This article reports the effect that the ranges presented in answer categories for survey questions can have on respondent answers. Response categories were manipulated in a split-ballot survey conducted in both telephone and mail modes. These categories, presented in the separate ballots, overlapped in one category; the other categories were unique to each ballot. The experiment was conducted on four questions: two frequent and mundane and two rare and salient. It was found that the response categories significantly affected the response for frequent and mundane questions. One question demonstrated a significant difference in response between the mail and telephone modes. For this question, a response scale with a limited number of socially desirable alternatives resulted in a social desirability effect in the telephone mode. Alternatively, the telephone mode demonstrated an extremeness effect when the response scale comprised a greater number of socially desirable alternatives.
Copyright 2017 Elsevier B.V., All rights reserved.