4

THE RHYTHMS OF INTERPRETIVE
RESEARCH I

Getting Going

The observation of which I shall speak is, for lack of a better term, interactive observation. It is not like looking through a one-way glass at someone on the other side. You watch, you accompany, and you talk with the people you are studying.

Much of what you see, therefore, is dictated by what they do, and say. If something is important to them, it becomes important to you. Their view of the world is as important as your view of that world. You impose some research questions on them; they impose some research questions on you.

That interaction has its costs—most notably in a considerable loss of control over
the research process. It also has benefits.
It brings you especially close to your data.
You watch it being generated and you collect it at the source.
It is not received data.

—Richard Fenno (1986: 4; paragraph breaks added)

With some sense of a research question in hand, along with the debates that inform it, a researcher is prepared to proceed to the more formal design elements of the research project: potential sources of data, and methods for generating those data. Note that opening: with some sense of a research question in hand. The extent to which this will be formalized depends on research communities, departments, specific advisors, funding agents, and one's own proclivities (including tolerance for ambiguity). We are in agreement with John Van Maanen (2011: 222) that “we should also allow our questions to determine our theories,” meaning that “one need not stake out a theoretical claim on how the world is before beginning a research project.” But at the same time, we recognize that this degree of openness and flexibility is likely to feel more comfortable among those who are old hands at interpretive (and in the case of Van Maanen's topic, qualitative ethnographic) research, whereas those more accustomed to the relative closure of experimental and quasi-experimental research designs are far more likely to desire more formalized research questions at the outset.

Due to the researcher's ongoing and evolving learning while in the field, as well as his or her limited control over settings and the persons in them, or over materials in an archive, interpretive research is, and has to be, much more flexible than other forms of research. This flexibility is a conscious, intentional strategy, and it applies not only to the need to respond in the moment to things said or done, but also to how the research process may be changing initial research designs and questions. Some research questions can lead the researcher to discover the most unexpected “answers,” which in turn can lead to revised research questions, also potentially unanticipated, which could not have been posed without having stumbled on the unexpected answer to the initial question. In abductive fashion, puzzles grow on the backs of other puzzles.

A design appropriate for this sort of research needs to reflect and make space for its iterative, recursive, and adaptive character. The design process itself is not necessarily as linear as the chapter's opening sentence suggests. Often, in the course of working out the design, the research question comes into focus in new or different ways; and that focus can send the researcher off in search of other literatures, other settings or archives, or both. The design process is, in other words, itself quite “circular”—the hermeneutic circle once again. We might even say iterative and recursive, each of its parts informing and folding back on the others, enacting the same sense-making spiral that characterizes the conduct of interpretive inquiry.

Perhaps because the process, the rhythm, of interpretive research manifest in these aspects is so contrary to the emphasis on control and other specific characteristics of positivist research, this iterative and recursive strategy and its significance for research flexibility are not widely recognized or well understood across the social sciences, something noted also by Becker (2009) in his assessment of recent US National Science Foundation recommendations for the conduct of qualitative research. Moreover, its requisite open-endedness and flexibility often make interpretive research designs appear—from the perspective of the much more closed-ended and controlled positivist design—underdeveloped. This, too, reflects a lack of familiarity with the different rhythms of these two research modes.¹

Much of the creative, intellectual work in positivist research projects (e.g., quantitative analysis, survey research, experiments) is “front-loaded” in particular ways: the choice of variables and how to operationalize them, the phrasing of survey questions and their ordering, the setup of experimental “manipulations”—all of these are established a priori in the research design or in the first stage of research, before the researcher engages with data sources.² Only when this prior “stage-setting” is completed does the researcher turn to the data themselves: measures are taken, survey questionnaires are administered, experiments are conducted. And only after that stage is completed does formal analysis begin: many of these approaches make a relatively clean distinction between data collection and data analysis. There may be some monitoring of results while the data collection phase of research is in progress (notably in double-blind health experiments, as required by human subjects protections monitoring); but most often, analysis begins when the data collecting phase has been completed, to avoid what in this view would be considered an error of ex post facto reasoning (which undermines the logic of tests of statistical significance).³

The rhythms of interpretive research are quite different. Its design revolves around front-loading of a different sort: the development of the researcher's a priori knowledge, so to speak, in the kind of clarification of expectations explored in Chapter 2, including, perhaps, preparations of body and emotions as well.⁴ These initial understandings—the researcher's provisional sense-making—will be “tested” in the field or archives, not literally, as in the case of significance tests in statistics, but by bringing them together with field realities, even to the point where those realities might take analytic primacy. Initial understandings are likely to be reformulated in light of new insights, new understandings, new knowledge acquired, and those reformulations will be subjected again to further inquiry, in that iterative, spiral-circular recursiveness of abductive reasoning described in Chapter 2. It is this continuous juxtaposition of conceptual formulations with field realities and the requisite flexibility that accompanies it that comprise the foundational rhythm of interpretive research.

In this dance of inquiry, data generation and analysis are ongoing and intertwined. Initial sense-making (analysis) commonly begins in the field or even beforehand, in the process of designing a research project, with the recognition that it will continue after fieldwork is completed, during the deskwork phase of research (formal analysis, working with fieldnotes and other materials, including the theoretical literature). And this iterative sense-making process continues on into the textwork phase, too (more argument-construction and word-smithing than analytic note-making and roughing out a draft), in which text-making takes on its “worldmaking” character (Goodman 1978). This means that writing is more than a “mere” description (“writing up”) of research “findings” (see also Clifford and Marcus 1986, Geertz 1983, Van Maanen 1988). These iterative–recursive processes take place within particular understandings of causality, hypothesizing, and analysis (taken up in Chapter 3)—which means, in turn, a different shape and sense to the research design.

In what follows, we engage the sorts of issues that interpretive researchers commonly consider in the first phases of research design: the selection of settings, actors, events, and/or texts, or archives, in which or among whom they are most likely to be able to address their research question or puzzle. This requires examining two features of the envisioned study: how the researcher expects and will arrange to gain access to the setting, actors, events, and/or texts; and in what role the researcher will do so. The question of access to archives takes on a different sense, and we touch on it in a separate section. We note those design elements and concepts that are shared by positivist and interpretive approaches, and we also mark the differences in ordering or “flow” between these approaches: the more step-wise process of positivist approaches compared to the distinctive rhythm of an iterative, recursive interpretive research process. In the chapter's final section we return to the matter of requisite flexibility in a research design, which we now elaborate on in terms of the agency of research participants: rather than a matter of lack of thought or planning, or even simple convenience, flexibility is essential to intelligent maneuvering in the field so as to pursue the situated, contextualized meaning-making of those whose lives, interactions, situations, written records and visual images, and so on are being studied.

In order to make the issues raised by these features as evident as possible, we draw on examples that provide the sharpest contrast to interpretive research practices, such as experiments and surveys, both of which entail a priori specification of cases based on an assumption of case homogeneity. We do so in recognition of the ways in which the positivist/interpretive dichotomy can, as noted previously, be an oversimplification when it comes to demarcating forms of research. Positivist ethnographic or case study research, the latter typically entailing heterogeneous cases, may share some research practices with interpretive researchers using related methods (but informed by different methodological presuppositions).⁵ For lack of space, we do not fully engage these similarities and differences.

Access: Choices of Settings, Actors, Events, Archives,
and Materials

In many interpretive research projects, the research question is intimately connected with a particular setting, a particular time period, a particular set of actors, and so forth—or with some combination of these. This would be the case with a research question developed out of a setting the researcher became involved with initially without any intention of doing research there, as discussed in Chapter 2. In that case, the necessary choices with respect to further particulars of setting, events, actors, and so forth develop reiteratively with the elaboration of the research question. The same holds for archival research, as inquiry follows what is found in the materials. In other cases, the researcher may have an idea about a topic and then go in search of appropriate settings and so forth in which to explore it: where the research will be carried out, with or among what actors or in which archives, investigating which events, and so on. Merlijn van Hulst (2008a), for instance, knew that he wanted to study an aspect of local governance, and he went in search of municipalities in which he could develop and then delve into his research question. Timothy Pachirat (2011) wanted to explore what he saw as the routinization of violence in the US, and thinking along those lines led him to consider carrying out his field research in a slaughterhouse.

All researchers focus on identifying settings, actors, archives, events, eras, and/ or texts that will enhance the likelihood of being able to explore the questions that interest them. This can include clock and calendar features that might affect research-relevant evidence. In all approaches, the rationale for choices made is a key part of the research design: the logic that links research question, setting(s), actors, documents, etc., and data-generating and -analyzing methods needs to be elaborated. For example, Kevin Walby (2010: 643) provides such a rationale when he writes that he chose to conduct his interviews against the background noise of coffee shops or outdoors at bus stops and park benches, settings that afforded anonymity and confidentiality. And Fujii (2009: 37) writes about her interview-based study that she “avoided market days because people were usually not at home” then. We underscore a central trait of interpretive research designs and their implementation already mentioned, to which we will return: these design specifications are initial formulations of the research question and choices of settings and so forth and of methods. Any of these may change once the researcher gets to the field and begins the encounters that comprise the research—something completely normal and legitimate in interpretive research (as discussed in Chapter 2 and the last section of this chapter).

But identifying settings and actors is only the first step. It does not automatically grant the researcher “access,” a permission-acquisition process—something researchers need to obtain at times from governmental and/or organizational entities at various levels, but also from individuals in a less formal, more interpersonal fashion. Researchers seek access to people in various kinds of settings: their homes (Soss 2000), a neighborhood corner, bar, or coffee shop (Liebow 1967, Zirakzadeh 2009, Walsh 2004), a community center organization (Yanow 1996) or gym (Wacquant 2004), a police department (Van Maanen 1978), government office (Rhodes et al. 2007), hospital (Barley 1986, Iedema et al. 2011, Katz and Shotter 1996), or factory, business, or corporation (Barley 1983, Dalton 1959, Kunda 1992, Roy 1959, Shehata 2009). The question for researchers is how to arrange this and what is entailed in doing so.

Initially, as the concept of “access” emphasizes, researcher bodies need to move through gates of various sorts. The term has customarily referred to securing formal permissions, a holdover from its methodological origins in archaeological and social-anthropological field research, especially in colonial settings. There, it was necessary to secure a traveler's visa to enter the field site to reside for an extended period of time, commonly requiring an administrator's permission. For interpretive researchers as well as for many working in a qualitative vein, however, “access” is not simply a matter of knocking on a door, literally or figuratively, in order to get in. As social anthropologists learned, this did not automatically gain them entrée into a village's social network. For that, the approval of the local “chief” was also needed, and this entailed more than just a piece of paper. Participant observer sociologists and others doing interview-based studies also came to understand “access” in a less literal sense, linked to the more interpersonal notions of establishing rapport with their interlocutors.⁶

Today, access is increasingly being understood in the context of the relational character of engagements with research participants in the field. In experimental, survey, or focus group research, the character of relationships is largely defined in terms of a “professional-expert researcher” who explains the conditions for participation to potential “research subjects.” The interaction is likely time-bound; and the implicit expectation is that the researcher remains in the proper, distant role of professional-expert throughout. The parameters of any relationality between researcher and researched are clear, even if not governed by explicitly stated codes of ethics. Moreover, the relationship has the texture of a contract, something made more formal in consent procedures developed in many human research participant protections policies (such as ethics review committees), whose signed documents, formal language, and potential oversight of researcher conduct mimic legal contracts and procedures.

By contrast, field relationships are less time-bound and more complex. Recurring interactions between researcher and researched mean the possibility of developing relationships that go beyond those in the model of researcher-to-“research subject” relationship characteristic of the contractual model of consent (see the discussion, below, of participants’ agency). This is a feature especially of projects in which the researcher dwells among the participants—participant observer or ethnographic research, in particular, but holding as well for interview research that involves repeated encounters with the same participants. In archival research, important relationships can develop between researchers and archivists (Bellhouse 2011). Relationship complications can arise when the researcher is acting in a consultant role or has the possibility of turning the research contract into a consulting one at its end (e.g., in school, hospital, or other organizational settings). The relational character of these interactions requires researchers to attend to the humanity of those who give of their time, and perhaps other resources, in helping the researcher gain greater understanding of her research topic—to treat them in their full human-ness and not just as means to an end (e.g., “my informants,” “my data”).⁷

Focusing on the humanity of research participants is part and parcel of treating them as agents in and of their own settings. More than that: just as “normal” relationships require ongoing maintenance, so do research relationships. This means understanding access as an ongoing process (Feldman et al. 2003), not a one-time-only unlatching of a door. It requires researchers to think about how “research friendships” are, or might be, different from non-research friendships. Walby (2010: 652), for instance, writes about trying to balance a degree of friendliness with self-presentation as a research sociologist “conducting a rigorous study.” It also means that researchers need to think about and plan for their leave-taking.

If this is a research friendship, does the researcher say goodbye forever when exiting the field? Or does he maintain contact and, if so, what kind? As Librett and Perrone (2010: 745) note, quoting Milne, “‘[C]aring interactions are established and maintained over time rather than a contract that once signed is forgotten’” (see also Henderson 2009: 292). Such engagements also bear on some aspects of research ethics, as well as the emotional side of field research (taken up in Chapter 7) and choices about what Burkhart (1996: 40) calls “the privileged grounds of ethnographic knowledge and appropriate ways to write about information gained through friendship.” The lines between researcher and friend (and perhaps even family member; Jacobs 1996: 296) may at times blur, and fieldwork relationships may also include elements of secrecy (Leap 1996). What, for instance, might a researcher have done to prompt a participant to exclaim, with whatever mixture of confusion, sadness, and even anger, “But I thought we were friends?” (Beech et al. 2009; see also Ellis 1995). D. Mosse's (2005, 2006) post-research troubles—in which colleagues with whom he had had professional relationships prior to the publication of his field research accused him of betrayal—illustrate this point.

Both institutional access-related negotiations and interpersonal ones can be or become political contexts that enable entrée for some bodies but not for others. Prior knowledge, including language skills and personal contacts (perhaps developed through previous non-academic field experiences), may make some projects doable for only some researchers, enabling them to gain access where others cannot.⁸ And formal permissions do not guarantee interpersonal access: the classic complete “stranger” of anthropological lore may not be welcomed, perceived with indifference, if not suspicion (and left without local family or friends to fall back on for emotional support; see Ortbals and Rincker 2009). These and other dimensions of researcher–researched relationships put the researcher in the position of supplicant in which he may be less powerful than those in the position of granting or denying access: the knowledge, institutional, economic, social, and/or cultural bases of the laboratory researcher's customary sources of power may not be operative in these circumstances.

Power and Research Relationships

At the design stage, researchers need to anticipate and consider the potential forms research relationships might take, as these may vary not only with the research question but also with respect to the relative power of the different individuals and groups studied. Within a research setting, power is relational, and the power of the researcher, the researched individuals, and the types of research relationships they may develop can vary considerably. In a single site, such as a school, hospital, or other form of bureaucratic organization, for instance, agency executives have different power resources than front-line workers and clients. Middle managers can enact different sorts of power relative to their employees than they do vis-à-vis their own bosses. In addition, researchers may have the initial power to set agendas, but participants may refuse to proceed with those agendas, reshaping them to their own purposes and meaning-making (see, e.g., Walby 2010: 654). How these situational variations will affect interactions and negotiations with the researcher cannot always be predicted, but need to be considered.

Interpretive researchers often study individuals with considerable social power—bankers (Abolafia 2010), lawyers (Pierce 1995), national security analysts (Cohn 2006), and other members of elite sociopolitical strata and areas of expertise. Although in such cases, those studied may expect the researcher to fulfill a “professional-expert” role—one who has mastered the substantive issues (e.g., regulations, executive orders, acronyms, and other situational jargon), with the status and power attendant on such expertise—experts and elites have considerable status and power themselves, and it is not clear that researchers will always have the upper hand in such situations. When scientists study other scientists, there is a complex interplay of power and knowledge dimensions. Drori and Landau (2011: 25) noted with respect to the nuclear scientists they studied that their own “role as researchers at Gamma resembled that of an equal participant in a conversation in which, regardless of the subject under discussion, the researcher and informant found themselves equally at ease.” But several of the authors in Bogner et al. (2009) remark on less equal dynamics in the political dimensions of interviewing experts. Indeed, Xymena Kurowska (personal communication, 4 April 2011) remarks on the “blurring between private and professional life” characteristic of the interactions among researchers, think-tank analysts, and practitioners within the landscape of European security research, who depend upon one another “for the reproduction of the setting,” at the same time as they occasionally envy one another's lifestyle.

Interpretive researchers also study individuals and groups with less ascribed social power, such as cocktail waitresses (Bayard de Volo 2003), secretaries (Kanter 1977), gang members (Venkatesh 2008), labor union members (Warren 2005), Native American women activists (Prindeville 2004), political campaign volunteers (Super 2010), neighborhood gangs (Whyte 1955/1947, Liebow 1967), homeless women (Liebow 1993), or professionals operating at the limits of the law (Cramer 2008). In such cases, the researcher may have considerably more social power than those with whom she interacts. But power, and powerlessness (Kanter 1977), are also situational, something that research settings may also reflect and determine. In one interview project, some “very old and frail people . . . were seen to have exercised considerable power over the course of the research and to have participated very much on their own terms” (Russell et al. 2002: 15, citing an earlier project of Russell's). As Fujii notes:

When American or European researchers, who are often white, educated, and highly privileged by local standards, enter the [non-American, non-European] world of the researched, they still enjoy structural forms of power. People with high social status vis-à-vis potential study participants cannot entirely control the world they enter, but they can certainly shape it to their liking in various ways.

(Personal communication, 29 August 2009, original emphasis)

These ways include getting instant appointments because of their ascribed status, having computers, cars, and other technologies at their disposal, and the option of paying participants (see Note 7, this chapter).

Interpretive research designs, then, need to anticipate the ways in which these potential relationships might affect the generation of evidence (and one's decisions to write about these encounters). Considering whether research relationships are likely to be neutral, friendly, professional, or possibly even hostile and combative can also be part of preparing oneself emotionally for the research project. As Drori and Landau (2011: 25) observe concerning their interactions with the nuclear scientists:

[W]e were attributed multiple roles, with distinctive characteristics, such as advocates, consultants, and researchers. The scientists related easily to our role as researchers, even offering their own suggestions regarding methodology and other questions of research. Our role as advocates, the most appealing to the scientists, served as an effective trigger for full-fledged research cooperation. It also provided them an outlet for expressing their own ideas regarding various organizational issues.

They quote one of the scientists who shared with them, extensively, his vision of the desired organizational structure: “‘Maybe many of us are frustrated social scientists, you ask me a question and I'll give you an analysis, you want information and I'll give you an interpretation.’” And they remark on the continual awareness of the responsibility demanded by this multiplicity of roles, requiring them constantly to monitor the data they were generating, clearly distinguishing between data for research and data for consulting purposes.

All of this means a requisite attention also to the “presentation of self” (Goffman 1959) in the field. Participants of all status and power levels will size up a researcher quickly, making initial assessments interpreting bodily characteristics (e.g., sex, stature, age, etc.), dress, tone of voice, and manner: all those elements that comprise nonverbal communication. Walby (2010: 651) writes about “policing [his] own gestures and speech acts to hinder respondents’ sexualization of [him],” and Lin (2000: 187–8) describes her concern not to convey nervousness to prison staff or inmates: “Both the staff and prisoners usually picked that up [the lack of nervousness] and responded with relief: a visitor who treads each step fearfully reminds everyone, uncomfortably, that the prison is abnormal and that their lives are on display.”⁹

Ascribed identity is not directly under the researcher's control: “. . . respondents can read different identities onto us before and during interviews, subverting our attempts to position [ourselves] as researcher[s]” (Walby 2010: 652), a point that holds for other forms of field research. And yet, identities, along with understandings of the research matter at hand, are both produced and contested (as Walby 2010 extensively notes), and researchers can sometimes use the identities participants ascribe to them to their advantage. Cohn (2006), for instance, turned some of her interlocutor's condescension toward her—a more senior, experienced, male professional facing a younger, less professionally experienced woman—in a more productive direction for her interviewing purposes, making herself his student on security issues. Fujii (2010: 233) recounts how the rumor that she was a local woman's long lost daughter led people to identify her as part Rwandan, “not as ‘Black’, ‘Asian’, or ‘Hispanic’ as I would be typed in the USA.” That understanding “provided a useful entry point for talking about ethnicity”—key to her research on the genocide in Rwanda. As these examples illustrate, a researcher's “presentation of self” is neither simple nor static, but an ongoing process that rests on self- and other-awareness, learning, and adaptation to the field. Others’ constructions of the researcher's identity may also shift over time, as the researcher becomes better known in the field setting.

Interpretive designs cannot be expected to predict and detail all of these intricacies, but those that articulate an awareness of the potential issues (and associated literatures) demonstrate the researcher's readiness to head into the field and negotiate the varieties of relationships that may develop therein. Because they enter others’ worlds seeking to understand from the perspective of those studied (rather than as journalists seeking facts, for instance), researchers need to be aware of the ethics entailed in field relationships (see, e.g., Blee 2002, Burgess 1989, Huggins 2002, Humphreys and others 1976, A. Kelly 1989, and Riddell 1989).¹⁰

Researcher Roles: Six Degrees of Participation

To speak of research in relational terms has implications for the ways in which researchers think about their roles in field settings. In the case of participant observation, the researcher needs to decide the extent to which observation will be coupled with some degree of participation and, if so, the degree and kind of participation in the activities and events taking place in a given site. In short, participation comes in different forms.

The degree and kind of participation may vary, ranging from a role in which the researcher participates as researcher alone, to one in which the researcher is present as both researcher and situational participant, but subordinates the researcher role to the situational one. Participation of the latter sort positions the researcher in dual roles: at the same time that he is participating in his situation-specific role, he is also always observing as researcher (Gans 1976). This includes observing himself as well as others. In the researcher-only role, when responding to something said or done, it is in the capacity as researcher. In the dual role, responses come out of the situation-specific role, rather than the researcher role. In between these two poles lies a repertoire of role combinations. Some studies may combine several different points along this continuum, linked together like the six degrees of relationship that reportedly separate any two individuals. The researcher studying community development, for instance, may be a “local resident” (de novo) participating in community meetings in one set of activities, whereas he is a researcher when interviewing city planning officials in a different set of activities. Autoethnography, which turns the researcher's membership in a class of actors into the object of study, is at the far end of that continuum, coming close to collapsing the distinction between the two roles.

A research design that includes a situation-specific role needs to assess what is feasible in that situation. Depending on the setting, aspects of the role may need to be negotiated with its members, including possible legal and/or ethical dimensions. Situation-specific roles often draw on researchers’ a priori knowledge and experiences, including prior membership, abilities, and inclinations. In turning a present or former member situation into a research project, a researcher would want to consider the impact this might have on fellow members, including colleagues and friends. Various factors may limit participatory roles, from physical ability and ablebodiedness (see Chapter 7) to culturally defined gender roles, from a lack of expertise in a project's central activities, where that cannot be learned on the job (e.g., Cramer's, 2008, study of midwifery), to not wishing to risk improper or illegal activity (such as drug-dealing; Venkatesh 2008). The legal and/or ethical implications of the latter need to be anticipated and considered, to the extent possible. Stuffing envelopes as a staff worker in a non-profit organization portends little risk of either legal or ethical problems (at least on the face of it); but a volunteer in, say, a rehab clinic might find herself asked for medical or other advice which, if she gave it, could raise both legal and ethical questions should she lack therapeutic training and a license to practice and put her in violation of written or verbal agreements as a researcher and of the access terms she negotiated. A researcher who has thought through these issues and incorporates them in a research design will be better prepared if faced with their actualization, and the research design itself is likely to be judged more favorably because the issues are considered there explicitly, transparently, and reflexively.

Although for most field researchers, the advantages of being on location are self-evident, in order to explain this to others who are less familiar with them, a research design might well make explicit what is to be gained by the participation being proposed, especially as participating can make research more complex. The primary justification for participation is that it enables researchers to learn more about participants and their relationships to the subject matter of the project, including their activities and ways of thinking about all of these, than they would know without it. Whether sitting with mothers whose children have disappeared (Bayard de Volo 2009), running the winding machine on the shop floor (Shehata 2006), or using an electric shocker to hurry the cattle through a chute to the kill location in a slaughterhouse (Pachirat 2009a: 152), the researcher is potentially able, physically, emotionally, and verbally, to access participants’ experiences—of grief and fear, monotony and exhaustion, or solidarity and laughter—and the local knowledge that is embedded and carried in these, including the tacit knowledge underlying embodied practices. As Fenno indicates, in the chapter's epigraph, you are there—in the midst of data generation, “at the source,” rather than being the more passive recipient of “given” data (the Latin origin of the term: data are things handed over, given). Reflecting on these participatory experiences may bring initial expectations or assessments into sharp relief, suggesting other ways of understanding than what the researcher initially anticipated.

Another justification is instrumental: depending on the situation, some degree of participation may engender greater trust in the researcher on the part of participants, and this may facilitate interactions important to the co-generation of evidence. Allina-Pisano (2009: 57–9), for instance, recounts how her silence in a day-long meeting in post-Soviet Ukraine unnerved those in attendance: it reminded them of secret police surveillance in earlier times. Her participation as a “normal” member would not have raised eyebrows; it was her silence, in fulfillment of her perceptions of an appropriate researcher role, which generated participants’ discomfort and distrust. (This example also shows that it cannot be automatically assumed that observation without participation is either “unobtrusive” or less “intrusive.”)

Despite the possibility of extensive degrees of participation in some kinds of settings, contemporary researchers understand that they are not, and usually will not become, full members of that setting. Even “native anthropologists” (Narayan 1993) or “at-home” ethnographers (see Alvesson 2009, Leap 1996)—researchers studying settings of which they are or were at one time members—are commonly only temporary members of the research venue, usually intending at some point to leave to return to their research base.

That said, what is the researcher to observe, and how? In some disciplines and research projects, it is sufficient in designing the study to identify settings, actors, events, texts, archives. It is assumed (and at times, is treated as self-evident) that once on site, the researcher will situate himself in some vantage point that will enable general observation of a range of persons and/or activities and events. This would be the classic case of a participant observer, interview, or ethnographic study of a neighborhood or community or of a school, hospital, or other bureaucracy. Consider Zirakzadeh, for instance, finding a bar stool where he could eat pintxos (the Basque version of tapas) and strike up conversations with ETA activists. Increasingly, however, researchers are thinking of such projects in more dynamic ways, in terms of “following” something. One form of research, “shadowing,” entails following a lead actor (or more than one) around through her day. This, for example, was Wolcott's (2003/1973) approach in studying a school principal. In more spatially-oriented disciplines, such as planning and human or social geography, following key actors through their own physical terrains, such as homeless men and their places of congregation, is developing into a distinctive method useful in other research inquiries: “footwork” (Hall 2009) or the walking interview (P. I. Jones et al. 2008; see also Kusenbach 2003, Pink 2008, Stavrides 2001). In some cases, researchers are using map-making activities as ways of concretizing such movement (Wood 2009). Other researchers, often influenced by science studies and the sociology of knowledge, think of following “facts” (an idea borrowed from Latour and Woolgar's, 1988, study of a scientific laboratory; see also Brandwein 2006), ideas, policy issues, and the like. Following in this way might also entail asking what work the entity being followed is doing in its context(s). This is another way of asking about the meanings underlying or embedded in the item(s) in question.

The shift from a seemingly more static study to a more fluid one appears to be linked to changes in research topics (or perceptions of them) from relatively more bounded settings (villages, neighborhoods, departments) to less apparently bounded ones (multi-national corporations, multi-level governmental entities, networks). This has itself given rise to a new term: “multi-sited” studies. “Rhizomatic” studies are another version of this, in which the researcher moves among linked sites, following actors, concepts, issues, objects, or other entities central to the research question (see, e.g., Nicolini 2009).¹¹ We anticipate that earlier “island anthropologies” and “village sociologies” were themselves not fully bounded, although they may have seemed that way from an outside point of view; researchers certainly moved from one “site” to another within them.

Still, considering a researcher's location in the field setting (discussed further in the next section) in relation to what is “flowing” or otherwise moving through it, and how and where she is going to move about with respect to that position can provide a useful perspective on the research question. We suspect that, given its links to researcher positionality and the anticipation of multiple interpretations (see discussion of intertextuality, Chapter 5), this attention to flow and to following things may catch on as a key characteristic of interpretive research design.

Access, Researcher Roles, and Positionality

Ranchers are notorious about sizing someone up within the first five minutes of meeting them. I've seen it happen over and over, growing up with them. I also knew from growing up around ranchers that as a group, they are notoriously reserved. . . . Oil representatives don't trust anyone. . . . And elected and non-elected officials are guarded when questioned by outsiders.

. . . [A] life lived in Wyoming [US] among ranchers and roughnecks told me that ranchers and oil-men viewed academics as having “agendas.” Particularly, environmental agendas. If either of these groups suspected that I meant to do anything other than relate to them and their story or seek their expertise to help inform my research, . . . they . . . would have never opened up to me as much as they did. . . . I needed them to “size me up” from the start as “one of them”—as someone who could be trusted with their story and tell/analyze/interpret it from their perspective, not mine.

Preparing for the interviews, it became clear to me that government officials and oil association executives had become targets of serious criticism from ranchers (their untraditional nemesis) and environmental organizations (their traditional nemesis). This same preparation also told me that ranchers had become wary, even skeptical, of anyone seeking information from them concerning the political conflict I was examining. Once in the room [with the ranchers, oil association directors, or politicians], I made small talk . . . about being a “born and bred” Wyoming boy, or about my first job out of high school as a “roustabout in the oil patch,” or that “my dad was a former city administrator in a very small town in Wyoming . . .”. As the interviewees found out about me, they saw me as something other than a Ph.D. candidate. . . .

As you might suspect, from my perspective the “presentation of self” is a process that begins first with getting to know whom you are interviewing and then having the wherewithal to step outside of the researcher role long enough to relate to the person being interviewed. It is then that the interview begins. And even then, it's really just a continuation of the conversation you began when you introduced yourself to the interviewee.

—Robert Forbis

(Personal communication, 16 December 2010)

For interpretive researchers, matters of access and researcher role raise an additional methodologically relevant issue, related to interpretive research's goal of building contextually grounded knowledge and to its acknowledgment that scholars are human beings with specific histories, capacities, and characteristics. Known as positionality, it has two aspects, one of which might be called demographic, the other, locational or geographic. Forbis's research narrative, above, includes both, as they relate to his desired access to particular field settings and the persons who inhabit them and to his researcher roles.

First, researchers’ demographic characteristics and personal backgrounds may be critical to accessing research settings and/or actors. Sex, age, education, physical agility and (dis)ability, class, religion, race-ethnicity, language mastery and accent, birthplace, possessing a driver's license, and other elements in a growing list of intersectional factors comprising individual “identity” can all contribute to generating access to research situations—or to having it blocked. As Schwedler (2006), Ortbals and Rincker (2009), and the essays in “Field Research Methods in the Middle East” (2006) show, none of these factors is an automatic, universal key to open all doors: each one can play either or both ways, sometimes opening doors, sometimes shutting them. Samer Shehata (2006), for instance, notes the extent to which being male enabled entrée to some circumstances in his Egyptian factory studies while shutting off access to other potentially research-related settings. For Walby (2010: 649), it was addressing or avoiding questions concerning his sexuality that either opened up or closed off the conversations he was trying to have.

Second, as field relationships develop and unfold, the types and degrees of access achieved position the researcher geographically in the field setting, which itself can shape access to other circumstances and groups; and this, in turn, can profoundly affect what the researcher sees or does not see, learns and does not learn. This process can take place as one is drawn more deeply into one network rather than another, opening some doors while simultaneously closing down other avenues. Zirakzadeh's (2009) happenstantial refuge with one group of ETA activists rather than another is one example. Alternatively, it may unfold due to organizational or community rhythms or rituals that create various opportunities: the holiday party that enables rubbing shoulders with certain people—a boss, for instance—who might not normally be encountered (Rosen 1988); the ending of a contract or a promotion (Pachirat 2009a). These cases illustrate the extent to which field realities cannot always be anticipated and planned for, requiring flexibility in the field. Despite such limits on planning, researchers need to proactively reflect on the ways in which demographic and locational positionalities might affect their access to research settings, persons, texts, and so forth and—as discussed further in Chapters 5 and 6—on the possible effects of such positionalities on data generation and analysis and ensuing knowledge claims.

Access and Archives

The language of access has been most prominent in ethnographic or other field research, whether within domestic locations or in a country other than the researcher's institutional base. Access issues are also critical for archival research (regardless of the methodological approach). Archives may be restricted; researchers may need an institutional affiliation, letters, or security clearances (the latter, e.g., for classified government documents) to secure entrée. Even here, however, both prior knowledge and flexibility often play a role. As feasible, researchers try to develop some sense ahead of time concerning which archives, and which documents in those archives, are relevant to their research question. At the same time, they need to be open to surprises as they encounter new and previously unknown information, including material that points them to other archives, elsewhere, which had not been part of their initial design plans. In addition, their research may depend on special skills to make sense of—to “access”—the meanings in those documents, such as knowledge of legal terminology, specialists’ jargon, or languages other than their mother tongue (see, e.g., Bellhouse 2011, Ferguson 2011, Wingrove 2011).

With respect to sources of artifactual materials (documents, paintings, photographs), difficulties of access can vary, and it is not always useful to equate accessing action settings (neighborhoods, organizations, governmental offices) and actors (experts, officials, town residents) with accessing materials in archives, newspaper morgues, or other locations. In “Western” countries in particular, where archiving and public access are commonplaces, broad public availability of documentary materials may render matters of access to archival materials methodologically insignificant compared to the need to decide which among these many texts or images to analyze. The language of “selection” is a better fit for this particular situation (see, e.g., Oren and Kaufman 2006 and the discussion in the next section). For other locations, especially in developing countries, restrictions on the public availability of documentary materials can subordinate selection challenges to the need to identify and locate materials to begin with.

Access versus Case Selection

In thinking through the selection of places and persons in which and with whom the research question might reasonably be explored, interpretive researchers might appear to be enacting the same task as their research colleagues who are engaged in “case selection” for theory testing; but interpretive sensibility and research purposes render the two activities quite different. For positivist researchers, selection and access are typically understood as separate and separable. Consider three examples:

1. •   In experimental research design, the gold standard is the random assignment of cases (individuals) to control and experimental groups in order to render the groups statistically equivalent such that differences arising from the experimental treatment can be assessed. Individuals who are not “accessed” because they decline participation are replaced by others who do agree to participate.
2. •   In survey research design, the bias associated with respondents’ self-selection (the bane of the call-in polling so popular in entertainment venues, among other settings) is prevented through random selection from a “sampling frame,” as complete a list as possible of the population under study from which the sample is to be drawn. Randomized selection of research participants achieves a sample that is representative of the population being studied, such that inference from the sample to the population is unbiased, in statistical terms. Here, too, those who decline to participate are replaced with others, again chosen randomly from within the sampling frame (although high refusal rates raise the further question of how those who do choose to answer surveys differ from those who opt out).¹²
3. •   In positivist “small n” case study research, selection of cases is complicated by their potential non-equivalence (evident in comparative studies of countries, schools, hospitals, and the like) and, depending on the research question, by the subsequent lack of guaranteed entrée to the places, archives, or individuals (often, elites) that constitute the cases. But the methodological literature is silent on the work involved in accessing such cases once selected.¹³ Instead, drawing on an experimental template, that literature (e.g., King et al. 1994, Gerring 2007) focuses on proper selection to ensure that descriptive and causal inferences are unbiased.¹⁴

In all three of these, access, if treated at all, appears relatively unproblematic, conceptually. It is as if such selection were entirely within the researcher's power and control, without access difficulties interfering; and one case is treated as if it were as good as another for the purposes of causal inference (which would not be so for constitutive causality). In these three positivist approaches, the concern with case selection is driven by the goal of building general theory: cases do not have value in and of themselves. As Gerring put it with respect to case study research, its purpose “is—at least in part—to shed light on a larger class of cases (a population)” (2007: 20).

For interpretive researchers, by contrast, choices of cases and access are often intertwined—reasonably so, given the research purpose of understanding meaning-making in particular sites. In addition, for interpretive research, both randomized selection and substitutability of one case for another are problematic, for related reasons. We illustrate the point with the example of documentary research, in which the notion of the random selection of texts (“cases”) is rarely appropriate or feasible. For one, as mentioned above, random selection requires a sampling frame. Depending on the research question, compiling a complete list of documents from which to sample assumes the availability of and unfettered access to collections that are open and whose materials are organized and catalogued, prior knowledge of what these contain, as well as a priori judgment about which documents are likely relevant—all of which are problematic in various ways from the perspective of interpretive research, as well as the realities of archives and other repositories, especially in the “non-Western” world.¹⁵

Even more importantly, the possibility of substitution is material: one text, one photograph or painting, or one person is not as good as any other, in all situations. Rather than seeking texts or other cases for purposes of generalization, the interpretive documentary researcher wants not just any text but those that matter (or mattered) to the agents under study— another way in which context is significant. An interpretive documentary research strategy follows the “intertextual” trail from initial documents to related ones—Ferguson's “Oh, look, there's another one . . .” experience (quoted in Chapter 1)—as the researcher's knowledge deepens and his or her research question(s) become(s) more nuanced (L. Hansen 2006, C. Lynch 2006; see Chapter 5).¹⁶ The same holds for following ideas and persons in an interview- or observation-based study.

Furthermore, the language of “case selection” implies considerable researcher control—warranted in the case of experimental and much survey research, but potentially problematic in field research where denial of access to particular archives, persons or other sorts of “cases” cannot always be solved by simple selection of an equivalent, replacement document, individual, or case. For these reasons, the language of “case selection” is not appropriate to interpretive research design: it fails to recognize the significant ways in which access may be contingent on the identity of the researcher, as if any researcher, in any circumstance, possesses the ability to select any case at will.¹⁷ Interpretive design needs to be concerned with the choices of settings, actors, archives, and so forth and with a focus on access and its relational dimensions. By contrast with the language of selection, the language of access recognizes the embodied and inescapably social aspects of the research process.

Regardless of methodological approach, denial of access to field or archival settings is possible no matter how assiduously contacts have been cultivated. For instance, a researcher studying public hearings on impending urban renewal might find those hearings delayed by months or even years. Or increased press scrutiny of an agency the researcher had planned to study could mean that previously supportive gatekeepers withdraw their approval. The promulgation of the Common Rule in the US in the context of Institutional Review Board policies and practices, by making consent more bureaucratic and legalistic, has also complicated researcher negotiations for access. Still, there remains a key distinction between positivist and interpretive approaches to the relationship between access and knowledge claims, and it hinges on the centrality of contextuality to the research. Because the meaning-making of those studied is intimately linked with context, the complex issues of researcher access—including the relative power of individuals and groups, the possible kinds and degrees of participation, and positionality—need attention from the very beginning in designing an interpretive research project. These issues are understood as methodologically relevant to the research project, and they therefore need to be considered and taken into account in its design.

Design Flexibility: Control and Requisite Researcher Skills

Assuming that access can be achieved to one extent or another, together with a research-appropriate role, an interpretive design needs to consider the implications of contextuality for the execution of the research. Specifically, a researcher's expectation (or that of evaluators, such as thesis or dissertation supervisors or grant reviewers) that the project will be implemented as designed is likely to be frustrated. Interpretive research designs must be flexible due to field realities, stemming from participants’ agency. The researcher lacks control over research participants as well as over unfolding events, as Fenno notes (in the epigraph). But that is a principal reason that the researcher himself goes to these settings: he is interested in how meaning-making occurs there—without the kinds of controls on participants that characterize positivist approaches.¹⁸ Three aspects of this matter are worth elaborating: the relationship between control and positivist approaches; the logics of control and interpretive research processes; and the resulting skills and competences needed for interpretive field research.

Control and Positivist Research Design

The general expectation of positivist research designs as presented in methods textbooks, at least, is that they by and large can, should be, and are implemented as initially designed. In a laboratory setting this is a feasible expectation. Hinging only on human subjects protections, the researcher utilizing a laboratory has complete control over the research setup, participants’ recruitment, the tasks they will be asked to perform or will be subjected to, and so on. It is because of these controls—and the typically time-limited tasks requested of experimental “subjects” (a term originating in medical and psychological research; Danziger 1990: 53–4)—that the Principal Investigator (PI) can hand over data “collection” work to trained assistants. The PI's role in all of this is to design the study, analyze the data, write the research report, and train and manage lab assistants, relieved of the responsibility of having herself to interact with every research participant.

All of the processes that the PI has detailed in the design—recruitment strategies, scripts for research assistants to read to participants, protocols for debriefing participants, etc.—are expected to be executed without significant variation. If variations in these processes do occur, it may mean that research results will have to be discarded: according to the logic of this research approach, such variations, rather than the treatment variable of interest, may have produced the experimental effects observed. This idea is extended to non-laboratory-based research, such as surveys, focus groups, and positivist case study research. In research designs for those kinds of projects, the underlying logic of inquiry remains the same in that variations in execution of the design are understood as liable to produce “flawed” results.¹⁹ For example, Yin (1989: 50) tells case study researchers that “if the relevant research questions really do change, the investigator should simply start over again, with a new research design,” noting that shifting research questions produce “the largest criticism of case studies.” Likewise, Gerring (2007: 149) recommends redesigning a study “after the new (or revised) hypothesis has been formulated” and then collecting additional evidence. Otherwise, in his view, the study should be termed “exploratory” or the evidence produced based on the initial design should be “jettison[ed], or deemphasize[d].”

Underpinning the expectations for invariant implementation of a research design may be its metaphoric source in architecture: a poorly designed building or bridge may crumble or collapse due to mistakes in the architect's blueprint; the designer's failure to understand the physical world—with whose constraints engineers have to contend—signals professional incompetence. So, it seems, is the positivist understanding of experimental research design: any need to vary from it in the execution of that blueprint is deemed a design flaw and a professional failure. This is part of what colors the negative and erroneous understanding of interpretive research design's flexibility.

The Logics of Control and Interpretive Research

Howard Becker (1998: 193) contests the kind of logic that requires researchers to “gather new data from a new sample” before they can take advantage of new insights:

Such an unrealistic requirement would, of course, put an end to qualitative historical research, because there is no gathering a new sample, and would make studies like Lindesmith's, based on interviews with hard-to-find addicts, impossible in any practical sense. More to the point, it treats as a sin what is actually a major scientific virtue: the willingness to revise your thinking in light of experience, the dialogue of evidence and ideas Ragin [citation] puts such emphasis on.²⁰

Yet it is not just openness to revisions that is at play in interpretive research. For one, the interpretive goal of understanding the meaning-making of those studied (as contrasted with the positivist goal of mirroring the world with degrees of certainty) eliminates the concerns associated with positivist forms of causality. Changes in design cannot, therefore, be understood as threats to the trustworthiness of a research endeavor that does not understand causality in this way (as discussed in Chapter 3).

But more than that, positivist expectations for inflexible design implementation—as well as associated judgments that variations from the initial design threaten its trustworthiness—do not fit the logic of interpretive research. Flexible design is central to interpretive research for two reasons.

First, the abductive and hermeneutic reasoning that undergird interpretive research processes are both built on successive phases of learning: the researcher starts with what he knows, as encapsulated in the initial design, but his investigation builds on itself in a reiterative, recursive fashion. Although he begins with insights from existing literature and/or prior knowledge of the field setting, if these insights prove unhelpful, it follows from this logic of inquiry that they must be modified, including being discarded, if necessary. Interpretive logics and processes see researcher sense-making and learning as deepening with ongoing experiences in the field. As Drori and Landau (2011: 25) write, their research relationships with the nuclear scientists in the organization they studied were “based on ongoing reciprocity and mutual learning,” a relationship that gained them “access to almost all divisions, branches, and laboratories,” which itself allowed them “to observe a wide range of organizational activities and situations, within certain limits.” Expanding and elaborating on the initial research question in its redesign is, in light of such ongoing learning, normal.

Second, a researcher entering the settings of research participants (rather than inviting them to the researcher's laboratory or controlling them through invariant survey questions or interaction scripts) lacks control over those settings and their individual and organizational inhabitants. Rather than being research “subjects” who participate in (positivist) research on the researcher's terms, in interpretive research it is the researcher who participates in the locals’ activities, in their settings, on their turf. Participants are recognized as having their own agency over their own movements, meaning-making, will to participate, and so forth. This means that they are understood as having the power to affect initial research designs actively in various ways, such as refusing to accede to requests for entry, for interviews, for reading permissions in archives, and so on. Even if they do not decline altogether, participants may still desire and choose to participate in ways other than those the researcher has designed for them. On the other hand, they may also be enthusiastic partners in the (co-)generation of evidence (as discussed in Chapter 5), assisting the researcher in learning about their settings, of which they have local knowledge.

But the researcher cannot know which of these (or what combination) she will meet in the field; and so the design—as well as the researcher!—needs to be flexible. As Walby (2010: 647, 645) observes,

Introduction of a speech act or gesture into the conversation between researcher and respondent can have unanticipated consequences for the rest of the dialogue. . . . However much scripts and roles guide our interactions with others, there is always an open tendency, such that the meanings which will be achieved from an encounter cannot be fully anticipated.

Interpretive Researcher Competence and Skill

That research participants are considered (and treated) as having their own agency is not problematic for interpretive researchers in the ways it would be for experimental and other positivist researchers, who depend on a level of compliance, and even submissiveness, from “subjects” in order to execute their designs invariably. The interpretive focus on participants’ worlds, their local knowledge, means that the researcher cedes control to them as the experts in their own lives. However, this very lack of control means that interpretive researchers need to cultivate particular competencies and skills to maneuver effectively in the field(s) inhabited by agential research participants, skills not required of those following a positivist methodology.

Accepting the limitations of researcher control in field settings—recognizing that participants are active protagonists in their own worlds—means conceiving of research as fundamentally and dynamically relational in form and character (rather than contractual)—with all the attendant rewards, uncertainties, and risks life in “natural” research settings entails. Whether conducting conversational interviews or observing (with whatever degrees of participation), recognizing the agency of research participants means the researcher has to adapt to the field setting (rather than the research participant adapting to the experimental setting or the categorical choices available in a survey questionnaire). Instead of the skill set needed for executing a research design as invariably as humanly possible (and, later, for choosing and running appropriate statistical tests), the researcher needs an openness to learning and change, the willingness to revise thinking in light of experience (noted above), and a high tolerance for ambiguity, together with improvisational skills and an understanding of research design that makes room for their use (and, later, the ability to manage the myriad details of observations, interviews, and documents encoded in words and to choose and apply the most data-appropriate method[s] from among the many available interpretive analytical techniques).

Let us turn, once again, to the research of Zirakzadeh (2009; discussed initially in Chapter 2) to illustrate these several points. When he set out to conduct research on ETA separatists in Basque country, Zirakzadeh arrived with a research design that was strongly informed by the literature that characterized his theoretical approach. He learned, however, that that literature was a poor guide to the lay of the land. Instead of starting over de novo (as some positivist methodologists, such as Yin, 1989, and others would have had him do), he improvised with respect to his existing plan, revising his research design in the field.

One story from that research, alluded to previously, clearly illustrates the kind of “on the fly” revisions that are commonplace and considered not only normal but legitimate in such research. When, by happenstance, he ducked for cover from police shooting at members of one ETA group and found himself—“in sheer desperation” (2009: 105)—sharing the doorway with some of them, he was taken for a sympathizer by group members. That evening, he found himself the center of attention at the bar they frequented, where he had hitherto been ignored. But now, he also found himself persona non grata among other ETA groups from whose perspective he was no longer “neutral.” A particular self-protective impulse, a particular doorway, made a world of difference to his research.

We see in this example an illustration of a researcher's willingness to revise his thinking, as Becker put it. This means being open to new experiences and insights and being open to learning new things as the research unfolds. Stacia Zabusky, in her ethnography of the European Space Agency (1995: 129), described the need for “being poised for movement, sensitive to flows and trajectories, . . . to disruptions and re-orientations. I had to be prepared to ride the ongoing currents of capital, people, and services . . . .” The inability to know ahead of time what will happen and to be able to plan for it—constraints that inhere in this openness and flexibility—requires a high tolerance for ambiguity. On a practical level, a rigid design that includes an invariable protocol for interactions in the field or in an interview would likely lead participants to see interpretive researchers as incompetent—“socially stupid,” lacking in social intelligence (see, e.g., Gardner 1983, Goleman 1995)—in their inability to respond in ways appropriate to the situation at hand (see also Burkhart 1996: 34). Or it may lead participants simply to be bored, as Zirakzadeh (2009) says his respondents were (see Chapter 2), because the pre-planned questions are out of touch with their own lived realities and the researcher is incapable—by virtue of the constraining research design, not intel-ligence—of translating them in ways that make more sense to those involved.

The need to interact effectively with dynamic human interlocutors in a dynamic field setting requires the ability to improvise. As musicians and other performance artists know, improvisation does not mean “making things up” de novo; rather, improvisation is a skill that is learned and perfected through practice (Yanow 2001, 2009). Improvisation's “yes, and” attitude, for example, requires a researcher to listen to her interlocutors and respond in ways that build on their ideas; a negative response—a “yes, but”—can block and even shut down the exchange. And improvisational skills cannot be applied invariably across all participants. Instead, the researcher needs to be attuned to various participants’ styles of interaction and sensibilities (e.g., which topics are “off limits”; Fujii 2010) and adjust her responses accordingly. Observational and fieldnote practices also need to be adapted to the context of the field setting (discussed in Chapter 5).

Interpretive field researchers, then, need listening skills, improvisational skills, and the knowledge and confidence that being flexible in the face of field realities is an appropriate and legitimate response, befitting this logic of inquiry.²¹ An orientation toward flexibility and even improvisation is also part of an interpretive sensibility in archival research. Moving from text to text in a sort of textual ethnography (see also Darnton 2003, P. Jackson 2006), the researcher looks for text-based clues that will lead in fruitful directions, and he wants and needs to be able to follow these.

Because this skill set is learned over time, through the doing of research in a kind of apprenticeship, guided by more experienced researchers (dissertation advisors, more senior colleagues), and because the researcher is herself the instrument of inquiry and sense-making, in interpretive research the PI interacts directly with research participants (perhaps with the aid of a translator, as needed). Even when research assistants are involved, they will more commonly work in such an apprentice-like fashion, closely supervised and learning their research craft alongside the lead researcher (see, e.g., Maynard-Moody and Musheno 2006).²² This presents a strong contrast with the training of research assistants in conducting experiments or executing telephone or field surveys: like the research design, their training is also “front-loaded,” with learning taking place prior to the “application” of the “instrument.” After training, they are commonly left on their own to execute the research (and their training entails learning not to react to the personal characteristics or inquiries of the experimental “subjects” or survey respondents).

Why produce a research design if flexibility is the rule rather than the exception? Does such flexibility render designing inconceivable? As discussed in Chapter 2, knowing that designs change does not excuse the researcher from asking significant and well-developed questions at the starting point in the circle-spiral of learning and understanding. First, interpretive research takes place in a theoretical context, and a research design is useful for considering previous research and placing the proposed project in its context, even if—or, as often, when— theoretically-based expectations are dashed by field realities. Second, it is certainly possible that a researcher's best attempt to access sources of evidence will fail, perhaps by chance, e.g., the vacationing, transferred, or suddenly-deceased senior executive whom the researcher had anticipated interviewing. Or evidence may not be available due to the agency of research participants and non-participants in the field who refuse access. Such unavailability may be unfortunate; Zirakzadeh's desire to research all ETA groups, for example, was stymied. However, although his actions cost him access to some groups because they no longer perceived him as politically neutral, he gained fuller access to one group, and all of the unfolding events led him to a greater appreciation for the complexity of the social and political landscape he was researching. Thinking through his initial research design provided him a baseline against which to compare field realities; it was the starting point from which learning could proceed.

Flexible interpretive research designs are, then, not only possible; they are needed, because they focus a researcher's attention on possibilities and limitations that need to be anticipated, including the flexibility to revise them in the field. Although at odds with the positivist conception of design, flexibility is a mark of competence in interpretive research and of a good interpretive research design. Its absence will stymie researchers in their efforts to generate data, the topic to which we now turn.