Abstract The use of real monetary incentives has long defined experimental economics, setting it apart from disciplines like psychology, where hypothetical choices are common. While full-payment and hypothetical designs represent two clear extremes, random-payment incentive schemes – where one of several decisions is randomly selected for payment – have become a prominent approach in economics. Yet the behavioral validity of random-payment schemes remains underexamined: Do they more closely resemble fully incentivized tasks or hypothetical ones? We compare full-payment (participants paid for every decision), no-payment (participants not paid for any decision), and random-payment (participants paid for one randomly selected decision) incentive schemes, using five standard economic games to measure social preferences (Dictator Game, Ultimatum Game, Trust Game, Public Goods Game, and Prisoner’s Dilemma). Results from Experiment 1 (n = 1,501), with £1 stakes, indicate no significant differences between incentive schemes in any of the games included or a composite measure of prosocial behavior. In Experiment 2 (n = 750), with £10 stakes, results were largely in line with Experiment 1, suggesting no consistent behavioral impact of incentive schemes at increased stakes. The one notable exception was responder behavior in the Ultimatum Game, where participants in the full-payment condition were less likely to reject offers compared to those in the no-payment condition, with random-payment falling in between. Our results challenge the rigid disciplinary norm that real stakes are essential for valid measurement and invite a more nuanced consideration of how and when different incentive schemes are necessary or appropriate in behavioral research.
Koppel et al. (Wed,) studied this question.