People belong to many different groups, and few belong to the same network of groups. Moreover, people routinely reduce their involvement in dysfunctional groups while increasing involvement in those they find more attractive. The net effect can be an increase in overall cooperation and the partial isolation of free-riders, even if free-riders are never punished, excluded, or recognized. We formalize and test this conjecture with an agent-based social simulation and a multi-good extension of the standard repeated public goods game. Our initial results from three treatments suggest that the multi-group setting indeed raises overall cooperation and dampens the impact of freeriders. We extend our understanding of this setting by imposing greater heterogeneity between groups through interweaving automated bot players amongst human subjects; whereby initial sessions of this amplify the aforementioned effects.
Berman, A. S., Iannaccone, L. R., & Modak, M. (2023). How personalized networks can limit free riding: A multi-group version of the public goods game. ESI Working Paper 23-12. https://digitalcommons.chapman.edu/esi_working_papers/392/