After reading this research guide you will be impressed with the volume
of methodological writings that are available. This has not always been
the case, and we have indeed come a long way. In anthropology, for example,
there have been many stories (some perhaps apocryphal) about the paucity
of advice available to earlier generations about how to do fieldwork. Some
of those stories are reproduced here.
In the late 1940s, Charles Wagley, one of the best ethnographers our discipline [anthropology] has ever produced, asked Alfred Kroeber for advice on teaching a course about fieldwork. "I was hoping for some wisdom as to how to organize my course," Wagley said, "but instead I was cut short. 'Some can and some can't, [Kroeber] said (if I remember his words correctly), and he passed on to a more interesting subject . . . I did teach the course on field methods but I cannot remember how it was organized or what I said, and I did not teach it again" (Bernard 1988).
Here is another.
The charming and intelligent Austrian-American anthropologist Paul Radin said that no one quite knows how one goes about fieldwork. Perhaps we should leave the question with that sort of answer. But when I was a serious young student in London I thought I would try to get a few tips from experienced fieldworkers before starting out for Central Africa. I first sought advice from Westermarck. All I got from him was 'don't converse with an informant for more than twenty minutes because if you aren't bored by that time he will be'. Very good advice, even if somewhat inadequate. I sought instruction from Haddon, a man foremost in field-research. He told me that it was really all quite simple; one should always behave as a gentleman. Also very good advice. My teacher, Seligman, told me to take ten grains of quinine each night and to keep off women. The famous Egyptologist Sir Flanders Petrie, just told me not to bother about drinking dirty water as one soon became immune to it. Finally I asked Malinowski and was told not to be a bloody fool (Evans-Pritchard 1937).
A somewhat more recent description of the preparation for fieldwork is provided by Elliot Liebow in his 1967 book about African-American men in Washington, DC. His advisor, Hylan Lewis, described his own vision of science: "The scientific method is doing one's darndest with his brains, no holds barred" (Liebow, 1967:235). Liebow goes on to describe other advice he got as he prepared for field work in an urban environment.
Having partially digested the project literature, I told the director [Lewis] that I was ready to get started. He suggested a neighborhood that might be "a good place to get your feet wet." His instructions were: "Go out there and make like an anthropologist." "Out there" was not at all like the Indian village of Winisk on the Hudson Bay in which I had done field work. I was not at all sure how one "makes like an anthropologist" in this kind of "out there." Somewhat wistfully, perhaps, I thought how much neater things would be if anthropologists, as they had done in the early thirties, limited themselves to the study of "wholes," a tribe, a village, or some other social unit with distinct boundaries and small enough to be encompassed in its entirety by direct observation (Liebow 1967:235).
Fortunately the advice available to researchers about these methods has improved. We hope this guide will contribute to that body of advice. It should be very useful to experienced, motivated researchers who are willing to peruse the topical and annotated bibliographies and to try and capture from them whatever techniques might be most pertinent to their situation. In spite of the progress of recent years there is still work to be done. There are many areas of methodology that are likely to see continuing development in the coming years. We hope the literature presented here will provide a basis for digesting those new methods as they come along, and for some perhaps, a basis for innovating those new methods.
The quotes above seem to turn on the question of whether qualitative methods can be made explicit, or whether they way they are carried out mostly involves intuition! Can the qualitative researcher can ever fully describe his or her methods? Whyte's (1955), in an artful appendix to his classic Street Corner Society, seems to come down somewhere in the middle. He attempts to give an account of his methods while warning the reader that the process can be a very personal one, not possible to fully explicate.
This account of living in the community may help also to explain the process of analysis of the data. The ideas that we have in research are only in part a logical product growing out of a careful weighing of evidence . . . Often we have the experience of being immersed in a mass of confusing data . . . We come up with an idea or two. But still the data do not fall in any coherent pattern. Then we go on living with the data-and with the people-until perhaps some chance occurrence casts a totally different light upon the data, and we begin to see a pattern that we have not seen before. This pattern is not purely an artistic creation. Once we think we see it, we must reexamine our notes and perhaps set out to gather new data in order to determine whether the pattern adequately represents the life we are observing or is simply a product of our imagination. Logic, then plays an important part. But I am convinced that the actual evolution of research ideas does not take place in accord with the formal statements we read on research methods. The ideas grow up in part out of our immersion in the data and out of the whole process of living. Since so much of this process of analysis proceeds on the unconscious level, I am sure that we can never present a full account of it (Whyte 1955:279-280).
Of course, this did not stop Whyte from later writing a text on fieldwork methodology! (See Whyte, 1984.) Of course Whyte was writing about ethnography, which is really a long-term combination of qualitative methods that at times is hard to distinguish from simply being part of a human community. Yet today qualitative researchers have been able to distinguish a wide variety of specific qualitative methods. Efforts have been made to carefully describe each one so that the need for intuition is minimized. The body of this document should help the reader to begin to understand the kinds of methods available and point to literature that will more fully introduce the reader to each.
FEAR AND LOATHING IN QUALITATIVE RESEARCH
Before we embark on the parade of qualitative tools available, we want to address first the apprehension that researchers may have in using qualitative methods at all. Our experience in consulting on qualitative and ethnographic methods is that no amount of readings and meetings can surmount a fundamental issue: Most researchers have concerns and fears about these methods. They have learned the canon of "classical" (read positivistic, quantitative, biomedical, experimental) research methods and design. They worry about bias; reliability and validity; sample size; sample generalizability; analysis techniques; and so forth. Qualitative and ethnographic work seems to have few or no answers to these kinds of concerns about "normal methods". No amount of consultation can be effective unless these kinds of fears and worries are addressed directly. We first discuss a most basic question: What are the purposes of methods? Then we turn to the terms that quantitative researchers bring with them, and explore how they might be thought to apply to qualitative research.
WHY METHODS?
Why do researchers use any method at all? First, methods are not primarily for the purpose of satisfying technical criteria, but rather to bring us closer to what it is we are trying to understand. Closer in every sense: more intimately, holistically and directly; more truthfully; more comprehensively; and so forth.
Second, methods are also to help us verify: to make what we think we have discovered through exploration, more plausible and probable. Third, methods are also here to discover objective truths about the world through assembling evidence, making testable statements, generalizable propositions, coherent arguments, and so forth. Qualitative methods can and do have this goal also. Perhaps the concern and special ability of qualitative methods to get at meanings and subjective experience may seem to make it less able to discover objective truths. It is likely, however, that contextual factors, and interpretations (subjective aspects) of social phenomena, are going to be a powerful part of any objective report about the world.
Fourth, methods exist to reassure audiences we want to reach for personal and professional reasons. We use methods to gain the attention and favor of these audiences. To the extent that methods are in part constructed by these audiences and are a kind of discourse in a social world of influence and power, we should recognize that choices of methods are made with these social worlds clearly in mind. Will journal X accept my study if I don't use a certain measure? If I use ethnographic data? Will Foundation Y or Federal Agency Z fund my proposal if I use such methods? Will Professor W give me a good recommendation or consider me as a student to work with her, if I do?. Although some researchers may fear that qualitative methods will be viewed as less "rigorous," or as competitive with quantitative methods, using these methods in complementary fashion can actually strengthen, not weaken, our abilities to reach important audiences.
Finally, we are personally anxious about dealing with the world and with ourselves. We select certain methods and shun others not only for the above reasons, but because we are anxious about doing them. We may be troubled by intimate unstructured interviews; participant observation; or confronting nonrational elements of human behavior. Hiding behind our professional roles, standard surveys, and assessments avoids contemplating these messy parts of human life. Often, newly-arrived researchers are unable or unwilling to easily see this connection between their personal styles and anxieties on the one hand, and "methods decisions" on the other (Devereux, 1968). But they are clearly connected in some ways. This ideas is captured well by Sechrest and Figueredo (1993) in their discussion of the effect Newtonian physics on later scientific inquiry, including social science. They suggest that the Newtonian image of an inalterable, mechanical universe biased social scientists toward avoiding the messy aspects of humanity.
[This Newtonian metaphor mentally prepared social scientists] for a bold exploration of the icy depths of interplanetary space. Instead, they found themselves completely unprepared for the tropical nightmare of a Darwinian jungle: A steaming green Hell, where everything is alive and keenly aware of you, most things are venomous or poisonous or otherwise dangerous, and nothing waits passively to be acted upon by an external force . . . The Darwinian jungle manipulates and deceives the unwary wanderer into serving myriads of contrary and conflicting ends. The sweltering space suits just had to come off (Sechrest and Figueredo, 1993: 647-48).
DISCOURSE AND LANGUAGE
One of the things we are trying to communicate here is that no research is free of a cultural context. It would be false to contrast traditional social science methods (generally quantitative) with those which seek to fill in the cultural context. All studies deal in material that is inherently comparative. So, quantitative research is best not thought of as research that is done where subjects can be extracted from a cultural context and made to produce data free of that context. Rather, quantitative research is simply where a phenomenon can be rendered into numbers (whether these are complex behaviors measured by observers, or yes/no answers on a questionnaire). Quantitative research is therefore more accurately contrasted with research that works with categories of things (Weisner, 1996:316).
Conversely, qualitative research should not be contrasted with quantitative research, but with particularistic research. The methods dealt with in this guide generally traffic in nominal, or categorical, data. These methods generally go to lengths to discover those categories although numerical data is used more and more to help model how people conceive of categories of things (multidimensional scaling, cluster analysis, cultural consensus analysis, some kinds of content analysis, etc.).
Another issue involves the terms used to refer to the people from whom information is being gathered by researchers. While these terms may be seen as interchangeable, they carry with them different histories and connotations. Throughout this guide we use the terms "subject," "respondent," and "informant" and "participant." The term "subject" enjoys a broadness that logically allows it to be used more often than it generally is in qualitative literature. In practice it is often reserved for experimental interventions for which there is a treatment group and a control group, or for laboratory observations in which the organization of the subjects' thinking--as expressed in his or her words or actions--is not central to the data being collected. "Respondent" is a term heavily used in sociology and psychology emphasizing the role of the research subject as one who responds to survey instruments. While this word implies the importance of the content of the subject's response, the range of responses allowed by survey instruments is usually constrained. Anthropologists and some sociologists prefer "informant" because it focuses more forcefully on the information that subjects both construct and provide to researchers and on the role of the informant as an insider with a perspective valued by the researcher. However, some don't like informant due to its unfortunate association with phrases such as "police informant, FBI informant", and also because the roles if an informant are clearly distinct from the researcher, who is separated, distant from the information provided. "Participant," more so than "informant," emphasizes the view that these individuals participate as equals in the research endeavor.
The information informants provide-usually gathered by open-ended or at least relatively unconstrained observational or conversational methods-is what qualitative research really feeds on. Rather than confirming or measuring categories provided by researchers, this information generally allows investigators build a holistic understanding of how informants see their world and act in it. Finally, in their role as constructors of information and understanding, informants may be considered "participants" as well. Without their (often active) participation in the research, the results of qualitative investigation would be impoverished. All four words--subject, respondent, informant, and participant--are used below partly to continue to illuminate these different styles of research. In multimethod research where styles may be intertwined, one who is an informant one day may be a respondent the next.
CONCEPTS AND TERMS
Researchers who use primarily quantitative methods are accustomed to relying on certain tenets of classical research design; for example, random sampling, standardized instruments with demonstrated validity and reliability, systematic and "non-biased" analysis of data, and interpretation of data based on assumptions of objectivity and generalizability. Here we discuss these concepts from the point of view of qualitative research. Some concepts are not especially problematic and can be resolved by challenging inappropriate assumptions of a dichotomy between quantitative and qualitative methods, others may be resolved by using the right method or procedure, and yet others by considering a different epistemological framework altogether.
While some qualitative researchers might argue that the concept of validity as used in the quantitative research paradigm has no place in qualitative research, there does exist some conceptual counterparts to validity in qualitative/ethnographic methods. However, the types of validity considered in qualitative methods are not isomorphic to their counterparts in the quantitative research paradigm; they are necessarily different because the epistemological framework in which they operate is different. Also, more than one term may be used by qualitative researchers to refer to similar concepts. However, for the researcher trained in the classical, experimental paradigm and unfamiliar with qualitative methods, these parallel concepts help to facilitate the understanding of the principles underlying the method. For example, the concept closest to internal validity in quantitative research is credibility (Lincoln & Guba, 1985). The credibility of the research tool, whether it be observation, interview, or ethnography, is judged by the informants themselves. In other words, it is judged from the point of view of the people closest to the phenomenon being studied. The conceptual counterpart of external validity, the validity of any set of research findings for a particular setting, is transferability or applicability (Lincoln & Guba, 1985). Among the types of validity used in quantitative research, external validity appears to be the least problematic when applied to qualitative methods. That is, in quantitative research the applicability of a certain research design or set of findings needs to be demonstrated empirically, case by case, and so too in qualitative research.
The general concept of reliability in quantitative research is similar to the concepts of dependability or consistency in qualitative research (Lincoln & Guba, 1985). In an ethnography, for example, the question of how dependable the data are is an important one. Some researchers at one end of the epistemological spectrum, however, may reject the notion of dependability or consistency based on the assumption that each phenomenon is rooted in a particular place and time, as viewed by a particular observer who is inextricable from the phenomenon, and therefore cannot be replicated or reproduced.
Quantitative research is predicated upon the assumption of objectivity, that the researcher does not bias the studied phenomenon or the results in any way. The closest principle in qualitative research to objectivity is confirmability (Lincoln & Guba, 1985). Confirmability refers to the quality of the data--whether the data can be confirmed by other observers or interpreters. Again, some researchers using qualitative methods may reject the notion of confirmability altogether based on different epistemological assumptions.
Whereas the notion of generalizability is taken for granted in quantitative research once random sampling is in place, the concept is actually more carefully considered in qualitative research. Instead of assuming that findings can be generalized, the transferability of findings to a particular group or population is evaluated on a case-by-case basis. Thus, generalizability does not depend on sampling strategy but rather on substantive data (Patton, 1980). Without the contingency of generalizability on random sampling, qualitative researchers are free to use a variety of sampling strategies depending on the purposes of the study. Sampling strategies include extreme or deviant cases, typical cases, maximum variation, critical cases, and sample of convenience (Patton, 1980).
These concepts and terms will be further discussed throughout the sections of this guide.