Standardized crime rates (e.g., “homicides per 100,000 people”) are commonly used in crime analysis as indicators of victimization risk but are prone to several issues that can lead to bias and error. In this study, a more robust approach (GWRisk) is proposed for tackling the problem of estimating victimization risk. After formally defining victimization risk and modeling its sources of uncertainty, a new method is presented: GWRisk uses geographically weighted regression to model the relation between crime counts and population size, and the geographically varying coefficient generated can be interpreted as the victimization risk. A simulation study shows how GWRisk outperforms naïve standardization and Empirical Bayesian Estimators in estimating risk. In addition, to illustrate its use, GWRisk is applied to the case of residential burglaries in Belo Horizonte, Brazil. This new approach allows more robust estimates of victimization risk than other traditional methods. Spurious spikes of victimization risk, commonly found in areas with small populations when other methods are used, are filtered out by GWRisk. Finally, GWRisk allows separating a reference population into segments (e.g., houses, apartments), estimating the risk for each segment even if crime counts were not provided per segment.
This is an open access article distributed under the Creative Commons Attribution License
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.