A new statistical approach to the problem of homicides in El Salvador

Abstract

Despite the high attention paid to the homicide rate in El Salvador, the empirical correlations of this variable with other explanatory variables have not been quantitatively substantiated. This study contributes to address this knowledge gap, calculating quantile regressions of the homicide rate per 100,000 inhabitants for each municipality in El Salvador in the 3rd, 50th and 98th percentile respectively, with population data for 2007. The resulting models show that the homicide rate is significantly correlated with the following variables: positively with the number of gangs per municipality, in the 3th and 50th percentiles; positively with the index of urbanity and negatively with access to drinking water, in the 50th percentile; as well as positively with the number of gang members per 100,000 inhabitants, in the 98th percentile. These results can be used to guide policy, but they only explain part of the problem. There is an urgent need for better statistics to correct some limitations in future studies.

Keywords

• tasa de homicidios
• el salvador
• regresión cuantílica
• regresión simbólica
• programación genética