Variability in interview discourse: a problem or a source?

In the course of a research interview, and especially in an open interview, an interviewee may construct very different, and often irreconcilable, positions and lines of argument regarding a single topic. Such variety and contradiction can be explained by appealing to the notion that discourse is essentially incoherent. This explanation isn't a very attractive one, however. After all, abandoning the assumption of coherence implies abandoning the assumption of any rationality in social interaction, and ultimately giving up the goal of trying to understand discourse. The crucial question is therefore: How to approach and analyze the discursive variety and contradiction that occurs within an interview?

Variability as a problem

In mainstream research methodology, variability is approached as a problem. Differences, and especially contradictions, in interview discourse are viewed as indicators of some kind of “measurement error” due to biasing factors, such as the possible inadequacies of the interviewer's behavior. After “purifying” interview discourse as much as possible from such errors (for example, by correcting for the supposed effects of social desirability), several procedures are recommended to construct an estimation of the “real” opinion or attitude of the interviewee. In the attempt to measure attitudes (or, more precisely, to construct estimations of supposed attitudes), these procedures transform different or contradictory interviewee statements into a consistent position that can be recorded on an attitude scale.