Sources of Variation in Ethnographic Interview Data

Food Avoidances in the Ituri Forest, Zaire

Bibliographic Data

ID7768504
AuthorsRobert Aunger (0000-0001-8136-9376, corresponding author)
Year1994
Volume33
Issue1
Pages65
Publication date1994-01-01
Peer ReviewedYes
Open AccessYes
TypeARTICLE
VenueEthnology (JOURNAL)
Journal identifiersISSN: 0014-1828 • E-ISSN: 2160-3510
PublisherJSTOR (PUBLISHER)
DOI10.2307/3773975
OpenAlexW143285492
LanguageEN
Citations received14

Traditionally, ethnographic researchers have relied heavily on the interview as a method of collecting data (Bernard 1988:203). This prevalence of interviewing is not just characteristic of anthropology: Certairnly, the survey-interview is the best-known method of data-collection in the social sciences (Dijkstra and van der Zouwen 1982:2). As a result, there is a vast literature on how to conduct interviews effectively (e.g., Bradburn et al. 1979; Dillman 1978; Foddy 1993; Rossi et al. 1983; Sudman and Bradburn 1982). However, regardless of the skill with which ethnographers conduct interviews, they often find that when they ask more than one information the same question, different answers are elicited, even when the question bears on cultural beliefs or values. How is this variability in response to the same question to be interpreted? Does it mean that different informants have different beliefs concerning the topic of the question? Should it be interpreted as spurious deviation from a culturally normative response arising from situational vagaries (e.g., the weather on the day the question was asked)? Alternatively, have some informants been unable to learn the appropriate response for some reason? Or does this variability result from a misunderstanding of the question and/or the anthropologist's lack of comprehension of what the informant responded? Similar interpretational difficulties arise when a single informant responds differently to the same question on separate occasions. Ethnographers have not had available the theoretical or empirical tools needed to determine the relative importance of different sources of variability in primary ethnographic data. Boster and Weller (1990:172) identify this as a central problem for ethnographic research: we believe there are still no clear criteria for identifying variation [in responses] as either cognitive, contextual, or simply noise. As a result, although many anthropological studies have tacitly recognized the existence of variability in the beliefs and behavior of different individuals from the same cultural group, most ethnographers have chosen to ignore that variability when conducting their analyses or presenting their results (Pelto and Pelto 1975:2). Instead, ethnographies have traditionally been constructed on the assumption that group-level processes dominate, with individual-level variability being ephemeral. However, when significant intracultural variability in belief does exist, the use of categorical statements to characterize cultural groups as a whole can seriously misrepresent the ethnographic reality being described. This misrepresentation is even more likely to occur when ethnographers follow the traditional practice of asking only a relatively few (i.e., key) informants about any particular topic. If informants do not respond consistently to questions, then the certainty of an particular response is low. Building an interpretation of some aspect of culture based on a series of such weak links can severely reduce the likelihood of arriving at a valid conclusion. This problem makes clear the importance of determining the reliability of responses to questions under field conditions. This article will show that the various influences on what informants report can be identified and their influence on those reports assessed. By using what I call the Reflexive Analytical approach to ethnographic data collection and analysis, the ethnographer can simultaneously solve the two related but distinct problems just discussed: namely, the problems of determining whether there is substantive intracultural variability in belief, and whether particular responses are reliable.(2) First, by partitioning variability in interview responses to different influences using a statistical model, the method determines whether all the observed variability is purely methodological, or represents, at least in part, meaningful differences in belief between individuals in a cultural group

Ethnography · Geography · Sociology · Variation (astronomy) · Anthropology · Cross-Cultural and Social Analysis · Qualitative Research Methods and Ethics

Tools

Open DOISci-HubOpen Access