Question-answer Sequences in the Survey Interview |
| Prof. Dr. Wil Dijkstra |
| Free University, Amsterdam & NIAS, The Netherlands |
| In many cases, answering a survey question posed by an
interviewer proceeds straightforward: the interviewer poses the question,
the respondent gives an answer that can unambiguously be scored by the interviewer,
and finally, the interviewer often confirms the answer (e.g. by repeating
it, or saying something like "OK"). Such a straightforward Question-Answer
sequence is called a paradigmatic sequence. Often however, the interaction deviates from a paradigmatic Question-Answer sequence. Such deviations may be non-problematic (e.g. the respondent may just repeat the question), or problematic. A further distinction is made between problematic deviations by the interviewer (e.g. altering the meaning of a question) and by the respondent (e.g. giving an answer that is not directly scorable by the interviewer). Research shows that most problematic deviations are initiated by the respondent, not by the interviewer. By far the most occurring type of such problematic respondent deviations are so-called mismatch answers: answers that don't fit one of the prescribed response alternatives, or cannot be scored unambiguously. These deviations in turn appear to be the primary cause of problematic interviewer deviations, e.g. suggestive probing. It could also be shown that in general, such question-answer sequences lead to less valid answers than non-problematic sequences. Three different causes of such mismatch answers are suggested. Firstly, a cognitive mismatch answer points to cognitive problems encountered by the respondent in answering the question. For example, in response to the question "How many days a week do you watch television on average", the respondent may have difficulty to retrieve an exact number, and answers something like "most days". Secondly, in many cases, there is no reason to expect cognitive problems in answering a question, but nevertheless, respondents may give mismatch answers. For example, in response to "What is your age?", respondents may answer with a conversational mismatch answer like "I'm retired", or "I'm near to eighty". Thirdly, respondents may have difficulty with the task itself, e.g. because the question is ambiguous, or their situation does not fit one of the response alternatives (e.g. "Do you own a car?" [yes/no] may be answered with "I own a car together with my neighbours". Relationships between the occurrence and kind (cognitive, conversational or task) of mismatch answers on the one hand and characteristics of the question and response alternatives on the other hand will be discussed. Finally, ways to improve the design and formulation of questions will be suggested. |