What Is a Clinic? Relationships and the Practice of Organizational Ethnography

Introduction: Topics and Methods of Organizational Ethnography

Organizations are a pervasive and vital part of contemporary social life. They constitute both the foreground and the background of our lives. We specifically seek out interactions with organizations to help us with an astonishing variety of tasks, along the way demonstrating considerable sophistication about variations among organizations in what they can do and how well. We praise companies that we regard as efficient, virtuous, or stylish and deplore others that seem to us incompetent, unethical, too powerful, or simply boring. At the same time, though, we are blissfully ignorant of the countless organizations that create and maintain the infrastructure that shapes many aspects of our lives. Which of us knows how and by whom product standards are created and policed, the Internet maintained and regulated, chargemaster rates set in hospitals, or currency rates adjusted? Or for that matter, before the publication of “The Lonely Death of George Bell” (Kleinfeld 2015), who knew just how many organizations were involved in the protracted processing of the deaths of socially isolated people?

Some parts of the social world simply cannot be fully understood without a thorough analysis of organizations, organizational ethnographers would insist. We cannot truly understand what happens in medicine only by looking at doctors, nurses, and patients, for instance. To complete the picture, we also need to study the clinics and hospitals that provide a framework for the encounters among these participants and to think about how these institutions both shape and are shaped by key actors. Often, we will need to do even more, because we will need to know how clinics or hospitals fit into a larger system that also includes other hospitals and clinics, medical schools, research facilities, pharmaceutical companies, regulatory bodies, insurance companies and other third-party payers, and the systems of conventions, rules, regulations, laws, and background assumptions that bind them together. Similarly, to understand education, we need to study schools, testing services, and curriculum suppliers as well as teachers, students, administrators, and ancillary staff. And we cannot fully understand soldiers without investigating military organizations and conventions of warfare; the working class without also looking at the labor process, the flow of capital, workplaces such as factories, associations such labor unions or business associations; or the first-line workers who granted subprime mortgages without also knowing about the complex institutional arrangements of lenders they worked for.

It is hard, then, to study any social phenomenon without being mindful of the organizations that form its backdrop and often organizations themselves will seem a worthy focus of research. Organizational ethnographies have given us a window on this vital but changing part of the social landscape. From the early days of the ethnographic research on the Hawthorne Project’s bank wiring observation room (Roethlisberger and Dickson 1947), through the classic works of Whyte (1948), Selznick (1949), Gouldner (1954), Roy (1959), Dalton (1959), Goffman (1961), and Blau (1963), to mention but a few, sociologists have used ethnographic methods to study organizations themselves, the activities they house, the people who inhabit them, and the gaps between formal and informal aspects of organizational life in an astonishing array of organizations including: factories, public relations firms, financial institutions, schools, universities, publishing houses, hospitals, clinics, mental institutions, nursing homes, government bureaucracies, police departments, courts, hotels, restaurants, and fast-food chains. ¹

Although these studies of organizations are often similar to other forms of ethnographic work in their emphasis on close observation and the importance of understanding the meaning of actions, words, and artifacts, they differ from other kinds of fieldwork in what they problematize. Much ethnographic work is focused on describing and explaining individual-level phenomena, such as the behavior and beliefs of certain categories of persons, and the patterns of interactions between individuals (think, for example, of Stack 1974; Duneier 2000; or Goffman 2014). To be sure, ethnographers who study groups and other collectivities, whether they be families (Hochschild 1990; Lareau 2003), neighborhoods and friendship groups (Whyte 1965; E. Anderson 1978; Pattillo 1999), or groups engaged in leisure activities (Fine 1998), are often interested in supraindividual phenomena but of a less formalized sort than ethnographers of organizations. Although nearly all ethnographers are attentive to context, the context itself is not as often the object of study. In contrast, organizational ethnographers often are studying the organization itself, not just what happens inside it. But, in the view of organizational ethnographers, the added formality and complexity are deeply consequential. Formality and complexity transform organizations into distinctive social formations whose unique properties need to be described and explained. Without careful ethnographic study, we have no way to know what it means that our social world is now populated with entities with complex internal structures, opaque relations among parts or between the parts and the whole, extended life spans, or immunity to many kinds of rewards and sanctions. Moreover, following Eisenhardt’s (1989) lead on the importance of contrasting cases, I suggest that the very extremity of formalization and complexity in organizations makes organizational ethnography a useful point of contrast for other fieldworkers. Training our focus on the effects of formalization and complexity in geographically and interactionally constrained spaces brings into sharp relief relational dynamics present in ethnographic research more generally.

In suggesting that increases in formality and complexity bring categorical shifts in social forms and that we should study organizations qua organizations, organizational ethnographers break with some of the truisms of ethnography. Grappling with the problem of how to move from “local truths to general visions,” Geertz famously argued that “The locus of study is not the object of study. Anthropologists don’t study villages (tribes, towns, neighborhoods…); they study in villages” (1973:21-22). But I think he overstates the point, especially as it applies to organizational ethnographers. Organizational ethnographers don’t just study in organizations; they in fact study organizations, both as social forms important in contemporary societies and, sometimes, as particularly influential individual institutions that have shaped large swaths of social life. This distinction between the warrants of organizational ethnographers and the warrants of other ethnographers is of course a matter of degree rather than bright lines—some organizational ethnographers study in organizations just as some anthropologists study the villages themselves, and many are interested in organizations (or villages) as both entities and as contexts.

But, however, important it is for ethnographers to study organizations—either as context or as a specific object of study—the methods by which this should be accomplished need scrutiny. As sociologists are quick to acknowledge, because organizations are not unitary entities, they cannot be studied using exactly the same methods used to research individuals and individual-level processes. Yet the alternative to organizations as individuals is not entirely obvious. Clearly, organizations are composed of semiautonomous actors with a variety of interests that are not entirely consonant. At the same time, though, it would be folly to suggest that an organization is merely the sum of those individual parts. Organizations also have technologies, product lines, routines, contracts, affiliations, identities, histories, and cultures, some of which are properties of the organization as a whole. Given these differences between organizations, on the one hand, and individuals, groups, and other collections of individuals on the other, it behooves us to rethink the methods ethnographers employ to study organizations.

My argument is that the relationships ethnographers form in the field fundamentally shape the nature and the quality of evidence they are able to gather. That much is surely true of any group of ethnographers. But organizational ethnographers confront a special challenge in sorting out who or what they should form relationships with. An organization is not simply its top brass, so ethnographers cannot get to know the organization by forming ties only with its leaders. But neither is the organization essentially the members of any other subgroup, such as the individuals it employs, sells to, or serves. Rather, it is more than the sum of these parts in the sense that the relations among elements are mediated by the organization’s mission, culture, history, policies, and structure. Organizational ethnographers, therefore, must distribute their attention differently than other ethnographers do. They need to know both more and less than other ethnographers: more, because they need to understand the organizational context and how organizational culture and policies shape what happens in the organization; less, because they are interested in people primarily as organizational participants whose interactions are inflected and limited by the organizational mission.

But how does an ethnographer relate to the supraindividual parts of the organization? Although in some senses, organizations live through their artifacts and policies, not all of these elements are equally important. Some parts of an organizational history are actively referenced and influential, for instance, while others are dormant but still available for use in the future. Ethnographers mainly know the specifically organizational elements through what people tell them about the organization and its policies, forms, documents, and artifacts. Artifacts usually are accessed through individuals and typically will need to be interpreted and contextualized by individuals. Thus, an ethnographer relates to the organization by forming relationships with people who know (“have a relationship with”) various departments, functions, or features of the organization because they are members of the organization or somehow interact with it or with some of its components. Most of these informants will see only a portion of the organization, bringing to mind the parable of the blind men describing an elephant. This parable seems especially apropos for ethnographers because it reminds us of the importance of forming relationships with many informants so as to aggregate the partial information that each can supply. Yet the parable also misleads by treating the elephant as merely a body to be palpated rather than as a series of parts that, acting together, can move loads or transport passengers. Informing about each of the parts, the blind men have neglected what it is that the whole can do. In effect, organizational ethnographers need to ask not only what the parts look like but also what they can do, particularly when combined into the larger, acting whole. And for this, “being there” is crucial.

This article examines the practices of ethnographers carrying out research in and, especially, on organizations. These practices arise from some combination of the bureaucratic and practical constraints of conducting research and the researcher’s strategic decisions about information gathering (summarized in Table 1). Although scholars have commented on some aspects of these practices (complaining about access delays, for instance), organizational ethnographers’ research practices and the relationships they form in the field have not been examined with an eye toward implicit assumptions about the nature of the organizations being studied, the downstream effects on what can be learned, or the knowledge they ultimately produce. In the light of recent research bringing the tools of the sociology of knowledge to bear on social science research (Camic, Gross, and Lamont 2011), it seems imperative to think about relationality in the production of ethnographic knowledge.

OPEN IN VIEWER

Table 1.

Epistemological Implications of Ethnographic Tasks and Practices.

Practices/Tasks	Pressures and Blind Spots
Submitting project for ethics review (IRB)	— Ethics review requires researcher to specify who/what is being studied — Assumes bounded organization — Assumes continuity of research subject over time
Gaining access to organization and participants	— Access to “organization” is not access to individuals — Top-down approaches may close off access to or affect rapport with low-level participants — Side-in access may seem disrespectful or insubordinate to bosses
Observing/shadowing	— Organization or informants may select/exclude activities/areas for observation — Activities vary in how easily they can be observed; more collective activities tend to more observable; work with things and people more observable than work with data
Interviewing, formal and informal	— Informants can only talk about what they are aware of — Informants may give distorted accounts of events—unrepresentative, cherry-picked, defensive, or boastful
Participating in organizational life	— Organization or informants may invite ethnographers to participate in the organization’s work, or “give back” in monetary or nonmonetary ways — Explicit requests for help may help identify subtle shifts of loyalty
Gathering/inspecting documents	— Documents vary in whether and how they can be accessed — Documents cannot be taken at face value; they often give official portraits of organization or policies — Documents are generally “touched” by multiple people as they are created, used, and interpreted. No single person’s perspective will give full picture

Note: IRB = institutional review board.

In this article, I follow the outline of Table 1, beginning with a discussion of ethics review and its effects on research practices. I next turn to the challenges of gaining access to the organization as fieldsite. In a third section, I explore the difficulties of participant observation and shadowing in organizations. Fourth, I discuss the limitations of interviewing organizational informants. I examine requests and pressures to participate in organizational life in the fifth section. Finally, I describe the obstacles researchers face in analyzing organizational documents and records. Throughout, I interrogate the epistemological implications of the constraints placed on organizational ethnographers at the various stages of the research process and provide some examples of methodological workarounds that ethnographers may use to overcome them. To illustrate my points, I draw on examples from the ethnographic literature and particularly from my own multisite research on the use of rules in HIV clinics. ²

Undergoing Ethics Review: Getting Sidetracked with Mandated Contacts and (Mis)Conceptions of Research Subjects

According to university offices of research, research cannot begin until researchers have submitted their proposals for approval by the proper authorities. These days, the “proper authorities” typically include regulatory bodies that review research ethics (for many disciplines), approve proposals for submission for funding, or approve the use of university resources and facilities for research purposes. It is especially ethics reviews that have garnered attention, and scholars worry about the chilling effect of such oversight, suggesting that some important research is never conducted or conducted in degraded form (see, e.g., Dingwall 2008:9) because of the onerous process of securing permission from the institutional review board (IRB) (Bledsoe et al. 2007; Hamburger 2007; Heimer and Petty 2010). This section considers the particular challenges that organizational ethnographers encounter in obtaining IRB approval and describes how the ethics review process itself may have unintended epistemological consequences for research projects.

Although the effects seem quite pervasive, those writing about ethics oversight have especially worried about discouraging particular groups of researchers (early and late-career scholars), research on specific topics (e.g., “sensitive” or politically charged topics such as sexuality, illegal activities, or immigration), and studies of particular (“vulnerable”) groups or sites (prisons). To move these concerns into the realm of organizational ethnography, we should recall that Dalton’s (1959) research, which showed the inconsistencies between how managers claimed to behave and how they actually behaved, was conducted covertly and would surely be viewed skeptically by an IRB today. Was it necessary to conduct the research covertly? As a comparison point, consider that 36 corporations denied Jackall access for his study of occupational ethics, reported in Moral Mazes. Only after he rewrote his proposal in “bland, euphemistic language,” (1988:14) sidestepping the questions about ethics and values that he actually wanted to study, was Jackall able to gain access through personal contacts and school ties. Had he followed the norms of full disclosure, this valuable organizational research almost certainly could not have been done. Or consider Librett’s proposed research on work-related problems (psychological disorders, substance abuse, etc.) of undercover police officers (Librett and Perrone 2010). Librett, who was employed as a police officer, had unique access and was exceedingly sensitive to the importance of protecting the confidentiality of research subjects. With only one or two undercover police in each department, it would be child’s play for chiefs to figure out who had participated in the study, opening the possibility of disciplinary action, dismissal, or even criminal prosecution. Librett proposed carefully designed safeguards to keep information from flowing back to the police departments in which the officers were employed. Yet the IRB refused to approve the research. “We absolutely must have letters of cooperation from the agencies involved,” they insisted (Librett and Perrone 2010:736). Ultimately, Librett conducted a study that was a “significant retreat” from the original proposal (Librett and Perrone 2010:737). On the one hand, organizational leaders may be unwilling to support research that could raise uncomfortable questions; on the other, seeking the permission of organizational leaders has the potential to harm subordinates who are the subjects of a research project.

To state the obvious, the process of undergoing ethics review is intended to induce reflection on the ethics of research. In the United States, IRBs will deny permission when they believe a project does not meet the ethical standards articulated in the Belmont Report, codified in federal statutes (45 CFR 46, also known as the Common Rule), refined in policies and guidance issued by the Office for Human Research Protections (OHRP), and implemented by the IRBs of universities and research centers. Beyond the intended effect on research ethics, though, the obligation to submit projects for human subjects review and the process of going through review also shape ethnographic research in unanticipated ways. The templates of IRBs, originally created for the review of biomedical research projects, have been adjusted only marginally to fit the needs of social scientists. In preparing for ethics review, then, social scientists often must recraft their proposals to make them legible to reviewers. Standard IRB templates provide a poor match with ethnographic projects in research objectives, research methods, and research subjects. Many ethnographers will be uncertain about how to describe their “protocol” and will worry that their projects will make little sense to reviewers when redescribed in categories that fit so poorly. How should ethnographers represent research that was intended as theory development rather than theory testing, for instance, on a form that asks about the researcher’s hypotheses? Especially important for organizational ethnographers, how can a researcher answer questions about how many “subjects” will be “enrolled” when the plan is to observe a social group or organization with porous boundaries rather than to study individuals, and when no one is in fact being “enrolled”?

Because IRBs implicitly assume that those engaged in human subjects research are studying autonomous individuals, their templates do not easily accommodate studies of groups, networks, or organizations. To be sure, IRB templates have been modified over the years, but for the most part, these adjustments take account of variations in populations of individual subjects—for instance, adjusting consent procedures for research subjects with diminished capacity to consent because of immaturity, mental deficiencies, or illness. But studies of supraindividual social forms continue to present more formidable impediments because it remains unclear whose assent or consent should be solicited and how. What does it mean to get the consent of a group or an organization? And how should a researcher proceed if most, but not all, members of a group consent or if one person later withdraws consent (Atkinson 2009)? Nevertheless, because of the pressure for researchers to demonstrate that they have taken ethical questions seriously, ethnographers often make awkward, inappropriate, or indefensible analogies between individual consent and consent in research on groups, organizations, or social movements.

Moreover, because the process of seeking consent is typically assumed to be a one-off event, researchers are also led to think about their research subjects in a static way. Yes, IRBs occasionally acknowledge that consent is an ongoing process and make provisions for individuals to withdraw consent. More problematically, IRBs imagine an essential continuity over time, with the entity giving consent being essentially identical to entity ultimately being studied. That assumption is not always realistic for organizational ethnography. Within organizations, departments may be combined or split or, on a larger scale, organizations may splinter or merge. Even more common are the flows of individuals into and out of task forces, work groups, or ephemeral gatherings of staff, and the flows of individuals into and out of organizations as they are hired, receive promotions, leave for jobs in other organizations, or retire. These exceedingly common events are not ones anticipated by ethics review designed for closed, controlled research settings, such as clinical trials or laboratory experiments, where the flow of people can be anticipated and managed (Librett and Perrone 2010). After a period of high turnover, either in employees, leaders, or clients, is the organization really the same entity? And if the organization undergoes a change, does an ethnographer need to renegotiate access and consent? In the rapidly changing world of HIV care, where I conducted much of my recent research, such reconfigurations are quite common.

Especially importantly, fulfilling the requirements of the IRB requires forming a specific series of ties, not because these ties are especially appropriate for answering the researcher’s questions, but because they are necessary for initial or continuing IRB approval. The vision of an organization is hardly neutral—imagine suggesting to an IRB that permission to study the organization could be granted by a low-level employee rather than by the head of the organization (or someone designated to handle such matters). The implication of this is that researchers approach organizations at the top, with initial contacts organized around a series of IRB-approved research documents (summaries of the project, consent forms with mandated but distracting, tangential elements, contact information for principal investigator and IRB officials, etc.), which all too often create either anxiety or irritation, however carefully researchers present them. To be sure, these constraints on entry points and early contacts especially affect the unit of research and the process of data gathering rather than the unit of analysis. Despite the validity of this analytic distinction, our thinking about units of analysis is inevitably reshaped by where and with whom we start (indeed, there would be little point in research if thinking was not affected by information gleaned in the research process), and we therefore must remain attentive to the effects of entry points.

My own research on governance in HIV clinics illustrates how preparation for IRB review and the review process itself shape research projects. As I went through the IRB process—six distinct reviews, some with multiple layers, for five clinics in four countries—I was asked to identify both my research site(s) and the groups from which my research subjects would be drawn. For most biomedical projects, the answers would be a medical facility (hospital, clinic) and patients with a specified set of traits (e.g., experiencing or at risk of a particular illness, receiving or eligible to receive a particular therapy). In some senses, because I was studying clinics, my research subjects and my research sites were one and the same. Although studying clinics entailed observing and interviewing individual clinic workers, these staff members were mainly of interest as routes to information about clinics and clinic processes. They were informants, “units of research,” but not really my research subjects or “units of analysis.”

Equally problematically, I filled out forms saying I was studying HIV clinics, as if these were well-defined entities. In fact, the boundaries of clinics are exceedingly ill-defined and defined differently by different people. At the outset, I assumed that a clinic was a geographically located site where outpatient medical care, often of a specialized nature, was given. And although this is one common understanding, “clinic” need not refer to the location but can instead refer to the event—the regularly scheduled occasion when a medical caregiver provides services. That is, medical workers commonly talk about various clinics—the mental health clinic, the diabetes clinic, the HIV clinic—but they also speak of being available for clinic, of what day or time is women’s clinic, and so on, without this being simply information about the hours when a particular facility is open. In saying I wanted to study HIV clinics, then, was I saying I wanted to observe the times when a group of caregivers offered HIV-related services as in “we have HIV clinic on Tuesday” or did I mean that I wanted to do fieldwork in a location whose primary activity was supplying outpatient HIV-related care? The distinction is important because one of these (observing “HIV clinic”) entails more narrowly circumscribed observations of HIV-related caregiving, while the other (observing “the HIV clinic”) is likely to include both core HIV caregiving and a broader array of ancillary medical, social service, and administrative activities. The second is more in line with the usual agenda of an organizational ethnographer.

Even had I understood this distinction between clinic as activity and clinic as location, I would still not have realized that people and activities are not as fully or firmly lodged in the clinic (as location) as one might expect. Staff working in or for the clinic might be employees of a university or hospital or of the clinic itself. Or they might be employees of more than one of these entities. They might carry out their work in the clinic proper or in the hospital or in university offices. In some clinics, it was the clinic itself that tested prospective patients for HIV; in others, testing occurred in a separate site. In some HIV clinics, patients who did not yet need antiretroviral therapy (ART) were cared for by the clinic; in others, they were returned to neighborhood clinics to receive prophylaxis against opportunistic infections during the early phases of the disease. As patient numbers grew, some clinics began to “down refer” patients who had been stabilized on ART for follow-up in affiliated facilities; other HIV clinics continued to manage stable patients in their own facility. These complexities were not ones I could have anticipated before beginning my research, but the pressure to represent my sites in relatively simple ways did not help me think complexly about what exactly a clinic is or how I should go about studying it.

To summarize, then, beyond its pervasive effects on most social science research, ethics review affects ethnographic research in organizations in two additional, unanticipated ways. It affects how ethnographers conceive the entities they are studying. It fosters an emphasis on individuals rather than organizations or other supraindividual social forms, views of organizations as having well-defined boundaries, and representations of organizations as static rather than evolving. And it affects the formation of initial ties, and sometimes even the possibility of conducting the research, by mandating conversations with particular organizational representatives on specified topics organized around documents and forms required and approved by the IRB. Ethnographers learn by forming relationships in the field. Neither misconceiving the entity they are studying nor placing too much emphasis on forming ties with organizational leaders seems likely to promote and support the sort of productive fieldwork relationships on which organizational ethnographers depend.

Gaining Access: Official Permission as Bridge and Barrier

Once IRB hurdles have been cleared, researchers can begin in earnest the process of securing access to their fieldsites. Ethnographers have mainly worried about simply getting in, but some researchers have also asked how access points affect what relationships can be formed subsequently and, therefore, what they will ultimately be able to learn from their sites. This section discusses the benefits and challenges that flow from obtaining “official permission” to conduct research as well as how the organizational status of fieldsite gatekeepers affects the knowledge ethnographers are ultimately able to produce.

Feldman, Bell, and Berger’s (2003) thoughtful collection on gaining access amply demonstrates that there are many routes into organizations. That said, because the members of organizations have relations with one another, how and through whom one enters can be deeply consequential. Because of this, Chambliss (1996) recommends “side-in” access rather than top-down access; the latter, he sagely notes, cannot be assumed to be the same as real access. Although in some official sense, those at the top may be empowered to grant research access on behalf of the organization, anyone at all acquainted with organizations knows that people lower in the organization have considerable discretion over how fully they cooperate with researchers. A reluctant informant is surely better than no informant but not by much. Some organizational ethnographers thus approach those at the middle and lower rungs directly rather than through their superiors.

An official stamp of approval may give a researcher the right to knock on doors but can sometimes make it more difficult to get through those doors. When tension between groups is high, being perceived as affiliated with one group may make others reluctant to talk. In effect, ethnographers need to convey that having received an official blessing does not make them the allies of any particular social group. Information about cleavages in the organization—sometimes grounded in very local experience and organizational history, sometimes in the usual social divisions of race and ethnicity, class, gender, age, and occupation—may suggest ways to counteract the worst effects of arriving with the too obvious blessing of the powers that be. In the HIV clinics where I gathered data, my record of research in medical settings helped with the top brass, my knowledge of insurance (also from previous research) improved my rapport with accounting staff, and my difficult-to-pronounce childhood Congolese name closed some gaps with nurses as I garbled their Zulu names.

In fact, for several reasons, ethnographers are likely to have to use top-down and side-in approaches in tandem. First, side-in access also carries a downside risk. As Chambliss (1996) illustrates with a near disaster of his own, a side-in approach will occasionally seem underhanded to supervisors and bosses who discover that a researcher has initiated contact with their subordinates without first consulting them. Profuse apologies, careful explanations, and requests for help often clear things up, but remedial actions occasionally prove insufficient. My side-in attempt to enlist a Thai professor as a research collaborator failed when a department chair interpreted my discussions as going behind her back. In Uganda, I was scolded by an eminent HIV researcher (not in my fieldsite) for scheduling an interview with a subordinate without his permission. When he insisted that I come talk to him about my mistake, all was forgiven, and a wonderful interview ensued.

Although top-down access to an organization is never sufficient, it has clearly become increasingly necessary with the institutionalization of ethics reviews and the corresponding diffusion of concerns about liability (Bledsoe et al. 2007; Dingwall 2008; Hamburger 2007; Heimer and Petty 2010). Not only will bosses and supervisors object to a researcher coming in without their express permission but subordinates also will worry about getting into trouble for talking with an ethnographer or saying the wrong things. In this environment, “no” may seem like the most prudent answer to a request for research access, and an early refusal can easily become a cascade whether or not the first refuser has any special authority to speak for the organization.

A first refusal can sometimes be turned around, though, if a researcher learns enough about the structure of the organization to locate someone able and willing to override the no (in a hierarchical organization), to persuade coworkers to change their minds (in a flatter organization), or simply to provide an alternate access point (in an organization in which people operate more autonomously). Even when the doors to an organization seem closed, a researcher may still glean sufficient information to enable contextualization or comparison with whatever organization is ultimately more open to being studied. Early in my research on HIV clinics, I was denied access to a clinic I had hoped to use as a fieldsite. Accepting the refusal, I scheduled an interview with a key researcher in the clinic, not intending to gather information about the clinic but only to discuss his role in HIV research circles. Appalled that his clinic had denied me access, he took the initiative to get the decision reversed, in this instance on the grounds that one part of the operation (the caregiving component) could not speak for the entire facility. At this early stage, I could neither have known much about the relationship between the treatment and the research components of the clinic nor about who was empowered to make decisions about research access. The relative autonomy of professionals in medical and academic settings—and one physician’s passionate support for research—worked to my advantage in this instance. Perhaps more importantly, though, this early introduction to clinic cleavages sensitized me to the importance of building relationships that would lead me into both the world of treatment and the world of research.

To reiterate, then, the pressure (including from subordinates) to adopt a top-down approach to access can all too easily reinforce a sense that those at the top can speak for the organization as a whole, that leaders, more than their subordinates, are the organization. Bosses’ self-aggrandizement is not limited to formal organizations, of course, but is particularly pernicious when the “speech” of complex organizations—organization charts, blogs, policy statements, websites, and newsletters—reinforces this skewed perspective, despite occasional bows to a more egalitarian ideology. Organizational ethnographers must guard against the reification of this view that leaders are the organization, including reflecting on how early contacts shape subsequent encounters and, ultimately, their portraits of the organizations they study. Fortunately, not all pressures push in the same direction, as we see when we examine the day-to-day activities of organizational ethnographers.

Making Observations: Organizational Constraints on Hanging Out

In depicting ethnography as primarily “hanging out” in a setting, ethnographers often gloss over the assortment of factors that shape where they can spend time, what they can see, and who they can talk to. These factors, which include but go well beyond IRB-imposed constraints and the imperative to seek official permission, may have little to do with an ethnographer’s preferences about what he or she would like to see and much more to do with the structure of organizational life and the inherent properties of work and talk. Some organizational activities are just harder to observe than others, and some topics are simply more difficult than others to access through talk. This section explores the constraints that organizational ethnographers face when carrying out their fieldwork, including institutionalized spatial, temporal, and moral boundaries and impediments arising from the organization of work activities. It also considers workarounds ethnographers can use to gain access to off-limit spaces and to the frequently private, silent, and “interior” activities of organizational actors.

In many organizations, people arrive at prespecified times to begin their workday, spend long periods carrying out their duties, and depart with clear plans for their return. This predictability of organizational life is a huge boon to organizational ethnographers. There is little point in arriving before others or planning to leave long after they depart, but an ethnographer can safely assume that there will be someone to observe or talk with during normal working hours, however those are defined. Moreover, because people are to some degree jointly responsible for what happens in that location, organizational participants will often fill in for each other, at the very least offering a cup of coffee and a place to wait, but often some conversation as well.

This apparent advantage of conducting research in organizations should not be overstated though. As Cohen, March, and Olsen (1972) pointed out in their analysis of organizational decision-making, participation in organizational life is structured. Organizational members are not equally present and equally available, and this shapes what ethnographers can do and see every bit as much as it affects organizational decision-making. ³ The structure of participation also varies from one time to another and from one kind of organization to another, of course. If the work of an organization is increasingly dispersed because of telecommuting and virtual meetings, for example, the possibilities for researchers are dramatically altered. When staff work remotely, what researchers can access, what they can see, and the kinds of relationships they can form are altered.

What do we know, then, about who is available and who is not? First, social status matters. Frontline workers—receptionists, guards, secretaries, and program assistants—are often more available, at least superficially, to researchers than are people who work in relatively private spaces. But even those who are relatively exposed to the public may not be as accessible as we might expect. Counters offer workers a modicum of privacy and some capacity to ignore members of the public for at least a few minutes until a task is completed. Such counters also separate truly public space from semipublic office space, which members of the public, including ethnographers, cannot enter without permission.

Getting into more private spaces—private offices, workrooms, staff lunchrooms, or meeting rooms—typically requires telephoning or e-mailing in advance and may require going through several people. Initial requests may land on the wrong desk or simply go unanswered. Often ethnographers only gain the information needed to route requests properly by doing the research. In practice, then, the allegedly efficient organizational ethnography usually cannot be carried out so very efficiently. The benefits of doing research in a physical site where people are regularly and predictably present only begin to materialize somewhat later.

These facts about the structure of participation in organizations affect core aspects of data gathering. They shape the possibility space of the relationships fieldworkers form with organization members and therefore who they are able to talk with and observe; they shape what actually happens in organizations at particular times ⁴ and therefore what ethnographers are able to see; and they shape access to the documentary evidence that allows ethnographers to understand the organization as an entity because documents come to ethnographers primarily through contacts with the people who create, work with, preserve, discard, destroy, misplace, file, or conceal these artifacts of the workplace.

As researchers go through the early stages in their fieldsites, suggestions about who to talk with are inevitably offered. When ethnographers go through official channels, organizational representatives have an opportunity to structure at least the initial episodes of observation with suggestions that the researcher will “find it useful” to start with a particular person who knows the organization well, is especially helpful, or happens to have some free time. Organizations may be considerably more proactive, though. In the Ugandan HIV clinic I studied, the senior staff of the clinic designated an expatriate staff member for me to check in with as I made initial appointments. Was this “minder” intended to keep an eye on me until they were confident I was trustworthy? Was she to steer me away from sensitive topics or prevent contact with people who might reveal too much? I was handed an organization chart and urged me to “spend time with” people in all subparts of the clinic before settling in for more intensive observation and interviewing. It is conceivable that I did not notice that some doors were closed because so many doors had been opened. Whatever their intention, I did in fact hear about the clinic’s troubles, mistakes, and embarrassments. The sole restriction, discussed more fully below, was that I could not accompany caregivers into examining rooms. ⁵ But my thinking was also shaped by my minder’s conception of the organization and by the organogram, both of which emphasized people and activities fully within the organization rather than alerting me to the porousness of the boundaries between the clinic and the other organizations.

Although minders and supervisors try to steer researchers, some activities and topics are simply more amenable to observation and interviewing than others, as my HIV clinic research illustrates. When doctors or nurses interacted with patients in examining rooms, I was able to observe what they did, though of course I could not actually hear what caregivers heard through their stethoscopes or feel what they felt as they palpated patients’ bellies. Likewise, I could quite easily observe the parts of research that involved interactions with patients—discussing whether all doses of medication had been taken and taken on time, whether the patient had run a fever or had any pain, and so on. Tasks that involved less interaction—preparing e-mail queries about deviations from research protocols, for example—were usually harder to observe. However, group discussions, for instance of protocol committees’ decisions about acceptable adjustments of research protocols to take account of patients’ symptom patterns, could bring previously unobservable activities (the e-mailed exchanges) into my line of sight. Observation can also be difficult because work occurs in close quarters—looking into a microscope, performing a delicate medical procedure where only a single person can be close to the task. But when people need to coordinate around such tasks, for instance, in surgery, cameras and projection screens can make the work observable to others. Ethnographers can take advantage of other people’s need to coordinate but are not ordinarily able to arrange for such observational aids beyond what others have already put in place.

Although the particulars will vary from one setting to another, barriers to observation occur in all fieldsites. Activities that take place in relatively public settings (meeting rooms, collective workspaces, lunchrooms, and reception areas), because they entail interaction and conversation among workers, will be easier to observe than activities that take place behind closed doors (in single-person offices or examining rooms). And, as a general matter, tasks that involve working with objects or people are likely to be easier to observe than those that entail manipulating ideas or data. ⁶ Work with ideas can be notoriously difficult to observe because, like the newspaper reading that B. Anderson commented on, “It is performed in silent privacy, in the lair of the skull” ([1983] 2006:35). One can see pills being dispensed in a pharmacy and hear the pharmacist review the medical regimen with a patient. One can listen as researchers discuss a paper being drafted for a conference. More difficult, though, is figuring out how to learn about the work performed in the “lair of the skull”—the initial crafting of a conference abstract or the thought that goes into a nurse’s delicate request for a physician to correct or elaborate a note in the medical record. In medical settings, much of this “interior” work is ultimately exteriorized when it is rehashed in meetings or hallway conversations. But this advantage may be balanced by the increasing amount of time that people spend in front of computer screens—or, even worse, typing on handheld devices. As researchers learn, to observe a person working with a personal computer or handheld device (or paper and pencil, for that matter) often means either intruding rather uncomfortably into their personal space or continually peppering them with questions. Thus, although the computer interface may facilitate the work of the caregiver, it tends to make the work of the ethnographer more difficult. In principle, ethnographers can request copies of templates, of course, though that may require a worker to seek permission from the boss.

Moreover, some work with people may be hard to observe because it is placed off limits by individuals, organizations, or even by law. This is a common but noteworthy feature of organizational life, worthy of investigation in its own right. Although respect for the sanctity of the examining room is an entrenched part of clinic culture, contemporary regulations and regulatory practices reinforce the boundary. Clinic staff nervously remind patients—and researchers—about HIPAA (the Health Insurance Portability and Accountability Act) ⁷ and informed consent. How such rules applied to my research was never entirely clear. Although my project had been approved by the relevant ethics panels, staff members seemed uncertain about whether it was legitimate for me to observe what happened in the privacy of medical examining rooms. Ultimately, four of the five clinics decided that, with oral consent of the patients, I could accompany staff members into examining rooms, though I was asked to leave at sensitive moments (e.g., when male patients were asked to “drop trou”). ⁸

All too often, though, a request to observe or shadow a worker leads to embarrassment and bafflement about what exactly the researcher wants to see. Workers may be incredulous that anyone would want to watch them fill out paperwork—if it is boring for them to do, how could it be interesting to an observer? And if filling out forms requires attention—as it does—how is the worker supposed to behave toward the researcher? People find it exceedingly awkward to work in silence as someone watches. For this reason, work that involves some opportunity for banter is usually easier to observe than work that is mostly silent.

By this principle, meetings are easier to observe than solo work, both because of the easier sociability that occurs with multiple people in the room and because of the diminished awkwardness when responsibility for interacting with the fieldworker is diffused (a rare instance when the diffusion of responsibility is an advantage). And here is where current work practices create fresh difficulties for ethnographers. When meetings involve some combination of people physically in the room and people participating remotely, for instance, an ethnographer’s presence can once again become salient. Changes in the ratios of people present to people participating remotely increase the likelihood that an ethnographer will be a too-visible presence in the diminished group of those actually present in the room. The same point is likely to hold for other modifications that decrease physical presence in the primary work setting.

To some degree, the awkwardness of observation can be reduced by conflating observation with existing organizational practices. In medical settings, new workers often shadow experienced colleagues; students and visiting colleagues often accompany staff members into examining rooms and meetings. Comparing ethnographers’ practices to existing organizational practices normalizes a fieldworker’s presence, making it more comprehensible to those being observed. ⁹ And because, in medical settings at least, status is signaled by having a group of students, residents, or fellows following in their wake, senior staff may be more accustomed to and willing to be observed than ethnographers would expect, given the protection that usually comes with high status. Conflating requests to observe with the common practices of an organization can create other challenges, though, a point discussed more fully below.

Conducting Interviews: Limits on What Informants Can and Will Reveal

Often the most useful conversations with informants are those that arise naturally during lulls in activity when an ethnographer is hanging out, with all of the documents, instruments, and activities of the site available as props for the conversation. When a more extended discussion is needed, though, the ethnographer may need to ask that a period of time be set aside exclusively for talking and may even need to request a formal interview. When conducting interviews, organizational ethnographers must confront resistance to talking openly with researchers in a more formal setting, the limited perspective that individual informants bring to the table once they arrive, and cognitive impediments to sharing “unavailable” information. This section explores these inherent constraints and discusses strategies ethnographers can use to work around them.

To begin, how formidable is this hurdle of simply asking informants to take the time to sit down for a lengthy chat? For some people, the request is difficult to accommodate because individual appointments are not normal occurrences in their work lives; for others, an interview simply gets folded into the schedule alongside other appointments. Likewise, the meaningfulness of appointment times can vary widely with local practices, some grounded in the routines of organizations and professions, others in local culture. An interview appointment with a physician is subject to the uncertainties of timing common in doctors’ schedules; appointments with government functionaries may be delayed by official business.

Whether interviews are formal or informal, though, what matters most is the willingness of respondents to talk openly. Reluctant respondents often simply decline a request for an interview, though they may opt for a “slow refusal” instead. ¹⁰ More difficult, though, are those circumstances where respondents are unsure about the rules of their organization and anxious about stepping over some ill-defined line, a response especially likely in lower echelon workers in highly bureaucratized organizations with strong accountability pressures. When respondents hesitate to talk about even relatively innocuous topics, the researcher gets valuable (but frustrating) information about the relations of workers to the organization and to one another, which topics are especially sensitive, and how boundaries are policed.

Additional impediments impinge on interviewing, whether it is done formally or informally. Just as some activities are easier to observe than others, some matters are easier to learn about through discussion than others. Generally speaking, respondents are more able to talk about things they notice—things that have not been thoroughly routinized, for instance—than things that escape their conscious attention. What we do automatically without much conscious thought becomes less accessible to us. Although such things can eventually be retrieved, they are not the matters that people will spontaneously discuss. Thus, in my research, informants were more able to articulate thoughts about the rules and procedures they found annoying than those that had been so thoroughly routinized that they had faded into the background (Heimer 2008: 41, 44). On the other side of the spectrum, as Chambliss (1996:194) notes, what ethnographers are told is often “more exciting than the ordinary drudgery” they see. Respondents, like other people, are affected by the availability bias (Tversky and Kahneman 1974; Heimer 1988)—they remember what is vivid and exciting rather than what is dull, ordinary, and mundane. To be sure, even though they are not representative, the dramatic moments can be important in defining and transmitting key values. But the mundane, which is often of special interest to sociologists, can be hard to recover and hard to place in the proper context without observation to complement material uncovered in interviews.

It is not just the mundane that is hard to recover, though. Access to an organization does not automatically give an ethnographer access to organizational secrets or the dark corners of the backstage. Although both individuals and organizations have secrets and backstage areas, organizational secrets may be more elaborately and (sometimes) more formally classified by levels of confidentiality. Moreover, the resources of the organization can be brought to bear on the production of such classification systems, on enforcement of silence, and on the punishment of breaches. During my time in HIV clinics, I regularly heard staff members strategize about what information to share with outsiders and how to present it. I observed preparations for monitors and site visitors, including relocation of files and other artifacts, with problematic objects moved to the deeper recesses of the backstage, and more conforming artifacts (such as groomed files) placed where they would be readily available for inspection (Heimer 2012; Heimer and Gazley 2012).

To summarize, then, organizational ethnographers encounter three core types of barriers that make observation and interviewing difficult. The first of these, institutionalized spatial, temporal, and moral boundaries, are especially likely to govern access to particular spaces and people and to prohibit access when the presence of an outsider would seem morally (or legally) inappropriate. Institutionalized patterns of work create a second type of barrier. Organizationally or occupationally based patterns of work shape how participants understand ethnographers’ requests to observe. In some organizations, watching someone else work seems quite normal; in others, watching another person work can seem almost voyeuristic. Finally, cognitive impediments can make observation and discussion difficult either because work is “interior,” carried out mostly in silence and often with no visible trace (short of magnetic resonance imaging) of what is going on in the worker’s mind, or because it has become unavailable to workers themselves by dint of routinization.

At first blush, these factors that shape what ethnographers can learn through observation and interviewing appear to be individual-level phenomena—after all, the availability effect is a well-known cognitive bias. In fact, though, they tell us much about organizations and the challenges of studying them ethnographically. In the case of the availability effect, for example, it is the processes of routinization in organizations that make some activities thoroughly automatic and mundane—just a boring matter of filling in some blanks—while other activities require considerably more deliberation. Likewise, which matters are brought to meetings for discussion, which topics warrant hallway discussions with colleagues, and which actions require formal consultation with colleagues or superiors depend on how things are arranged in a particular organization. And which staff members are considered too important, too insignificant, or just the right candidates to spend time with the nosey ethnographer also tells us much about who matters in the organization.

In relating to an organization and its members, then, organizational ethnographers, like others in the organization, are necessarily always also interacting with the organization’s formal structure and policies as well as its less formal norms. For an organizational ethnographer, these impediments and mandates supply important information in their own right (Sallaz 2009), but they also suggest how ethnographers can work with rather than against the barriers. Careful acknowledgment of organizational norms may, for instance, decrease barriers associated with temporal, spatial, and moral boundaries. Although ethnographers’ practices may seem bizarre in some organizations, ethnographers can typically find some points of comparison to make their activities more legible. And cognitive barriers to observation can sometimes be surmounted by co-opting the routines used by organizational participants themselves. Workers exteriorize otherwise “interior” or “unavailable” processes in meetings, in training sessions, and in reports, all of which may be accessible to ethnographers and may provide an opening for further discussion.

Participating in Organizational Life: Giving Back and Going Native

Ethnographers inevitably must find some role for themselves that works for both them and those they are observing. Often in organizations, this entails conflating their requests to observe with other common practices of the site. Yet this may create fresh challenges. In the normal course of events, shadowing easily flows into participating for those who are medically trained, and even for ethnographers, who may receive requests for help when their growing knowledge of the local scene and mastery of specialized language occasionally mislead people about the nature and extent of their expertise. Under these circumstances, increased pressures to “give back” can create ties and loyalties that may shape what ethnographers are able to see in both advantageous and disadvantageous ways. In this section, I ask how requests to “give back” in organizational settings differ from similar pressures experienced by other ethnographers and assess the magnitude of the downstream epistemological effects of such requests.

All ethnographers will feel some pressure to be friendly and accommodating—that is, after all, part and parcel of the process of developing rapport. And, depending on the circumstance, ethnographers might be asked to give back materially, for instance, by making cash donations to needy individuals or local causes. ¹¹ But for organizational ethnographers, these pressures may be somewhat more likely to mark similarities rather than differences. In the clinics I studied, staff members would occasionally be invited to contribute funds to support clinic work. In Uganda or South Africa, staff members might pass the hat to raise funds for the families of particularly needy clinic patients; in the United States, staff members might be reminded that high participation rates made the clinic look good in a general fund-raising campaign. In both of these scenarios, my participation signaled general support for clinic activities, though it did not directly help any of the people I was studying. Rather than contributing to the group (or to individuals) I was studying, I contributed alongside other members of the group. The contribution marked me as similar to my research subjects rather than as having deeper pockets or other kinds of privilege.

Organizational ethnographers are often asked to give back to individuals as members of the organization, rather than as private persons, or to the organization itself. They may be asked to pitch in on some task, and this may alter what they are able to do and see. Bosk (1985), for instance, noted that he was episodically asked to modify his usual research practices, for example, to refrain from taking notes when he was needed as a legal witness. Ethnographers may also be asked to offer advice, assess a document, or make an introduction. In the clinics I studied, clinic staff occasionally solicited my views on organizational design, research strategy, research ethics, and the social worlds of their patients. As my time in the field drew to a close, some clinics also requested a presentation about my initial findings, premature though that seemed to me, but neither pressured nor advised me about what my presentation should contain.

Do these exchanges affect what can be seen—either by modifying the observer’s capacity to see or by altering the behavior of people in the fieldsite? And are these effects different for those studying organizations than for other ethnographers? Given the size and complexity of organizations, the effects of exchanges are likely to be limited. Even if relationships with a few research subjects are modified by these exchanges, many more will remain unaffected, particularly when relationships are conceived as “professional” or work related rather than as more fully rounded friendships. Yet the effects of these explicit requests may well be easier to track than the more subtle shifts of loyalty that occur over long periods as one becomes familiar with a site and comes to feel that one is part of the team. Elliott, for instance, suggests that as bioethicists become part of the medical bureaucracy, “it is reasonable to assume that the duties, allegiances, and professional identities of bioethicists will be shaped by the institutions in which they are employed” (2005:381). Although there is no reason to expect that ethnographers would be immune to such shifts of loyalty, in important ways they remain outside the organizations they study. They generally are employed elsewhere, split their time between the organizations they study and the places where they work, and usually exit their fieldsites completely at some point. And in the end, the fact that organizational ethnographers’ focus is on the organization more than on individuals may facilitate the distance that is essential for writing up the fieldwork in an objective or even critical way.

Engaging with Documents and Paperwork

Much of what goes on in organizations involves documents—organization charts, mission statements, brochures, annual reports, budgets, time sheets, expense forms, standard operating procedures and other written policies, grade sheets, requisitions, inventories, announcements, calendars, and so on. Documents and other organizational artifacts are important resources for ethnographers in managing the practical problems of studying social configurations larger than small groups (Brown-Saracino, Thurk, and Fine 2008). Organizational documents can give researchers crucial information about people’s responsibilities and relations to one another, the constraints under which people work, and the culture of the organization. Yet, as many ethnographers know, organizational documents are anything but transparent (Riles 2006; Hull 2012). Scholars need to look at as well as through documents, historian Ben Kafka suggests in reviewing the recent work on the history of paperwork (2009:341). In this section, I discuss how organizational ethnographers can critically analyze the records produced by the organizations they study, and thus avoid some of the common interpretative traps these documents pose.

Since Weber’s (1978, Chapter 11) foundational work on bureaucracy, documents have been understood as tools to exert control and to coordinate activity. Crucial as they are in getting work done, documents are equally important in bringing organizations and other entities into existence (e.g., in articles of incorporation) and helping to form their special character (e.g., in vision statements and orientation materials). Documents also play a role in distinguishing among entities (e.g., in systems of classification) and specifying appropriate relations among them (e.g., in employee handbooks and standard operating procedures).

For the purposes of organizational ethnographers, documents are especially important because of what they can tell us about supraindividual matters. They may, for instance, help ethnographers understand how organizations see themselves and represent themselves to insiders and to outside audiences, a point well illustrated by Strathern’s (2006) examination of university mission statements and Kunda’s analysis of the corpus of texts and recordings that made up the “articulated and relatively enduring inscriptions and codifications of the company point of view” (1992:53) of the engineering organization he studied. As Fayard and Van Maanan note, top managers experience a “pressing need to speak holistically about the company” (2015:19). Yet as Kunda shows, even when organizational texts are broadly consistent, organizational ideology is not adopted fully or uncritically by organization members. Moreover, as documents prepared by energy companies (Hall 2015; Mulvey and Sullivan 2015; Banerjee et al. 2015), tobacco companies (Glantz et al. 1996; Weishaar et al. 2012; Hurt and Robertson 1998), and automotive manufacturers (Hakim and Ewing 2016; Wikipedia contributors; Ivory and Tabuchi 2016) show, internal conversations can diverge quite sharply from statements for public consumption. Even when few people believe the ostensible message of a document, a “fantasy document” may nevertheless enable an organization to move on with its work (Clarke 1999). Texts cannot be taken at face value, then; organizational researchers need to consider not only how, by whom, and with what purpose organizational texts are produced but also who they are intended for, how they are disseminated or concealed, and how they are received and used. In short, organizations’ texts and artifacts need to be analyzed as cultural products (Griswold 1987).

Documents created for one use are often redeployed for other purposes, sometimes in other organizations. But these apparently efficient redeployments of existing materials also subtly transform the documents, as the history of medical records shows. Initially created as handwritten logs of a physician’s daily activities, ledgers with medical records were typically stored in the physician’s private office and used for teaching and research rather than for guiding clinical decisions. Eventually and with some resistance from doctors, medical records were reorganized as patient-centered case files and relocated to the ward where they could be used in case management (Timmermans and Berg 2003). These records now included not just physician inscriptions but notes on the observations and interventions of other caregivers and information on tests and medications. Beyond their use for tracking and guiding patient care, records gradually came to be used in governance and billing. As record keeping became standardized, courts concluded that medical records were objective representations (previously they had been categorized as inadmissible hearsay) and could be used in legal cases (Timmermans and Berg 2003:45), and this in turn led to further standardization.

Pressure to modify documentary practices continues unabated, as does resistance to that pressure. Administrators urge physicians to write fuller records to support claims that encounters are “complex” and should be billed accordingly (Heimer 2008:39); medical professionals debate over enlarging the group permitted to write in medical records and what restrictions should be put on those entries. ¹² Staff members worry about what patients and their families (and their lawyers) will glean from medical records and adapt their inscriptions accordingly. ¹³ Medical records have thus evolved from a physician’s personal record into institutional boundary objects (Bowker and Star 1999) used for tracking what has been done, deciding what to do next, communicating with other staff members, training students, supporting the submission of bills to insurers and other payers, and a host of other medical and administrative functions. As documents are redeployed and move from one department or organization to another, labels, tables of contents, tracking sheets, file folders, and binders get added to categorize documents, track their flow, guide transformations of data, and ensure that data and documents are used in permissible ways by authorized staff members. ¹⁴ Paperwork begets paperwork.

Perhaps more surprising is the seamless mixing of instrumental and symbolic uses of official documents. The very documents that play a role in coordination and control also support individual and collective status and identity claims. The right to participate in preparing important documents supports workers’ claims about their standing in the workgroup. For instance, it was a coup for nutritionists when they received permission to add their notes to the medical records in one of the neonatal intensive care units I studied (Heimer and Staffen 1998). Likewise, having a reputation for good documentation was important to the status of both the clinics I studied and the individuals who worked in them (Heimer and Gazley 2012). Ambivalent, as they were about research monitors, HIV clinic research nurses were nevertheless proud to be praised for meticulous record keeping. The regulatory affairs specialists of clinics proudly showed me their elaborate and meticulously maintained records and filing systems. Although some documents were lightly modified versions of widely used templates, others were thorough revisions of imported templates or completely original local creations. The Ugandan clinic’s standard operating procedures, for instance, covered all of the topics mandated by the U.S. National Institutes of Health but had been carefully adjusted for local conditions. And one South African nurse had a well-deserved reputation as the go-to person when a new form needed to be designed.

Finally, beyond looking at overt messages, ethnographers must attentive to the pattern of silences and lapses in documents (Espeland 1993:315). As “repositories of interests” (Espeland 1993:314), texts can preserve ideas over time, making it easy to resurrect a proposed dam project, for instance (Espeland 1993:315). Thus, ethnographers need to be attentive to the life cycles of documents, noting which documents are used and which have instead been left to languish in file drawers or on computers, perhaps creating an inventory of policy ideas for future use (Feldman 1989).

All this suggests that organizational records, those “most despised of all ethnographic subjects” (Latour 1990:54), lead a far more complex life than many ethnographers have imagined. Documents and paperwork are one of the means by which organizations “live.” In a deep sense, organizations do not simply have documents; rather, organizations are constituted through the interaction of documents and people. Getting to know an organization thus inevitably means getting to know its documents and forming a deep relationship with them. Documents and paperwork cannot be treated as disembodied artifacts or transparent collective representations, but instead need to be examined carefully to discern how people make, use, and live with texts (Smith 1984).

But where does that leave organizational ethnographers? Briefly, it leaves them with the obligation to ask a long series of questions that will enable them to read organizational documents and paperwork more critically. How did the documents come into the hands of the ethnographer? Was the ethnographer led to some documents but away from others? Is there any reason to think that some documents have been concealed? Document creation is often an iterative process. Is the ethnographer able to observe all parts of the creation and modification of documents? Very often documents are used by multiple people (that is, after all, how they coordinate the work of multiple people). Are some uses more observable than others? Organizational documents typically will be subject to more than one interpretation. Is the ethnographer likely to be privy to some interpretations but not others? Is an organization’s mission statement, for example, a revered text, a silly public relations document, or both? Does such a document supply information about organizational aspirations or about cleavages?

The complexity of these questions and the answers to them unfortunately means that documents cannot be assumed to tell us—at least in any straightforward way—about organizational life or organizations as supraindividual entities. Organizational documents cannot in fact do much to increase the efficiency of organizational ethnography. Once again, ethnographers must reflect on the relationships they are able to form and how those relationships affect their access to and understanding of organizational documents. Moreover, they should mine these artifacts, in large part because they may be able to reveal critical information about the relationships among individuals in the organization, and between individuals and the organization as a whole. Rather than providing quick and definitive answers, then, documents and paperwork should be thought of as providing hints about where to look and what to think about. Among other things, they should be understood as mediating the relationships among individuals and thereby helping to constitute the organization. In effect, documents and paperwork have to be treated more as enticing clues in the long treasure hunt of organizational ethnography rather than as the ultimate prize.

Conclusions: The Relational Work of Organizational Ethnography

This article argues that fieldwork relationships are the core technology of organizational ethnography and that the challenges of studying organizations differ from the challenges of doing ethnography in other settings or with other analytic purposes precisely because of the complexity of relating to organizations as well as to the individuals who are their members. Because ethnographers work through relationships, they must necessarily cede some control; the organization and its activities shape what aspects of the organization the researcher can and will study. For this reason, the lives of ethnographic research projects, like the lives of individuals and organizations, are likely to unfold in the manner described by Dorothy Smith (1987) and Mary Catherine Bateson (1989)—not carefully planned and executed but rather “composed” from the elements at hand, some purposefully created or gathered, to be sure, but others gleaned from the offerings of the local environment. This gap between plan and execution has been noted by others. Zussman, for instance, observed that grounded theory (Glaser and Strauss 1967) and the extended case method (Burawoy 1998), common touchstones for ethnographers, are “honored more in citations than in practice” (2004:356). What we have failed to see clearly, though, is that although our attempts to plan are sometimes thwarted by research bureaucracies and the like, they are at least as often thwarted by the necessity of coordinating with and adjusting to the schedules, needs, and plans of others and taking advantage of the serendipities of fieldwork. We have often seen this gap between plan and execution as a failure rather than as part and parcel of the process of doing fieldwork. A relational methodology necessarily introduces uncertainties and reduces researcher control.

In this article, I looked at how ethnographic work unfolds and, in particular, how it unfolds when conducted in organizations. I looked first at the early tasks—undergoing ethics review and gaining access—that set the stage for a project and then turned to the realities of data collection in organizations. For each of these stages and research activities, I noted the successive constraints and challenges that are introduced as projects unfold and the points at which projects can go off the rails as ethnographers attempt to form relationships with organizations and the people in them. In my own project, for instance, my early (mis)understanding of what a clinic is and who, therefore, can grant research access nearly cost me access to a very valuable site when I formed ties more selectively than was appropriate, given the bifurcated research and treatment structure of the organization.

The impediments I discussed in this article arise in three quite different ways. First, regulatory regimes—some internal, others external—may erect barriers to prevent unauthorized research or to place some people, activities, topics, or documents off limits. Second, social arrangements and physical barriers may be put in place to protect the integrity of work, work time, workspace and work materials, and to prevent others from intruding on organizational routines. Finally, inherent properties of social life can make some activities difficult to observe and some topics difficult to discuss, whether one is a researcher, a fellow worker, or a supervisor. Some of these barriers have been put in place with the express purpose of excluding research that has not been appropriately reviewed (by an ethics panel or organizational representative), but others are simply “naturally occurring” features of the organizations and activities in organizations, not specifically intended to make fieldwork or other kinds of research difficult. Overcoming them may nevertheless impose extra costs on both the ethnographer and the research subjects and so may decrease access. Crucially, though, all of these impediments work through their effects on researchers’ capacity to form relationships with organizations and the people in them.

But what about the specific challenges of learning about organizations? Zussman makes the case for multileveled research that “requires not only multiple sources of data but different orientations to that data” (2004:359). As he explains, “Places are not simply aggregations of the people who live and work in them. Neither are people simply products of the places where they live and work. Successful research on people in places requires data that speak to both” (Zussman 2004:359). With minor changes in wording, Zussman’s commentary fits organizational ethnography remarkably well. Organizations are, after all, one kind of place where people live and work.

The centrality of the relationships formed in the field cannot be overstated. Noting that most of urban ethnography had been done by men who had no access to the worlds of women, Lofland (1975) argued that male researchers missed important mechanisms for fostering and maintaining community ties. It was women after all, not men, who peeked out from behind the curtains to track activities on the streets where their children played. If ethnographers and their bodies are indeed the key instruments of ethnographic knowing (see, e.g., Van Maanen 1996:380, 2011:219), then it is especially through the relationships they form with people in that social world that they come to understand it. To be sure, ethnographers’ own direct experiences of the social world are crucial. As Van Maanen puts it, “by subjecting the self…to a set of contingencies that play on others…one can more or less see, hear, feel, and come to understand the kinds of responses that others display (and withhold) in particular situations” (2011:219). But important as those direct experiences are, they are mediated by relationships with others in the field—the people who give access, teach skills, share advice, and reflect together on the experience.

The distinctiveness of organizational ethnography lies, then, not so much in a requirement for a unique set of skills as in the more intensive use of certain skills that are already in the repertoire of ethnographers. Difficult as it may be to form relationships with relative strangers, this is something ethnographers more or less know how to do. Much ethnographic lore focuses on how to create rapport, how to work with differences that might create too much distance or similarities that might afford too little, how much or how little to participate, and how to assess the effects of those choices. But we believe we know less about how to create ethnographic relationships with organizations qua organizations. In our everyday lives, we do of course have relationships with organizations, but those relationships are generally constructed with specific objectives and so lead to knowledge about small segments of an organization rather than the organization as a whole. For this reason, our everyday relationships with organizations may not provide the kind of guidance about how to know an organization that everyday relationships with people do for how to know individual-level phenomena.

Knowing an organization depends primarily, I suggest, on a more intensive construction of relationships that together permit a thoroughly layered understanding—each set of primary observations supplemented and interwoven with secondary observations that allow the ethnographer to see the relationships among people, the structure of positions, the historical evolution of work, the construction and maintenance of crucial ties across organizational boundaries, the evolving and varied understandings of the organization’s policies, practices, and culture. Yes, an organization is more than the sum of its parts, but ethnographers nevertheless access the parts, the something more, and the constructive process through relationships with the people who inhabit the organization, remaining continually sensitive to how those relationships are mediated by organizational structures, formal policies, norms, and culture. Once they have secured the raw materials for their analyses, ethnographers must indeed do the sifting, sorting, juxtaposing, abstracting, and reordering—what Dauber (1995) describes as “bureaucratizing the ethnographer’s magic.” It is this that allows ethnographers to learn things that people in the organization truly do not know about themselves and their organization and may even be chagrined to learn. And ethnographic accounts would be pale and wan without the ethnographic details gleaned from rubbing shoulders in the field. But fundamentally, the whole enterprise depends on constructing, exploiting, and maintaining relationships. Because “it’s turtles [relationships] all the way down” (Geertz 1973:28-29), organizational ethnographers need to be clear eyed about the subtle and not-so-subtle pressures that shape their access to organizations and the relationships they are able to form.