What works in offender profiling? A comparison of typological, thematic, and multivariate models

Utilizing a sample of 85 stranger rapists, three models (Hazelwood's (1987) Power and Anger FBI model, the Behavioral Thematic evaluation of Canter, Bennell, Alison, and Reddy (2003), and the Massachusetts Treatment Center: Rape classification system revision 3 (MTC:R3, Knight & Prentky, 1990)) were contrasted with a multivariate regression approach to assess their ability to predict an offender's previous convictions from crime scene information. In respect of the three aforementioned models, logistic regression and AUC analysis indicated that the Power and Anger FBI model was the most effective, followed by the MTC:R3, and then the Behavioral Thematic evaluation. However, predictive analyses based on a multivariate approach using a mixture of crime scene behaviors, as opposed to the grouping of behaviors into themes or types as in the three models, far exceeded the predictive ability of the three models under AUC analysis. The results suggest that emphasis should be placed on further exploration of the predictive validity of each of the individual behaviors that comprise existing thematic, typological, and multivariate classification systems, especially those that are subject to inter-situational variation.