Model Uncertainty in Ecological Criminology: An Application of Bayesian Model Averaging With Rural Crime Data

Abstract

In this study we explore the use of Bayesian model averaging (BMA) to address model uncertainty in identifying the determinants of Midwestern rural crime rates using county level data averaged over 2006-07-08. The empirical criminology literature suffers from serious model uncertainty: theory states that everything matters and there are multiple ways to measure key variables.By using the BMA approach we identify variables that appear to most consistently influence rural crime patterns. We find that there are several variables that rise to the top in explaining different types of crime as well as numerous variables that influence only certain types of crime.