Poverty Matters: A Reassessment of the Inequality–Homicide Relationship in Cross-National Studies
Dozens of cross-national studies of homicide have been published. Virtually all have reported an association between inequality and homicide, leading scholars to draw strong conclusions about this relationship. Unfortunately, each of these studies failed to control for poverty, even though poverty is the most consistent predictor of area homicide rates in the US empirical literature and a main confounder of the inequality–homicide association. The cross-national findings are also incongruent with US studies, which have yielded inconsistent results for the inequality–homicide association. In the present study, I replicated two prior studies in which a significant inequality–homicide association was found. After the original results were replicated, models that included a measure of poverty were estimated to see whether its inclusion had an impact on the inequality–homicide association. When effects for poverty and inequality were estimated in the same model, there was a positive and significant poverty–homicide association, while the inequality–homicide association disappeared in two of three models. These findings were consistent across different samples, data years, measures of inequality, dependent variables (overall and sex-specific homicide rates) and estimation procedures. The new results are congruent with what we know about poverty, inequality and homicide from the US empirical literature and suggest that the strong conclusions drawn about the inequality–homicide association may need to be reassessed, as the association may be a spurious result of model misspecification.