The Source and Significance of Confusion in Public Goods Experiments

Economists use public goods experiments to develop and test theories of individual preferences and institutional design. Previous work demonstrates many participants in public goods experiments contribute to the public good out of confusion. We design experiments to provide insights on the consequences and causes of confusion. We establish that confusion amounts to more than statistical noise and does not dissipate with repetition (i.e. learning). Confused subjects use experimental parameters and the behavior of other players as cues, which confounds treatment effects and traditional strategies to identify other-regarding preferences through exogenous parameter changes and the modeling of reactions to other subjects' decisions. We argue that confusion stems from an inaccurate understanding of game incentives ("failure of game form recognition"), which is a consequence of the framing and inadequate payoff information in standard instructions. Modified instructions can substantially reduce confusion, and, in turn, change the distribution of contributions to the public good.