Betting on others to reveal oneself

This thesis tests Top/Flop and Target betting, a new approach to elicit agents’ private signals when the truth is unverifiable. The method elicits an agent’s private signal based on the relative performance of an item against a comparable but unknown item using bets to incentivise subjects. We test this method in the domain of student evaluations of teaching under the implementation of an exogenous default bias. Our results indicate that Top/Flop and Target betting can increase data quality in regard to subject’s response time. Further, upon testing the assumptions of the method, we conclude that they are reasonable to hold in practice. These experimental outcomes underscore the discussion of finding simpler methods in the Bayesian truth-inducing incentive literature. Our results might help in designing more reliable methods of opinion elicitation and encourage further testing of the Top/Flop and Target betting method in large-scale sample populations.