A Procedure for Revising Data-Based Priors in a Group

Abstract

A group of individuals is interested in predicting the outcome of a current problem. Each individual has access to private data that are used to form a prior probability over possible outcomes. While individuals may be reluctant or unable to disclose their private data, they are willing to publish their priors. We characterize a procedure for revising the individuals’ prior probabilities based on the published priors of others and discuss its implications regarding the formation of a common prior in the group, both in the short and long run. This paper was accepted by Manel Baucells, behavioral economics and decision analysis. Funding: G. Gayer gratefully acknowledges support from ISF [Grant 1443/20].