Joint Spatial Mapping of Multiple Crimes Using a Multivariate Conditional Autoregressive (MCAR) model approach

Abstract

In this study, a multivariate Bayesian spatial modeling approach was used to jointly model the counts of two types of crime, i.e., armed robbery and theft (stealing) across sub-national level in Nigeria. The approach explores the geographic pattern of crime risks and relevant risk factors. In contrast to the univariate model, which assumes independence across outcomes, the multivariate approach takes into account potential correlations between crimes. Five co- variables are included in the model as potential risk factors include unemployment rate, economic index, young male 18-35 population, population density, education index and number of police area commands. The overall result showed that the multivariate approach outperforms the univariate model in term of smaller Deviance information criteria (DIC). Unemployment and young males are found to be positively associated with crime rates, while the number of police commands per state would reduce (negatively) the crime rate although it was not significant. In addition to the risked factors, the proposed approach further estimated the conditional correlation between the two comes, spatial dependence and geographical pattern of variation of individual crime.