Research has shown that altruistic punishment is an effective mechanism for norm enforcement in social dilemmas.From various experimental studies, treatment effects are qualitatively consistent: altruistic punishment significantly promotes cooperation, generosity and reciprocity; yet they are qualitatively different under different experimental settings, and a single study can only focus on one or two impacting factors.These differences in experimental settings further help us understand the factors that motivate altruistic punishment.And a systematic review, namely a meta-analysis is needed to deal with the difficulty of combining empirical results.In this meta-analysis, we collected 97 journal articles published in over 20high-rank economic journals that investigate altruistic punishment behaviors under public goods game.A set of rules were used to search for and screen studies,and weights based on objective mathematical criteria are assigned to each study for the regression analysis.From the regression results, we found that experimental settings and designs that directly and indirectly determine peoples income are key factors that affect individuals altruistic punishment decisions, that is to say, the key mechanism of impacting altruistic punishment behavior is through income.Culture and religion also contribute to the explaination of heterogeneity in altruistic punishment behavior.