Ethnographic Traditions and Common Misconceptions

What constitutes an ethnography? “Ethnographers are committed to going out and getting close to the activities and everyday experiences of other people. ‘Getting close’ minimally requires physical and social proximity to the daily rounds of people's lives and activities” (Emerson, Fretz, and Shaw 1995, 1–2). In the hands of the interpretivist, ethnography is “the art and science of describing a group or culture,” the aim being the explication of meaning (Fetterman 1989, 11; Geertz 1973, 5). For the positivist, it is a tool to get the causal story right (Laitin 1998).

We deliberately employ a loose definition of ethnographic methods meant to capture their contribution to both the interpretivist and positivist intellectual traditions.³

They are those methods that seek to uncover emic (insider) perspectives on political and social life and/or ground-level processes involved therein. Ethnographic methods typically include some combination of the following: living in the community being examined; learning a local language or dialect; participating in the daily life of the community through ordinary conversation and interaction; observing events (meetings, ceremonies, rituals, elections, protests); examining gossip, jokes, and other informal speech-acts for their underlying assumptions; recording data in field notes that attempt to produce daily accounts of social and political life; in other ways letting trust and emotional engagement be of benefit to the research project. In short, ethnographers immerse themselves in their research setting and attune themselves to the daily rhythms of their subjects' existence.

Three points of possible confusion about ethnography loom. First, ethnographic approaches use qualitative data, but not all qualitative methods are ethnographic. An institutionalist interested in the early evolution of the U.S. Congress, for example, may gather qualitative data by examining congressional minutes, but this approach is not ethnographic in nature. If the same institutionalist turns to private papers, letters, or biographies of important early congressional actors, s/he may reveal insider perspectives. But this, too, falls short of ethnography. Ethnographies deal with living individuals, social groups, and their environments, rather than historical figures. The ethnographer interacts with subjects regarding his/her observations and asks questions for clarification. A further difference is that the ethnographer is keen to record the mundanity of what is observed, probing interactions that may be deemed irrelevant or too ordinary to require comment by the actors themselves. In sum, while oral history, archival research, and open-ended interviews are qualitative methods of data gathering, and while all of these methods may complement the use of ethnographic methods, they are not ethnography.

Second, today's approach to ethnography is crucially different from the methods used by early cultural anthropology. Anthropology historically privileged the immersion of the ethnographer in the remote, isolated village that was often assumed to be culturally static rather than dynamic and changing.⁴

As many sociologists and anthropologists now point out, in a world of tremendous complexity such an assumption is both inappropriate and increasingly misleading. There is a widely-recognized need to go beyond micro-scale studies that artificially freeze the “ethnographic present” and instead learn how everyday life is linked to, and changes in relation to, political processes that may transcend the boundaries of community, region, and state. The irony is that although political scientists, as students of power and politics, are well positioned to consider these links, the discipline tends rather to ignore them.

A third fallacy about ethnographic methods is to equate them with a more recent trend in cultural anthropology. In parts of that sub-discipline, the ethnographic method has achieved star status. Allowing for ongoing dialogue (“reflexivity”) between subject and object, the method has become the defining feature of entire bodies of research. In its worst forms, this reflexivity can become a sort of transcendental principle that rivals any methodologically narrow navel-gazing practiced in other disciplines.⁵

These excesses, however, are the exception. The norm is for ethnographic work to offer potentially profound contributions to the body of knowledge about social and political life—no matter the intellectual tradition of the researcher.

If these tools still seem to belong more to the anthropologist than to the political scientist, this is testimony to the uneasy place that ethnographic approaches have had in our discipline. While most political scientists, for example, recognize the contribution of Richard Fenno (1990) to our understanding of the U.S. Congress, such recognition is the exception rather than the rule; ethnography generally takes a back seat to other approaches. Most “methods” courses at Ph.D.-granting institutions do not cover ethnography (Schwartz-Shea 2001).

On a spectrum from purely deductive to purely inductive reasoning (with most studies falling somewhere in the middle), ethnography invites inductivism. While many sociologists have embraced the use ethnographic methods as part of “shopfloor studies” to build mid-level theories from the ground up, the deductivist turn in political science fosters a reluctance to use these methods; they are tools of the trade that move from data to theory-construction, rather than the other way around. But perhaps there is some irony in the fact that rational choice perspectives— which many view as the epitome of deductive theorizing—are particularly well suited for use in combination with ethnographic methods. Both approaches tap the micro-foundations of political processes and social life. At present, such a methodological marriage rarely occurs.⁶

Explanation from the Inside Out: What Ethnographic Methods Can Do

Far from becoming outmoded as quantitative data are increasingly available, ethnographic methods remain critical to political analysis. At a minimum, insider perspectives on political processes are important for three reasons. First, attention to insider perspectives posits some agency for social and political actors. This is a welcome change. For many political researchers, the best explanation is a structural explanation; an agent-centered explanation is assumed to rest on the idiosyncrasies of individual actors that are, ipso facto, considered resistant to generalization. Unless we believe that agents have no bearing on political outcomes, we must employ tools that uncover, rather than presuppose, individuals' motivations and behaviors.⁷

By examining how actors themselves view the myriad political situations in which they are involved, we bring to the fore their influence on political outcomes, constrained as it is by structural factors.

Even for those who prefer structure-centered to agent-centered explanation, ethnography is a powerful check on analytic reasoning. Ethnographic methods help the researcher discover whether a given explanation is consistent with insiders' viewpoints. To be sure, good research does not merely mimic these viewpoints, but neither should it be out of sync with the latter. If outsiders' explanations are fundamentally inconsistent with insiders' perceptions, then a specter haunts the explanation—that of attributing insider beliefs to “false consciousness.” Ever human, ethnographic informants can be misguided or simply wrong, but it does not follow that we should dismiss their perspectives wholesale.⁸

Indeed, the inconsistencies between outsider and insider beliefs, between rich and poor's self-representations, between men's and women's version of events can become an analytical focal point enriching our understandings of power relations and politics at the local scale.⁹ Methods such as ethnography help us access the same events or processes from multiple points of view, which ultimately may provide great insights into questions of power and perception.

Second, beyond the reasons why giving voice to insiders is important as an end in itself, consideration of such viewpoints can help to explain behavioral outcomes. Take, for example, the social movement literature. Elites frame the interests and goals of a movement, using popular tropes and symbols to increase its mass resonance. This resonance, in turn, hinges on how individuals link their self-understandings to the interests and goals of the movement. Without a sense of how elites “frame” their movements in daily interactions with potential and actual supporters, we lose key insights into the origins and trajectories of mobilization.¹⁰

Third, some forms of political competition hinge directly on self-perception. Here, we have in mind what has come to be loosely called “identity politics,” i.e., the power dynamics that surround claims to membership in ethnic, national, racial, religious, sexuality, or gender groups. In such contests, the collective meanings that underpin groupness are critical. What political meanings are linked to being a woman in urban Iran, or a peasant in rural Russia, or a Hutu in contemporary Rwanda? How do these varied meanings influence political processes? For identity politics, competition should not be reduced to group demographics and membership in one or another identity group; it is also about what membership in one or another group is understood to mean and how these meanings influence interests and behaviors. Insider perspectives, uncovered through ethnographic methods, help address these issues.

Critically, the value of ethnographic methods extends beyond the ways in which they capture insider perspectives. First, for positivist studies, they help to highlight causal links. Where other research tools are effective in charting out correlations, ethnography can put flesh on these relationships. It can establish a micro-causal story that adds credibility to our theories. Potentially, therefore, ethnography does more than simply propose theoretical connections between our variables; it actually views how these variables play themselves out in the real world (see, for example, Laitin 1998).

Second, ethnography brings to the fore alternative causal paths. As Ragin (2000) points out, in much statistical analysis the assumption is made that causal paths are homogenous. Useful as this assumption may be, ethnography entertains the possibility that we may arrive at any given political outcome through a variety of mechanisms. Close familiarity with case-data—which is the hallmark (though not the monopoly) of ethnographic work—can help to justify or challenge this assumption. Thus, for example, Villalón (1995) successfully charts out an alternative relationship between the political system and Islam precisely because such an alternative becomes evident through ground-level familiarity with the role of Sufi orders in rural Senegal.

Third, in cases where government statistics are suspect, media outlets are controlled by political interests, and poverty, lack of infrastructure, illiteracy, or political violence impede survey research, ethnographic approaches are often the most reliable and practical means of collecting data. Examples abound. Berliner (1957) famously examined the operation of Soviet enterprises from the perspective of the shop floor and contributed lasting insights about the nature of the Soviet political economy. Walder (1988) more recently examined similar issues in the Chinese context. In related ways, Wedeen (1999) was able to use ethnographic data to provide critical insights about authoritarianism in Syria. These insights—generated at the ground level—have initiated an ongoing reconceptualization of the relationship between regime legitimacy and authoritarian rule.

The fact that ethnography is particularly able to penetrate opaque circumstances, however, should not lead to the mistaken conclusion that ethnography is an analytic tool of last resort—a crude implement to make use of in the absence of more refined instruments.¹¹

Ethnography focuses the research lens at a level of analysis that is often ignored or assumed to be insignificant—the micro-level. It has helped, for example, Weaver (2003) and Bayard de Volo (2001) penetrate the rhetoric and formal structure of prominent organizations to reveal the multi-layered conflicts that influence their political dynamic. In fact, even in relatively transparent advanced industrial societies close ethnographic work could help to shed new light on government statistics, media portrayals, and the influence of poverty and lack of education on political processes. The dramatic accounting scandals in the United States in 2002 remind us that the power enacted in informal networks of individuals may have a profound social, economic, and political impact.

If I Had a Hammer: Ethnography's Limitations and Their Management

An aphorism holds that, to the hammer, everything looks like a nail. Likewise, no methodological tool can do everything. Like other approaches, ethnographic methods are most successful when combined with other tools. Here, we address a series of possible reasons why some researchers might choose to avoid using ethnographic methods. We suggest that while some of these reasons have surface validity, the judicious combination of ethnography with other methods can help to overcome important research challenges.

A first objection to ethnographic methods is that they rarely conform neatly to ideally scientific research designs. The personal, time-consuming, face-to-face encounters that are the hallmark of ethnography clearly limit research to a relatively small-N, which ultimately inhibits our ability to generalize to grander theory. This is true if ethnographic methods are used in isolation, but the thoughtful researcher either combines them with other research tools or engages a dialog with the work of others who do. Moreover, the potential ethnographer may be weighing a tradeoff: ground-level research may produce higher-quality data at the expense of generating higher quantities of it. Depending on the research question, this may be a perfectly justified strategy. Agrawal and Goyal (2001), for example, offer an ethnographically well-grounded examination of cooperation on natural resource issues in the Himalayas.

A related objection is that the ethnographer does not select informants through random sampling. Rather, s/he seeks to establish relationships with informants based on mutual trust. On this count, the critic objects that ethnographic data is inherently biased, both in a statistical and a normative sense. Here, too, there is a tradeoff. What the ethnographer loses in statistical reliability (and no thoughtful ethnographer ever loses sight of the peculiarities of the population being researched), s/he seeks to gain in validity. Extended interaction with informants can give the researcher a better idea of whether or not fruitful questions are even being asked. Furthermore, much ethnographic work seeks to understand the range of human interaction in a population, and random sampling could impede such understanding (Stewart 1998, 35). Finally, random sampling may have unintended negative consequences in practice for the ethnographer immersed in a community over a significant period of time: it can produce situations in which certain individuals are ignored, thus creating a strained research environment and potentially foreclosing access to data-rich locales in the future.

In addition to bias in the statistical sense, ethnographers are also at times accused of “going native”—that is, of normative bias. Indeed, deep familiarity with the microprocesses and perspectives of a given context may lead researchers to adopt, without critical self-reflection, insider viewpoints. Maintaining analytic distance from such viewpoints can be difficult, especially when the researcher becomes personally familiar with political and social actors. This challenge is even more acute when the researcher comes to rely on these same actors for future access to research sites and resources. Ultimately, however, this is scarcely different from other types of data-collection. In research about human beings, we may come to rely on some of those human beings. This is a professional hazard, and we can only hope that colleagues are familiar enough with the empirics to identify how such reliance might affect findings.

A frequently heard further objection is that intensive (rather than extensive) data collection and interpretive analysis cannot produce insights that “travel” well to other situations. This objection begins with an important premise: ethnographic data are closely tied to specific temporal and spatial contexts. But lack of portability is only one possibility among many; it is ultimately an empirical question whether locally generated insights speak to other locales and regions. Ideally, through intensive fieldwork, strong familiarity with similar cases, thorough review of research findings in a wide variety of contexts, and comparative analysis, the ethnographer offers a contribution that speaks broadly across space and time, sparking cross-case dialogue and debate, if need be.

Most ethnographers, however, recognize that pure replicability of results is impossible. “Ethnography no longer claims to describe a reality accessible by anyone using the right methods, independent of the historical or cultural context of the act of describing” (Agar 1986, 19). Those researchers interested in establishing causality (whether couched as a contextualized causality or a more general theory) thus use ethnography in conjunction with other methods. Yet, those who work in the positivist tradition should not mistake ethnographic work for an entirely subjective endeavor. To the extent that it is interpretive, the focus of interpretation is a concrete group of people in concrete situations. Although two ethnographies on the politics of the same community may differ, those differences are limited by the fact that both study the same community. While the ethnographer cannot aspire to laboratory-like controls, insights must be subjected to critical judgment and peer evaluation. Specification of the research environment and procedure is most important here (Stewart 1998, 30). The research context and methods should be spelled out transparently, so that reviewers and readers can themselves judge the quality of the data and its interpretation.

Contrary to an often-heard criticism, ethnographic methods do not merely generate “stories.” At best—the criticism goes—they merely give rise to hypotheses; it is a preliminary (read: “inferior”) sort of research because it cannot “test” propositions. In a limited sense, there is something to this criticism. Given small-Ns and of necessity less systematic data-collection techniques, few ethnographers claim to be offering conclusive tests of alternative hypotheses. In a more profound sense, the criticism is wide of the mark. Ethnography is readily employed to test hypotheses to determine whether and how well general theory applies to a specific case. For example, James Scott (1985) and Susan Stokes (1995) both use ethnography (combined with other methods) to “test” the Gramscian theory of hegemony. The ethnographer offers an intensive evaluation of possible causal chains. Unless we are to rest content with statistical correlations, such in-depth evaluation of causality is critical. Further, the ethnographer uncovers locally attached meanings; in the absence of knowledge of these meanings, the research can miss or misinterpret important political processes.

Ethnography can be combined with other methods in such ways as to produce richer understandings or enhance validity (Seale 1999, 53–56). Such methodological “triangulation” involves the use of several data-gathering methods, in effect producing a checks-andbalances system such that biases, misperceptions, false statements by informants, and other potential problems are more readily revealed or rooted out. Using one-on-one interviews in addition to participant observation, for example, can enhance our understanding of the politics of everyday life in that an individual might be more candid or forthcoming about power relations in a private interview than he or she would be in public.¹²

The methods employed may all be qualitative—for example, participant observation, openended interviews, and document analysis. Or the research might employ a combination of quantitative and qualitative methods—for example, comparing large-N survey research with findings based upon ethnographic methods.

There are many examples of multimethod political science research that includes ethnography. Laitin and his research team (1998), for example, successfully incorporated ethnographic methods into a broader research design. Susan Stokes (1995) combined participant observation in a Peruvian shantytown with one-on-one interviews and a large sample survey. After discovering—through ethnographic work and interviews—a clientelist-radical dichotomy amongst leaders, she used a large-N survey both to determine whether this political culture was reflected in the mass population and to test for alternative explanations for this dichotomy. James Scott's (1985) ethnographic detail of Malaysian peasant everyday resistance is combined with data on state, regional, and local political structure, economy, and agricultural innovations gathered from primary and secondary sources to provide a fascinating analysis of peasant resistance. Through use of ethnographic work, semi-structured interviews, and surveys in two U.S. schools, Clarissa Rile Hayward (2000) challenged prevailing theories of power relations.