ABSTRACT
Most attempts at developing typologies of criminal behavior have not involved empirical research. This paper describes an exploratory empirical approach to identifying patterns in criminal behavior. Two data-reduction techniques, factor analysis and cluster analysis, are applied to the official arrest records of a Danish birth cohort of 28,879 men. Four factors emerged from the factor analysis: GENERAL CRIME, TRAFFIC OFFENSES, WHITE-COLLAR CRIME, and SEX OFFENSES. The cluster analysis revealed GENERAL CRIME and TRAFFIC OFFENSES clusters. A substantial number of offenses are shown by both analyses to be independent of any pattern. The results show good split-sample cross-validation and for the most part are robust across the two analytic approaches.