Research

Discrete choice experiments are widely used to elicit consumer preferences for products or services. Usually, the characteristics of the goods in question are referred to as attributes and the possible values that these characteristics may take are referred to as levels. Yet, how the assignment of attributes and levels may affect respondent understanding and researchers’ preference elicitation remains underexplored. This study addresses this gap by comparing two designs that alternate the presentation of the same information between attributes and levels. We implement a design where sustainability measures and outcomes are defined as either attributes or levels. We find that the design using sustainability measures as attributes is modestly easier to read than the design using sustainability outcomes as attributes. Results on preference consistency and stability are broadly similar across designs, with a few meaningful differences concentrated in a small subset of sustainability measures. These findings suggest that attribute–level assignment can affect reading burden and, in some cases, elicited preferences, even when the substantive content of a design remains constant.

Hypothetical bias (HB) is a common concern in stated preference studies, where respondents often overstate their willingness to pay for attributes of interest. Past studies suggest that participation incentives may influence respondents’ behavior, but their effects on HB have not been systematically tested. We conduct two split-sample experiments in two online discrete choice experiment settings: an alligator-hide product study and a donation study. In each study, respondents are randomly assigned to one of three treatments: a hypothetical treatment without an incentive, a hypothetical treatment with a participation incentive, and a real-payment treatment in which respondents purchase or donate using the incentive amount provided while keeping any remaining balance. HB is measured by comparing willingness-to-pay estimates from the hypothetical treatments with those from the real-payment treatment. Our results show that participation incentives lead to behavioral changes, including higher opt-in rates and stronger perceptions of consequentiality. However, although HB persists across both contexts, offering a participation incentive in a hypothetical treatment does not produce statistically significant differences in HB.