Interviewers' question rewording: not always a bad thing

Although correct question reading is a fundamental assumption of standardized interviewing, in surveys, interviewers will not always read all questions exactly as worded. In this study the deviations in question reading by interviewers were analyzed. In addition, we studied if these deviations were based on conventions in ordinary conversations. Data of the Nebraska Annual Social Indicators Survey 2006 (NASIS) were used for analysis. The results indicate that interviewers tend to shorten lengthy questions, but also add parts to specific questions. In this contribution we will argue that (at least some of) these changes have specific interactional functions: they increased both the cohesion and the coherence within the questionnaire. We will argue that this restructuring of language arises from the fact that an interview is seen as a communicative coherent whole, and not just as a set of individual items. Due to these changes, respondents appeared to be better able to formulate an answer. With the knowledge of this study, the structure within and between questions in surveys can be improved which will enhance response quality