Model-Rich Approaches to Eliciting Possibly Weak or Incomplete Preferences : Evidence from a Multi-Valued Choice Experiment

This paper contributes to the elicitation of a decision maker's strict preferences and their possible indifference or incomparability/indecisiveness. Every subject in both treatments of an incentivized lab experiment could choose multiple alternatives from each of the 50 distinct menus of popular gift-card pairs that they saw. Subjects in the non-forced-choice treatment could, in addition, avoid/delay making an active choice at those menus. Applying a non-parametric optimization method on data collected from 273 subjects, we find that nearly 60% of them are well-approximated by an indifference-permitting model of complete- or incomplete-preference maximization. Most recovered preferences are unique, have a non-trivial indifference part and, where relevant, a distinct indecisiveness part. The two kinds of distinctions between indifference and indecisiveness uncovered by this method are theory-guided and documented empirically for the first time. These findings suggest that accounting for possible indifferences and/or incomparabilities in the data-collection process and analysis can be useful in eliciting transitive weak preferences. Two aspects of the experimental design, finally, allow for interpreting an additional 10% of subjects as revealing a systematic preference for randomization or satisficing