questionnaire

A document containing all the questions, closed and open-ended, for a survey . Normally, a separate questionnaire is used for each respondent to a survey, providing enough space for answers to be recorded, and subsequently coded for computer-based analysis of all replies to each question. Questionnaires range from the postcard, with a few questions to be filled in by respondents, to long documents to be filled in by trained interviewers. Good questionnaires require a great deal of care and effort, to ensure that the questions are clear and easy to answer, to exclude leading questions unless by conscious design, to prompt and probe respondents' recollections of events that may not always be very recent, and to shape the interview overall so that it is a pleasant and interesting experience for respondents. Special techniques have been developed for questions on sensitive topics, interviewing on life-cycle events and work histories, questions on attitudes, values, and preferences. Questionnaires must also be structured to ensure that people are correctly filtered into or past particular sections-for example, someone who has not been in employment for many years should not be asked questions about their work, and so on. Questionnaires help to standardize interviews , increasing the consistency of enquiry and response, but they cannot completely eliminate interviewer bias .

There is a large and specialized literature on questionnaire design. Experimental testing has shown that (among other things) question wording, the order in which items are presented, the use of intensifiers, and the number and distribution of response categories can all affect the distribution of answers produced by questionnaires. It is well established, for example, that one elicits a different pattern of responses to a question about class identification by asking people to choose between ‘upper class, middle class, and working class’, on the one hand, and ‘upper class, middle class, or lower class’, on the other. (Fewer people are willing to describe themselves as ‘lower class’ than ‘working class’.) Similarly, studies have shown that people respond differently to questions which use intensifiers (‘do you feel reallyunsafe … ?’, ‘have you been particularly dissatisfied … ?’), as compared to the same questions formulated without intensification.

The distribution of response categories that is offered to respondents may also affect the distribution of responses. Some studies suggest that when informants are offered alternatives in numerical form (‘How many hours each week do you watch television-(a) not at all (b) 1 to 3 hours (c) 4 to 6 hours (d) 7 to 9 hours (e) 10 or more hours?’), they think they are being offered a choice that reflects population norms, and may therefore be influenced to avoid the (apparently extreme) end-categories. Someone who watched 15 hours of television each week might feel guilty about appearing to be an outliner, and so choose the ‘7 to 9 hours’ response category instead. However, if the above scale was continued in like fashion and in such a way as to place ‘13 to 15 hours’ in the middle of the range of response options offered, a more honest answer would be elicited. Conversely however, and no less problematically for the researcher, it is also clear that non-numerical and vague quantifiers (Do you watch television ‘hardly ever’, ‘not much’, ‘regularly’, ‘quite a lot’?) mean different things to different people. Ten hours each week may suggest ‘regular’ watching to one respondent but ‘not much’ to another.

These and many of the other design factors that should be borne in mind when constructing a questionnaire are discussed in, Questions and Answers in Attitude Surveys: Experiments on Question Form, Wording, and Content (1996). See also closed response ; open response.