An Empirical Test of the Homology Assumption in Criminal Profiling
Criminal profiling (CP) involves predicting a criminal's characteristics based on crime scene evidence. One fundamental assumption underlying CP is that offenders who exhibit similar crime scene actions will share similar characteristics - i.e., the homology assumption. In a profiling task, some profilers first categorize crime scene behaviors into a behavioral typology (e.g., sadistic rapist). The resulting typology is then used to invoke a standard set of background characteristics (e.g. type of previous convictions). Despite the importance of this assumption, no study has tested the ability to generalize existing typologies. This study tests the ability to generalize of three existing profiling typologies. Demographic and crime scene behaviors were compiled for three data sets: arson (n = 87), burglary (n = 36) and robbery (n =177). Crime scene behaviours associated with each offence were used to classify offenders into a typology and background information for each offender across the various typologies was compared. Results provide little support for the homology assumption. The implications of these findings for the credibility of CP and police investigations are discussed.
01 Jan 2006
31 Dec 2008
Strategic Research Theme
Governance and Public Policy