Attributes of Questions and Interviewers as Correlates of Interviewing Performance

Abstract

A basic tenet of survey research is that respondents must answer standardized questions. Consequently, interviewers are trained to ask questions verbatim. Yet evidence from monitoring studies shows that interviewers frequently depart from the prescribed wording (Brenner 1982; Cannell, Lawson, and Hausser 1975; Marquis 1971). A few studies have explored the correlates of this error, such as question features or interviewer characteristics. Cannell and Robison (1971) found open questions were less apt to be misread than closed ones. The same effect was observed by Mathiowetz and Cannell (1980) and Bradburn, Sudman, et al. (1979), but it was very small. By contrast, Oksenberg (1981) reported that open questions were more likely to be misread than closed ones. In their experimental comparison with sensitive items, Bradburn, Sudman, et al. (1979) found interviewers made somewhat fewer reading errors when questions were short than when they were lengthened by nonsubstantive introductions and when they contained standard language as opposed to colloquial terms. In addition, they reported that older interviewers and those with the longest experience made more errors, but these differences were not statistically significant. The present study replicates parts of this earlier work and extends it to other interviewer and question characteristics. We reasoned that the less exposure interviewers had to a question the less likely it would be read verbatim. If this is true, then questions that are part of frequently invoked skip patterns would be misread more often. Two considerations led us to expect that the later a question ap-