Overcoming Incentive Constraints? The (In-)effectiveness of Social Interaction

We experimentally study behavior in a simple voting game where players have private information about their preferences. With random matching, subjects overwhelmingly follow the dominant strategy to exaggerate their preferences. Applying the linking mechanism suggested by Jackson and Sonnenschein (2005) captures nearly all achievable efficiency gains. Repeated interaction leads to significant gains in truthful representation and efficiency only if players can choose their partners.