Let the Others Do the Job: Comparing Public Good Contribution Behavior in the Lab and in the Field

Under the assumption of payoff-maximization, standard game theory predicts that groups playing a public good game (PGG) do not manage to coordinate on the social optimum of full cooperation and play the pareto-inferior Nash-equilibrium of zero contributions instead. Contrary to this, numerous lab experimental studies have shown that subjects contribute on average positive amounts to a public good, which might be due to pro-social preferences. In our study we compare individuals’ public good contribution behavior in a lab and field setting. University students working on an in-class group assignment, which is graded on the group level, are essentially playing a PGG. While in this setting investing a maximum amount of time and effort is the socially optimal strategy, free-riding on the efforts of the others may be individually more beneficial. We test if students’ behavior in a standard PGG is correlated with their contribution behavior in the group assignment which is part of a university course. The PGG is incorporated into an extensive online survey which is also used to measure the students’ performance in the group assignment and to collect data on other relevant factors. Among others, we control for individual motivation and possession of course-relevant skills, the social cohesion of the group, and the use of sanctions towards underperforming group members. Preliminary results suggest a positive correlation of the contribution behavior in the online linear PGG and the behavior in the group assignment, suggesting that individual characteristics, such as preferences, matter for the behavior in both settings. Students who are more cooperative in the online PGG also contribute a significantly higher quality of work to the group assignment. On the other hand, we do not find a correlation between conditional PGG contributions, measured with the strategy method, and field behavior. This finding challenges the generalizability of this method.