UC Irvine

Objectives: Current strategies for locating where offenders live either focus exclusively on individual suspects or generalize to entire neighborhoods. However, better estimates of where offenders are located may improve models of the ecological distribution of crime, and forecasts of the locations of future crime incidents. Methods: We propose a novel reverse spatial patterning (RSP) strategy that estimates where offenders may live based on the spatial locations of crime events. We rely on a distance decay function – based on the consistent finding that offenders do not travel far to commit crime – and Hipp's (2016) general theory of spatial crime patterns, to work backwards from the locations of actual crime events to make predictions about where offenders may live in subsequent years. We then use these estimates in models predicting crime locations. We create two versions of the RSP: one which assumes everyone is equally likely to offend, and another that creates an estimate assuming disproportionate offending across persons. Results: We test the effectiveness of our proposed strategy for these two measures using offense and arrest data from St. Petersburg, FL, and assess how well they predict the location of offenders (proxied by arrestees) and future crime events. We find consistent evidence that our RSP strategy provides better predictions of the locations of where offenders are located and also future crime incidents across a variety of crime types compared to existing strategies. Conclusion: The RSP approach is useful for creating estimates of where offenders live, which allow for better predictions of the locations of future crime incidents. These better forecasts will allow for more efficient allocation of police resources and targeted crime suppression efforts.