Validated RIS: A within subject test of incentive compatibility in the random (lottery) incentive scheme

We propose the Validated Random Incentive Scheme (VRIS)—an enhancement of the standard Random Incentive Scheme (RIS)—that enables internal validation of RIS’s incentive compatibility (IC) without altering an experiment’s core design. The key innovation is that researchers can observe each subject under both RIS and single-choice incentives, enabling direct withinperson comparisons that form the basis for testing IC. We derive a set of testable predictions from assumptions about the structure of random noise in decision making. In an online experiment involving three risky decisions and a perception task, we use the VRIS design to compare behavior under the RIS with that under single-choice incentives. We find that the RIS distorts the marginal distribution of choices in two of the four decisions and the joint distributions in five of six decision-pairs. In contrast, our novel VRIS directly measures all four marginal distributions under single-choice incentives and enables recovery of undistorted pairwise distributions in all but one case.