Classifying heterogeneous cooperation in social dilemmas: experimental evidence and simulation insights
摘要
In this paper, we introduce a revealed strategy method to capture heterogeneous behavioral patterns in repeated public goods experiments. This method identifies an individual’s strategic type—a stable decision-making tendency in interactive settings—by analyzing behavioral profiles. We conduct a repeated public goods game to collect participants’ behavioral data and then apply both k-medians and hierarchical clustering algorithms to classify them into five distinctive types: free riders, strong cooperators, above-average conditional cooperators, below-average conditional cooperators, and hump-shaped players. These classifications are then used to construct and calibrate an agent-based simulation model. Our simulation results suggest that policies targeting how conditional cooperators respond to perceived social norms may effectively promote cooperation. The revealed strategy method offers a useful tool for experimenters to better characterize group composition and understand strategic interactions, while also providing empirical validation for agent-based models and theoretical studies.