A B S T R A C TThis study presents a new approach to developing a typology of criminal activity. The distinguishing feature of the analysis is that it concentrates on determining types of activity rather than the amounts of activity over the life course. The methodology involves investigating criminal activity in a succession of five-year periods rather than the conventional approach of summarizing a 'lifetime' of crime. This provides scope for assessing changes and pathways of criminal activity as offenders grow older, and gives new insight into the concepts of specialization and versatility. The Home Office Offenders Index birth cohort for 1953 provided official conviction histories up to 1993 (age 40), and latent class analysis identified a fixed number of types of criminal behaviour separately for males and females. The patterns of offending varied markedly between males and females. Male offending (with nine identified types) showed greater diversity than female offending (with three identified types). For the males, each type of offending had a distinct age profile, but this was not evident with the females. A new definition of offending specialization is given, and is shown to increase for males as offenders grow older. A case study on one of the male offending types illustrates the potential for identifying pathways of crime. K E Y W O R D STypologies / Criminal Trajectories / Clusters / Latent Class Analysis / Pathways / Specialization.

The aim of the study This study focuses on developing a typology of criminal activity. It aims to identify a fixed number of types of criminal behaviour separately for males and females. Age profiles for each type of criminal activity are also constructed. Finally, the study probes the notion of criminal pathways. What proportion of offenders are specialists within one sphere of activity? How many tend to migrate from one sphere to another as they become older? The distinguishing feature of the analysis is that it attempts to describe criminal activity over a five-year period rather than the conventional approach of summarising a 'life-time' of crime. The five-year summaries of criminal activity are likely to have a greater practical use, for they will help practitioners, such as the police and probation officers, to understand a recent criminal history. Methods Data from the 1953 and 1958 birth cohorts drawn from the Offenders Index was analysed. Conviction histories for individuals were divided into five-year strips defined by the age of the offender at the date of conviction. Using latent class analysis, and examining males and females separately within each cohort, the approach succeeded in identifying clusters of offending behaviour.