Predicting and understanding risk of re-offending
This report examines the accuracy of tools used to assess the risk of violent offenders re-offending; the prevalence of offenders potentially classifiable as dangerous and severely personality disordered (DSPD); the reconvictions made by people in this group after release; the element of risk attributable to DSPD; and the likely relationship between DSPD and indeterminate sentences.
Title: Predicting and understanding risk of re-offending: The Prisoner Cohort Study
Authors: Jeremy Coid, Min Yang, Simone Ullrich, Tianqiang Zhang, Amanda Roberts, Colin Roberts, Robert Rogers, David Farrington
Series: Ministry of Justice Research Summary 6
Number of pages: 9
Date published: October 2007
Availability: Download full report 
PDF 187Kb
Key Points
Offenders with Dangerous and Severe Personality Disorder
- 15% of the prisoners (212 of the 1396) fulfilled criteria for DSPD. This is the first time such a benchmark has been identified, albeit provisionally.
- Comparison between DSPD and non-DSPD offenders revealed statistically significant differences in age, ethnicity, socio-economic class, and marital status.
- Statistically significantly more DSPD offenders were reconvicted after release into the community. They accounted for statistically significantly more major violent and acquisitive convictions.
- The risk of violent and acquisitive reconviction statistically associated with DSPD is considerable. Successful treatment or management of the link between recidivism and DSPD could therefore lead to a significant reduction in the number of reconvictions in the DSPD group, and hence in an offender population with similar characteristics to the prisoner cohort.
- There was a significant overlap of offenders classifiable as DSPD and those potentially qualifying for extended/indeterminate sentencing according to the Criminal Justice Act 2003, Sections 225 and 227. This has implications for the targeting of interventions.
Comparison of the predictive accuracy of risk assessment instruments
- All instruments predicted reconvictions for violence, theft and drug-related offences, and for a combined category of any offence. All instruments studied were moderately accurate and statistically better than chance, even though not all the instruments were intended for individual prediction.
- The Offender Group Reconviction Scale (OGRS-2) ranked as the most accurate instrument for predicting all outcomes of reconviction.
Risk prediction at item level
- Analysis of prediction at the item level showed that the power of each instrument to predict reconviction was often concentrated in a subset of items, amongst which static variables predominated.
- A combination of actuarial and clinical assessment is highly recommended in formulating risk for any given individual. Future research should be aimed at investigating further the predictive properties of static items in risk instruments and their relationship with violent offending, and at identifying additional dynamic factors which are powerful predictors. This could help to improve the management of dangerous offenders for public protection.
Getting a copy
Download Predicting and understanding risk of re-offending: The Prisoner Cohort Study PDF 187Kb
Last update: Thursday, October 25, 2007