Experimental Insights on Anti-Social Behavior: Two Meta-Analyses

We conduct two meta-analyses on antisocial behavior in experimental settings in which such behavior is not rationally motivated by pecuniary incentives. We investigate the impact of its potential determinants. The first meta-analysis employs aggregate data across experimental settings from 93 published and unpublished studies (22,200 participants), using laboratory, field and online experiments carried out since 2000. We find that antisocial payoff destruction varies depending on the experimental setting, being highest in vendetta games and possibly lowest in social dilemma games. There is significant heterogeneity across the studies, including within game classes, making inference difficult. The second meta-analysis includes only money burning experiments (for which we have the largest number of observations: 46 studies and around 15,000 participants). It finds evidence of negative discrimination against outsiders, of exogenously disadvantaged subjects destroying more often, and of more antisocial behavior in one-shot interactions. The strategy method biases antisocial behavior upwards. We do not generally find publication bias, either in aggregate or in relation to money burning experiments. Field studies display more antisocial behavior than laboratory experiments. Taken together, our results point to the value of more laboratory experiments that systematically build on paradigmatic experimental designs to enable comparability and the identification of key economic drivers of antisocial behavior.