These data were gathered to study six-person Prisoner’s Dilemmas (PDs) in which subjects endogenously decide whether to implement a peer punishment institution in their group and whether the punishment institution, if implemented, implies more or less severe punishments. We consider PDs with perfect information on other subjects’ previous behavior, and PDs in which subjects observe each other’s behavior with noise.
In the experiment, subjects participated in series of six-person PD games with endogenous punishment institutions. In half of the sessions, the PD with noise was employed, while in the other half, subjects participated in the PD without noise. Subjects were informed on whether and how noise was implemented. Payoffs were presented as points that were translated into monetary earnings at the end of a session. Throughout each session, subjects were randomly re-matched in different groups of six after every interaction. Subjects interacted for 40 periods in the PD.
The experiment was programmed using z-Tree (Fischbacher 2007) and conducted at the ELSE laboratory at Utrecht University. Subjects were recruited using the online recruiting system ORSEE (Greiner 2004). 156 subjects participated in total: 78 in the noise condition and 78 in the condition without noise (41% male, 85% students, 30% economics students). Subjects earned €12.50 on average (minimum of €7.50, maximum of €15.50).
The data show that without noise, the majority of groups choose a punishment institution with severe punishments, while with noise most groups do not implement a punishment institution. Both with and without noise, cooperation and earnings increase when a punishment institution is implemented, especially with opportunities for severe punishments. However, subjects in the noise condition perceive lower earnings under severe punishments than under the other options. Also, with noise, observing that cooperators are punished, discourages subjects from implementing a punishment institution in subsequent interactions.
