Citation

Loftin C, McDowall D. Homicide Stud. 2003; 7(4): 353-367.

Copyright

(Copyright © 2003, SAGE Publishing)

DOI

10.1177/1088767903256553

PMID

unavailable

Abstract

In 1998, Cohen described a study showing that in the United States, regionally distributed culture controls the direction of the relationship between social stability and homicide rates. In the South and West, where, according to Cohen, there is a culture of honor, strong community and family bonds increase honor-related violence. Conversely, in the North, where the culture of honor is rare, strong social bonds restrain violence of all kinds. This article describes a reanalysis using Cohen's data sources that finds little association between measures of social stability and homicide rates. Cohen's findings are due to errors in the measurement of homicide that lead to an excessive number of zero values and a few extremely high values that heavily influence the slope estimates. Alternative estimates using larger counties or a logistic regression model that eliminates the impact of the extreme values lead to estimates that are inconsistent with the theory.