Criminal Consistency and Distinctiveness

Abstract

Crime linkage is the process of grouping together crime events that share a common offender. It is the first step in investigation, profiling, and being able to predict the characteristics of future crimes. In the absence of DNA evidence, crime linkage is carried out by considering the characteristics and features of the crime, crime scene, or offender as recorded by police investigators. The ability of police to link the crimes from an offender depends on two aspects: how consistently the offender carries out their crimes and how distinctive their crimes are from the crimes committed by the other offenders operating in a region. The more consistent and distinctive an offender behaves, the easier it is for police to link their offenses. Likewise, the crime features (e.g., location, crime type, point of entry) that have the most consistency and distinctiveness across all offenders will be the most useful for linkage models. This paper develops two metrics for measuring the consistency and distinctiveness of a crime series; the consistency score based on Simpson’s index and the distinctiveness score based on the Kullback-Leibler Divergence. A Monte Carlo method is also developed to evaluate the statistical significance of the scores. We calculate the scores for the offenders in a mid-sized US county, identify the most important linkage features, and analyze the distribution of consistency and distinctiveness score for all offenders. These results can help police understand which crime features are most useful for linkage, measure the potential for linkage success in their jurisdiction, and identify the type of offenders that will be most difficult to apprehend.