Interaction characteristics of the difficult-to-interview respondent.

In a recent book Thinking About Questions (Sudman, Bradburn and Schwarz, 1996), the authors consider answering a survey question as a rather complex task. Interpreting the question to understand its meaning, retrieving information, generating an opinion or a representation of the relevant behavior, formatting a response, and editing the response are the main components of the task. The performance of each task involves a cognitive and/or communicative process. As a result, answering a survey question is a cognitive and communicative process. When a researcher decides to collect data by means of survey interviews, he or she makes the assumption that all the respondents have sufficient cognitive and communicative skills to execute the respondent's role adequately.

However, there is empirical evidence that some respondents have problems accomplishing their task as respondent and that they are difficult to interview (Groves, 1989). Krosnick (1991) uses the concept of `satisficing' to describe the fact that some respondents are not motivated to expend the mental effort necessary to generate optimal answers. Satisficing may lead some respondents to employ inadequate response strategies (i.e. selecting the first response alternative, acquiescence, and saying `don't know' instead of reporting an opinion). These respondents are difficult to interview. To describe and to explain the profile of these difficult-to-interview respondents, age and education are used as proxy indicators of motivation, cognitive ability, and communicative skills (Alwin and Krosnick, 1991; Loosveldt, 1995). According to this explanation, it is more difficult to interview older and less educated respondents because they do not have the same or enough cognitive and communicative skills.

In this article there is an attempt to validate this explanation. To do that, more direct information about the respondent's cognitive and communicative skills is needed. One method for determining cognitive processes in survey interviews is behavioral coding of the respondent-interviewer interaction (Sudman et al., 1996). From this kind of interaction analysis some objective aspects of respondent behavior as indicators of the respondent's skills are used. The relationship between some interaction characteristics and the two background characteristics (age and education) used to describe the difficult-to-interview respondent will be analyzed.

RESEARCH PLAN AND RESEARCH QUESTIONS

The starting point is the construction of a typology of the respondent's interaction characteristics. It is hypothesized that it is possible to distinguish several types of respondent on the basis of some objective aspects of respondent behavior and that these types can be characterized in terms of cognitive and communicative skills.

To validate this typology, the interviewer's evaluation of some aspects of respondent behavior is used. We assume that objective respondent behavior influences the interviewer's evaluation. As a consequence it is expected that the interviewer's evaluation will be correlated with the typology.

The main research question concern the relationship between the typology and two background characteristics of the respondents (i.e. age and education). It is predicted …