海角黑料

Public Good Games and Punishment: A Large-Scale Empirical Reassessment

This paper offers a robustness assessment of Herrmann, Thöni, and Gächter (2008) using one unified dataset from ClassEx, an educational experimental platform. We analyze a large set of sessions that implement the canonical public good game with costly peer punishment using instructions and parameters that mirror the original design (endowment, MPCR, punishment technology, round structure). The study asks three simple questions: (i) does allowing punishment raise average contributions relative to no-punishment rounds in this single-source dataset? (ii) how frequent and how severe is “antisocial” punishment (sanctions of equal or higher contributors) within groups? and (iii) to what extent do punishment–retaliation dynamics erode efficiency? We compute the same core summary measures as the original study and compare distributions and effect sizes within the ClassEx data. The contribution is to provide a clean, single-dataset check of whether the qualitative patterns emphasized by Herrmann et al.—cooperation gains tempered by antisocial punishment—emerge when the game is implemented at scale in a standardized classroom environment.