Abstract:
Several studies provide evidence that heat is positively associated with criminal activity. However, the empirical literature does not provide conclusive evidence about the effect of high temperature on homicides. I examine 156 estimates from 20 studies on the relationship between temperature and homicide rates. In particular, in this meta-analysis I study publication bias using linear and nonlinear techniques together with Bayesian model averaging to explain the heterogeneity in the estimates. After correcting estimates from the publication bias, I find no significant effect of temperature on homicide rates. Moreover, monthly data produce larger estimates. Conversely, studies using data from Asia or the OLS estimation method lead to smaller estimates.