LAUSR

Predictive policing is a tool used increasingly by police departments that may exacerbate entrenched racial/ethnic disparities in the Prison Industrial Complex (PIC). Using a Critical Race Theory framework, we analyzed arrest data from a predictive policing program, the Strategic Subject List (SSL), and questioned how the SSL risk score (i.e., calculated risk for gun violence perpetration or victimization) predicts the arrested individual's race/ethnicity while accounting for local spatial conditions, including poverty and racial composition. Using multinomial logistic regression with community area fixed effects, results indicate that the risk score predicts the race/ethnicity of the arrested person while accounting for spatial context. As such, despite claims of scientific objectivity, we provide empirical evidence that the algorithmically-derived risk variable is racially biased. We discuss our study in the context of how the SSL reinforces a pseudoscientific justification of the PIC and call for the abolition of these tools broadly.