Author Wai-Yin Wan and Don Weatherburn
Published September 2016
Report Type Crime and Justice Bulletin No. 198
Subject Assault; Recidivism / Re-offending; Statistical methods and modelling
Keywords Violence, longitudinal study, survival, cure model, re-offending, loglogistic



To determine:

(1) the long-term risk that someone charged with a violent offence will commit another violent offence
(2) what factors influence the likelihood of desistance and the length of time to the next violent offence for those who do re-offend.


All 26,472 offenders who were born between 1986 and 1990 (inclusive) and who had at least one violent offence proved against them in New South Wales (NSW) before December 31st, 2014 were followed up to December 31st, 2015. An offence was counted as proved if at the index contact it resulted in a caution, a youth justice conference or proven court appearance. The mean follow-up time for offenders in the study was 6.35 years (range = 21.3 years; interquartile range = 4.7 years). Bivariate correlates of time to re-offend were identified using log-rank tests. Multivariate analysis of survival time was undertaken using a cure fraction model with a loglogistic distribution of survival time.


In the median case, after 20 years, an estimated 23 per cent of violent offenders committed a further violent offence. However the risk of violent re-offending varies greatly across different offender groups, being much higher for Indigenous offenders, those who were aged 17 and under at the time of their index contact and those whose first contact with the criminal justice system occurred when they were 12 years of age or younger. There is little evidence of specialisation among violent offenders in the sample. Most have committed a wide variety of different offences prior to their conviction for a violent offence and those who do re-offend commit a wide variety of offences.


Authorities charged with responsibility for making bail, sentencing and parole decisions in relation to violent offenders need to pay close attention to the characteristics of the violent offenders they are dealing with. Evaluations of violent offender programs should include both short-term and long-term follow up. Prison is not a very effective instrument through which to reduce violent offending