Skip to main content

Rob Ranyard - Dimension-based models, intransitive preferences and decision processes - ONLINE UNTIL 17th March

Date

Rob Ranyard has given a talk at the London Judgement and Decision Making Seminar series on Wednesday, the recording is available until next Wednesday, here.

Abstract

Tversky (1969) presented the first evidence of systematic and predictable intransitive preferences. To explain the violations of weak stochastic transitivity (WST) that he observed, he constructed an extended additive difference model. However, he did not investigate the extent to which the model was a good fit to his data. This question remained unanswered until we recently tested the goodness of fit of a simplified additive difference (SAD) model which predicts transitive or intransitive preferences depending on its parameter values. We reanalyzed Tversky’s lottery study and six replications by estimating the SAD model’s maximum likelihood parameters for each individual choice data set. The model had a very good fit to most individual choice data sets analyzed, with many predictably violating WST. Work in progress is investigating the extent to which the SAD model predicts satisfaction or violation of the triangle inequalities condition, a testable consequence of transitive mixture models. We are also comparing the SAD model with a stochastic lexicographic semiorder heuristic with respect to predictions of intransitive preferences, and we are revisiting verbal reports and neural value responses that were elicited in previous replications. Our preliminary analysis suggests that for most of the participants across the seven studies whose lottery choices were stochastically intransitive, a compensatory additive difference strategy offers the more satisfactory account.