Duration of the time to reconviction: Evidence from UK prisoner discharge data

Abstract

Duration models were used to investigate reconviction risks for a sample of 34,126 offenders released from prison in England and Wales during 1998. Different versions of the Cox proportional hazards model were applied to a comprehensive data set covering several offense types. Factors such as age (modelled using bands rather than a linear or quadratic format), gender, convictions-history, and offense type were found to be strongly associated with the risk of reconviction. Interactions between factors were also included and it was shown that the discrepancy between the risk of reconviction for males and females became weaker as the number of previous convictions increased. The study helped identify the subgroups of offenders for whom reconviction risks are greatest and the times when they seem to be most vulnerable to reconviction. Implications are explored for the design of prisoner resettlement programs and the prioritization of offenders for more intensive forms of intervention.