Testing transitivity of preference in individuals.

Abstract

This study presents a new experiment testing transitivity of preferences in individuals using the stimulus design of Butler and Pogrebna (2018). Each individual responded to each choice problem 60 times, replicated twice in each of 30 sessions. The individual true and error model was used to estimate the incidence of transitive and intransitive preference patterns and error rates for the choice problems for each person. Although the data of most participants were consistent with transitivity, a few participants showed convincing evidence of intransitive preferences patterns at least part of the time, and several participants showed clear evidence of changing true preferences over time. This study also tested and found violations of the assumption that choice responses are independently and identically distributed over repetitions, an assumption used previously in certain random utility models and statistical analyses. Violations of iid are compatible with Markov True and Error Models in which parameters drift gradually over time.