Investigating the Effective of Cross-Jurisdictional Data Mapping on Predictive Crime Mapping Accuracy

• In Event: The Value of Integrating Data across Agencies and Jurisdictions to Improve Public Safety

Abstract

Prospective crime mapping applications have only recently been adopted by police agencies that are moving towards a risk-based policing strategy. The accuracy and utility of these predictions, however, are significantly impacted by the quality and variety of data that are used in modeling risk of crime, and in the geographic distribution of the data—or whether the data are artificially truncated at a jurisdictional boundary (Rengert & Lockwood, 2009) even while crime hotspots, offender movements, or target-rich environments may not respect such arbitrary boundaries. This study employs both cross- disciplinary and cross-jurisdictional data to highlight changes in accuracy of predictive crime mapping when information is integrated across law-enforcement jurisdictions. Predictive crime mapping is facilitated with geospatial preference modeling and spatial naïve Bayes modeling. Implications for ways such predictive mapping practices can improve the efficacy of policing are discussed.

Authors

• Meagan Cahill, Urban Institute

• Samuel Bieler, Urban Institute