Allocating Police Resources While Limiting Racial Inequality

Abstract

Police targeting hot spots of crime tends to disproportionately burden minorities via stops and arrests. This work attempts to reduce disproportionate minority contact by formulating a crime hot spots spatial allocation strategy for police that prioritizes areas of high crime, but constrains the targeted hot spots given different levels of acceptable racial inequality. This racial inequality constraint is measured as the proportion of minorities likely to be stopped in those areas prioritized by police. Using data on stops and crime in New York City, I show that police stops can be more equitably distributed according to race, but there are fundamental trade-offs. One cannot gain a racial distribution of stops proportionately equal to the residential population without large decreases in the efficiency of targeting high crime areas. More modest gains can be had though in reducing the proportion of minorities stopped while still targeting high crime locations.

Mapping the Risk Terrain for Crime Using Machine Learning

“…The current approach of this paper-as with the majority of crime and place based research-does not aid in accomplishing that objective. Note however, that incorporating such individual aspects into the predictive models comes with additional concerns of disparate impact that can occur as the result of predictive models (Ridgeway 2018), although such concerns are not entirely absent from spatial predictive models either (Brayne 2017;Harcourt 2007;Kochel 2011;Wheeler, 2019a). Interpretable model summaries can effectively address how different factors contribute to the prediction, even if they cannot depict the entire complexity of a machine learning model (Wachter, Mittelstadt, and Russell 2018).…”

The effect of public surveillance cameras on crime clearance rates

“…Ultimately having public CCTV results in some loss of privacy relative to no cameras, but that loss will likely be more palatable to the general public if there is evidence that cameras result in additional benefits, which include both overall crime reductions, as well as improving the likelihood that a crime is solved. Future research should attempt to identify those reasonable trade-offs in efficacy given more cameras in place, especially as it relates to surveillance in minority communities (Wheeler, 2019).…”

Learning to rank spatio-temporal event hotspots

“…While hotspot policing has been shown to yield crime rate reductions, there is the possibility of unwanted side effects of hotspot policing such as traffic stops that unfairly target minority populations, stop and frisk, and other police activities that have negative societal consequences. There has been some recent work on improving fairness of spatial crime forecasting algorithms (Wheeler 2019;Mohler et al 2018) where a fairness penalty is added to the optimization algorithm. Future research may focus on incorporating fairness into learning to rank models of crime, similar to methods that incorporate fairness into learning to rank for information retrieval (Zehlike and Castillo 2018).…”