Measuring respondent uncertainty in discrete choice experiments via utility suppression

Discrete choice experiments (DCEs) are an important methodology in survey research. Although DCEs assume that respondents know their true preferences and can make choices precisely, in reality, respondents may be uncertain. Respondent uncertainty has long been recognized as an issue in the DCE context, but unfortunately, its explicit modelling has received only limited attention. This paper proposes using hierarchical Bayes multinomial probit models to construct respondent uncertainty measures based on the utility difference and a concept called utility suppression, which is quantified by the derivative of the inverse Mills ratio. Two empirical studies with different DCE design formats are conducted to assess and compare the performance. The first includes a no-choice option, and the second is forced-choice formatted with a simpler design. The hold-out validation shows that the utility difference and our proposed measure, which are consistent with the random utility maximization assumption, have better overall performance, particularly in identifying the high-uncertainty respondents. We further show how the information about respondent uncertainty can be utilized in practice.