A statistical model examining repetitive criminal behavior in acts of violence.

A simple statistical model for examining repetitive criminal behavior in acts of violence is described. The units, called "parameters," are nonquestionable data concerning environment of the crime, personal properties, and postmortem findings of the victim, obtained by double-blind investigation performed by two forensic pathologists. Parameters shared by two or more criminal acts allegedly committed by the same assailant were compared with the same parameters recorded from 50 or 100 other mutually independent criminal acts committed by other known assailants. This allowed an evaluation of the probability (p) of a crime pattern expressed as a parameter score to recur in mutually independent cases. The distribution of the score, when plotted on a logarithmic scale in all examples, showed an approximately normal distribution. The relation between probability (p), the estimated mean (means), and standard deviation (SD) yielded a normal curve. Different patterns of action by different perpetrators and patterns indicating repetitive behavior could be obtained. The method is applicable during investigation of crimes in which the perpetrator acts in a repetitive manner, as in serial murders.