Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
Multiple group memberships are the rule rather than the exception. Within a linear public good game, we experimentally investigate two possible factors that impact the decision to cooperate in a smaller, local or a larger, global group: diverging marginal per capita returns, resulting in different social returns, and social feedback information. If social returns are equal across groups, subjects prefer to contribute to the local group that offers social information on individual contributions. An increase of the social return in the global group initially attracts more contributions, but this tendency quickly unravels in favor of cooperation in the local group. Cooperation in the global public good can only be sustained if contributions of global group members can be observed. We thus identify social feedback information as a key factor for institutional design to foster cooperation.