Author Joanna JJ Wang and Suzanne Poynton
Published October 2017
Report Type Crime and Justice Bulletin No. 207
Subject Recidivism / Re-offending; Sentencing; Statistical methods and modelling
Keywords intensive correction order, re-offending, propensity score weighting, bivariate probit model

Download this publication



To compare reoffending rates between those who received an intensive correction order (ICO) and those who received short prison sentence (less than two years).


Offenders’ demographic characteristics, index offence characteristics, prior convictions and penalties, LSI-R score and re-offences were extracted from the Re-offending Database maintained by the NSW Bureau of Crime Statistics and Research. Using logistic regression model with inverse probability of treatment weighting, the effect of penalty choice on re-offending was estimated. Doubly robust estimation and bivariate probit model with an instrumental variable were also used to address potential model misspecification and endogeneity of penalty assignment. As a sensitivity analysis, separate modelling was performed for offenders who were in medium to high risk categories and the prison group was restricted to those with 6 months or less fixed term.


There was a 11%-31% reduction in the odds of re-offending for an offender who received an ICO compared with an offender who received a prison sentence of up to 24 months. The bivariate probit model with an instrumental variable
did not reveal a significant effect or evidence of endogeneity.


These results further strengthen the evidence base suggesting that supervision combined with rehabilitation programs can have a significant impact on reoffending rates.

Download this publication