ABSTRACT
RATIONALE AND OBJECTIVE: In the United States, gun violence claims thousands of lives each year and is a pressing public health issue. To gain a better understanding of this phenomenon, this study spatially analyzed county- and state-level predictors of yearly gun violence rates and gun-related casualty rates. METHODS: This study modeled hypothesized predictors of gun violence incidence and casualties across four years. Data sources included the Gun Violence Archive (gun violence data in the United States for 2014-2017), the U.S. Census Bureau (socioeconomic, demographic, geologic features), ICPSR (crime reports), the U.S. Geologic Survey (elevation data), and U.S. gun laws and ownership. Random forest analyses identified relevant additional interaction terms to include. RESULTS: The extent to which counties are urban was the most robust predictor of both gun violence incident and casualty rates. Similarly, places characterized by greater income disparity were also more likely to experience higher gun violence rates, especially when high income was paired with high poverty. CONCLUSIONS: Community- and state-level features are markedly associated with gun violence. Gun violence is higher in counties with both high median incomes and higher levels of poverty; poverty did not seem related to gun violence rates in counties with relatively low median incomes. Some of these findings may well be due to racial segregation and concentrated disadvantage, due to institutional racism, police-community relations, and related factors.