Stochastic Choice and Preferences for Randomization
We conduct an experiment to investigate the origin of stochastic choice and to differentiate between the three main classes of models that account for it: Random Expected Utility; Mistakes; and Deliberate Randomization. Subjects face the same questions multiple times in two ways: 1) following the literature, with repetitions distant from each other; 2) in a novel treatment, with repetitions in a row, telling subjects that questions will be repeated. A large majority of subjects exhibited stochastic choice in both cases, and stochasticity is strongly correlated in the two cases. Our results support the class of models of Deliberate Randomization.
A previous version of this manuscript circulated under the title "Stochastic Choice and Hedging." We thank Mark Dean, David Dillenberger, Federico Echenique, Kfir Eliaz, Guillaume Frechette, Yoram Halevy, Mark Machina, Efe Ok, Gil Riella, Kota Saito, Chris Shannon, Leeat Yariv, and participants at various seminars and conferences for helpful comments and suggestions. Ortoleva gratefully acknowledges the financial support of NSF grant SES-1156091 and of that of the Caltech's Gordon and Betty Moore Foundation through Grant GBMF1158.
Accepted Version - RandomChoice.pdf