The Offender Group Reconviction Scale (OGRS) is the name of a series of actuarial risk assessment instruments for use with convicted criminal offenders. Each version of OGRS has been developed for and used by the adult correctional services of England and Wales, directed since 2004 by the National Offender Management Service, which in 2017 became Her Majesty’s Prison and Probation Service (HMPPS). Versions of OGRS have been used to predict general and sometimes more serious recidivism among probationers since 1996 and prisoners since 2009. OGRS scores are of considerable importance in court reporting as well as in allocation of offender services and research. Successive updates to OGRS have amended the measurement and scoring of its risk factors to improve its usability and validity as a predictor of reoffending, as well as to ensure that it reflects contemporary patterns of recidivism. This article examines the application and scoring of the OGRS and provides an overview of its development.
Applying a Static Actuarial Risk Assessment Instrument
OGRS fulfills the demand for an actuarial risk assessment instrument with high levels of predictive validity (as confirmed by prevailing standards in forensic psychology), scored only from easily available static risk factors. Scoring each version of OGRS requires the offender’s gender; dates of birth, conviction, and discharge from custody (when this sentence has been passed); current offense; and a small number of criminal history variables. The resultant scores are estimates of the probability of recidivism within 2 years of discharge from custody or sentencing to a noncustodial court disposal. At court, probation services include OGRS scores in presentence reports to the sentencing judge or magistrate. (Presentence report assessments always assume that a noncustodial sentence will be passed.) After sentencing, minimum OGRS scores form part of the criteria determining suitability for offending behavior programs, operationalizing the risk aspect of the
Risk-Need-Responsivity paradigm upon which HMPPS programs are based. OGRS score thresholds are also referred to when determining an individual’s overall level of probation service and, in aggregate, when producing management information reports used in the commissioning of rehabilitative interventions and other offender services. OGRS scores are used by researchers to control for recidivism risk when evaluating interventions and have been used to compare outcomes between those receiving different sentences. They are also a key part of the Payments by Results contractual agreements between the Ministry of Justice and private probation providers, known as Community Rehabilitation Companies (CRC). These agreements specify that the CRCs will receive additional payments, known as Payments by Results, when cohorts of offenders have recidivism rates significantly lower than their OGRS baseline. The reoffending rate of each cohort is compared with that of the baseline 2011 cohort, and a comparison of OGRS4/G scores for the baseline and current cohort is used to adjust the current reoffending rate to reflect changes in the case mix of offenders being supervised.
The Development of the OGRS
Version 1 of the OGRS (OGRS1) was published in 1996. Its development involved several stages common to all versions. The criminal records of tens of thousands of offenders given sentences involving custody or probation management were traced on a central database to score risk factors and determine recidivism status within 2 years of community sentence or discharge from custody, weighting risk factors in a logistic regression model. At that time, the source database was the now defunct Offenders Index, which listed only the dates of convictions for relatively serious offenses. A key risk factor derived from these criminal histories is the Copas rate, named after a coauthor of OGRS1. An offender’s Copas rate, and thus OGRS score, is higher when the offender has many criminal appearances and when the offender’s criminal career (i.e., from first to current appearance) is short. Younger, male offenders and those with current offenses of burglary and theft also receive higher scores in all versions of OGRS.
Published in 1999, OGRS version 2 (OGRS2) was designed to be scored using personal computers rather than pocket calculators (which were used with the OGRS1) and involved a more complex set of risk factors. An additional scale, OGRS2-SV, estimated the combined risk of sexual and violent recidivism.
OGRS version 3 (OGRS3) was published in 2009, and it introduced several substantial technical and practical improvements. The Offenders Index was replaced with a research extract of Police National Computer data, which records a wider range of offenses and includes those resulting in formal cautions as well as convictions. As Police National Computer data include offense dates, the new recidivism outcome, proven reoffending, measures offenses committed (not convicted) within 2 years that eventually result in caution or conviction, while criminal history counts now include cautions. As further validation showed that the OGRS2SV Scale failed to predict proven sexual reoffending, it was withdrawn and not replaced. The underlying algorithm was revised to model age separately for men and women, increasing predictive validity for female offenders. Efforts to lessen burdens on assessors led to a reduction in the number of data items required from nine to six, with trivial impact on predictive validity. OGRS3 was the first version implemented in prisons as well as probation, improving the integration of sentence plans for released offenders.
Published in 2015, OGRS version 4 (OGRS4) models reoffending over any given 2-year period, rather than only the first 2 years after sentence or discharge. The amount of offense-free time since sentence/discharge is therefore an additional scored factor. Development of an evidence-based classification of nonsexual violent offenses allowed the introduction of a violent reoffending predictor, OGRS4/V, with the general reoffending predictor now named OGRS4/G. Instruments combining static and dynamic risk factors, versions 2 of the OASys Violence Predictor and OASys General Reoffending Predictor, were developed using the same sample. Comparable violent and general recidivism predictions are therefore available for all offenders, with the slightly superior versions 2 of the OASys Violence Predictor and OASys General Reoffending Predictor recommended in preference to OGRS4/V and OGRS4/G where possible.
Future Research
As of early 2018, OGRS3 is in use across HMPPS. OGRS4 is used to calculate Payments by Results and awaits operational implementation through upgrades to HMPPS case management software. Future developments being considered by HMPPS analysts include the use of alternate statistical methods (e.g., random forests) to estimate risk and the development of live links to the Police National Computer to score predictors automatically. Improvements to case management software should also facilitate more frequent updates to OGRS and other actuarial instruments, as complex code revision processes will be replaced by a dedicated risk-scoring software module.
References:
- Howard, P. (2015). OGRS4: The revised Offender Group Reconviction Scale. In R. Moore (Ed.), A compendium of research on the offender assessment system (OASys) 2009-2012. London, UK: Ministry of Justice.
- Howard, P., Francis, B., Soothill, K., & Humphreys, L. (2009). OGRS 3: The revised Offender Group Reconviction Scale. Ministry of Justice Research Summary 7/09. London, UK: Ministry of Justice.
- Kershaw, C. (1999). Interpreting reconviction rates. In M. Brogden (Ed.), The British criminology conferences: Selected proceedings. Vol. 2. Papers from the British criminology conference, queens university, Belfast, 15–19 July 1997. Retrieved from http://www.britsoccrim.org/volume2/005.pdf
- Travers, R., Mann, R. E., & Hollin, C. R. (2014). Who benefits from cognitive skills programs? Differential impact by risk and offence type. Criminal Justice and Behavior, 41, 1103–1129.