Answering attitudinal questions: Modeling the response process underlying contrastive questions

Abstract

We analyse reaction time distributions and responses for attitudinal survey questions, which were part of a self‐administered questionnaire about medical and ethical issues. Two contrastive versions of each question were asked, whether an issue should be forbidden or whether the government should allow it. Logically, the answers to these contrastive questions should oppose, but numerous investigations have shown that they result in the so‐called forbid/allow asymmetry: respondents tend to rather say ‘no’ to both questions. We present a mathematical model, based on point process theory, which formalises attitude representations in memory and different stages in the response process. The data of the allow questions are used for parameter estimation, while the forbid data are used to test the predictive power of different model versions. The result is a model that describes the cognitive processes underlying the asymmetry. It indicates that the forbid/allow asymmetry is caused by the use of an increased response threshold in forbid answers, and that the asymmetry size varies due to both respondent characteristics and the issue at hand. This model is capable of simultaneously predicting the asymmetry in the reaction time distributions and in the response scores for the answering categories.