One prison service has come up with an innovative and fascinating risk-based approach to sentencing.
One of the ultimate challenges in risk management is to predict the chances that a felon will reoffend. With so many socioeconomic factors at stake, one prison service has taken a new risk-based approach.
According to a report by US-based The Oregonian, a pilot study is underway to determine how a new risk model can be used by judges at sentencing, or by prosecutors and defence lawyers negotiating a plea deal, to assess the chances that convicted felons released from prison will commit a new crime or be rearrested – based on their age, gender and criminal history.
The actuarial tool, called the Oregon Public Safety Checklist (PSC), measures the likelihood an offender will be convicted of a new felony within three years of release from prison or imposition of probation, measured from estimate patterns of 345,000 offenders released from prison or sentenced to probation since 1980.
The PSC presents correlations between 10 current risk factors with the level of risk for future crime, set against the percentage of offenders who have a lower risk of future crime.
An algorithm is then used to weigh the offender's predictive factors and give a ‘composite score’ for an offender who presents these characteristics, much in the same way as in life insurance actuarial tools, on three levels:
the risk the offender will be convicted of a new felony within three years of release;
the person’s risk of being arrested for a property crime within five to 10 years of release;
and his or her risk of arrest for a crime against a person within five to 10 years of release.
The PSC does not account for dynamic factors such as rehabilitation and drug treatment, and currently does not include first-time offenders or offenders who only have an arrest history and no conviction history.
However, Oregon State found the new checklist is more reliable. The old case management system was accurate about 63% of the time, while the new checklist has a 78% accuracy rate.