Closed-Ended Questions

Survey questions come in two varieties: OPENENDED QUESTIONS, in which the respondents provide their own answers, and closed-ended questions, in which specific response categories are provided in the question itself. Although there has been considerable research on the relative merits of the two types of questions, the substantial preponderance of questions that appear in any survey are closed-ended questions.

All survey research involves asking questions for which the responses are categorized to facilitate analysis. In closed-ended questions, the researcher makes prior judgments about what the appropriate categories might be and offers them immediately to the respondent in the question wording. There are a number of potential problems with this format, not the least of which is that it typically describes the “world” in dichotomous terms that sometimes reflect an oversimplification of possibilities. This kind of constraint does not exist with an open-ended question, and for this reason Schuman and Presser (1996) conclude that they often provide more valid data. However, the cost saving of not having to CODE verbatim responses and the quicker access to analysis make close-ended questions the preferred form.

When designing or using close-ended questions, a number of standard suggestions reflect considerable research on the matter. Using forced choices is preferable to asking a respondent to agree or disagree with a single statement. A middle alternative should generally be offered, except when measuring intensity, and an explicit “no opinion” option should be offered as well. Researchers should use multiple questions to assess the same topic, remaining sensitive to the effects of question order. And when in doubt about the possible effects of question wording or response options, research should include split-sample versions in their questionnaires so that comparative analyses can be run.