA AND STUART L. MORRISON (1978), The Image and the Reality. London: Oxford
University Press.
STRAUSS, ANSELM (1978), Negotiations. San Francisco, CA: Jossey-Bass.
, LEONARD SCHATZMAN, RUE BUCHER, DANUTA EHRLICH AND MELVIN SABSHIN (1963), in Eliot
Freidson (Ed.), The Hospital in Modern Society, New York: Free Press, 147-169.
________ ________ ______ _ _ _ AND (1964), PsychiatricIdeologies and Institutions,
Glencoe, IL: Free Press.
TRIST, E. L. AND K. W. BAMFORTH (1951), "Some Social Psychological Consequences of the Longwall
Method of Coal Getting," Human Relations, 4, 3-38.
VAN MAANEN, JOHN (1979), "The Fact of Fiction in Organizational Ethnography," AdministrativeScience
Quarterly,24, 539-550.
(1988), Tales of the Field: On WritingEthnography,Chicago: Chicago University Press.
WEBB, EUGENE J., DONALD T. CAMPBELL, RICHARD D. SCHWARTZ AND LEE SECHREST (1966),Unobtrusive
Measures: Nonreactive Research in the Social Sciences, Chicago: Rand McNally.
WIENER, H. (1979), Findings in CT: A Clinical and Economic Analysis of Computed Tomography, New
York: Pfizer.