Design, Evaluation, and Analysis of Questionnaires for Survey Research, 2nd Edition

5

RESPONSE ALTERNATIVES

So far, we have been discussing requests for answers. As was indicated in Chapter 3, the requests can have many different forms, which in turn can create the same response alternatives for the respondent. However, the fact that the same response possibilities are present does not mean that the requests for an answer measure the same thing. Along the same line, it is not immediately clear whether requests for an answer that are identical but differ in the set of possible responses measure different variables. This is an empirical question that has to be answered for different measures. Saris (1981) showed that at least some sets of response scales, although different, will give responses that are identical, except for a linear transformation suggesting that roughly speaking, these measures are indeed identical.

Another issue studied by many people is whether it makes sense to present the respondents with more than only a few categories. Most textbooks suggest, in reference to Miller (1956), that people cannot use more than approximately seven categories. Cox (1980) has argued that Miller’s rule does not apply at all to this problem. He suggests that more information can be obtained if more categories are used. This opinion is shared by a few more researchers (Saris et al. 1977; Andrews 1984; Költringer 1995; Alwin 1997).

Finally, there are people, Krosnick and Fabrigar (forthcoming), who suggest that it would be advisable for certain problems, or in general ...