Importance of categories of crime for predicting future violent crime among handgun purchasers in California

Abstract

Background Prohibiting the purchase and possession of firearms by those at risk of violence is an established approach to preventing firearm violence. Prior studies of legal purchasers have focused on convictions for specific crimes, such as violent misdemeanors and driving under the influence (DUI). We broaden that line of inquiry by investigating and comparing the associations between prior arrests for most categories of crime and subsequent arrest for violent offenses among legal handgun purchasers in California. Methods In this longitudinal cohort study of 79,678 legal handgun purchasers in California in 2001, we group arrest charges prior to their first purchases in 2001 according to categories defined by the Uniform Crime Report (UCR) Handbook. We use a gradient boosting machine to identify categories of offenses that are most important for predicting arrest for violent crime following firearm purchase. For each category identified, we then estimate the difference in risk of subsequent arrest for a violent offense using survival regression models. Results We identified eight crime categories with high predictive importance: simple assaults, aggravated assaults, vehicle violations, weapon, other crimes, theft, drug abuse, and DUI. Compared to purchasers with no prior arrests, those with a prior arrest for any one of the eight important categories and no other categories were found to be at increased risk of arrest for a Crime Index-listed violent crime (murder, rape, robbery, aggravated assault), with the greatest estimated risk corresponding to the simple assault UCR category (adjusted hazard ratio 4.0; 95% CI 2.8–5.9). Simple assault was also associated with the greatest risk for subsequent arrest for firearm violence (adjusted hazard ratio 4.6; 95% CI 2.4–9.0) and any violent offense (adjusted hazard ratio 3.7; 95% CI 2.7–5.0). Conclusion The findings of this study suggest that prior arrests for a broad array of crimes, both violent and non-violent, are associated with risk of subsequent violent crimes, including Crime Index-listed violent crimes and firearm violence, among legal purchasers of firearms. Current policies aimed at restricting access to firearms for individuals at increased risk of violence should be re-examined considering these findings.