Risk of Reconviction

Statistical models which predict four types of re-offending. (1999)

 

The ROC*ROI Measures - Explanatory Notes

The ROC*ROI is a measure of risk currently in use throughout the Department of Corrections.

The ROC*ROI measure is in fact a combination of two measures. ROC equals Risk of ReConviction, while ROI equals the Risk of Imprisonment. The asterisk (*) means "multiplied by". Hence the term ROC*ROI is a measure that means: the offender's Risk of ReConviction multiplied by the offender's Risk of Imprisonment.

These measures derive from the mathematical relationship between basic social and demographic variables (e.g. age, gender), criminal history variables (e.g. age of first offence, time free in community since thirteenth birthday, seriousness of previous offences, length of time between offences) and future offending. The ROC*ROI has been used because research has shown that even simple risk scales (i.e. a checklist of risk factors) invariably outperform the clinical or professional judgements of trained experts and experienced correctional staff when making predictions about future offending.

The ROC*ROI measure is an expression of the likelihood that a person will be both reconvicted in the future and be sentenced to a term of imprisonment for that offence. As a combined measure, it is quite possible that any individual may have a very high chance of re-offending (say 90%) but a very low chance of also being sent to prison for that offence (say 10%). In that case, the actual chance of someone being both reconvicted for an offence, and being sent to prison for that offence would be only 9 percent. Conversely, of course, it is possible for a person to have a very low chance of offending, but a very high chance of receiving a prison term if they do. Again, the combined value expressed by the ROC*ROI measure would result in a low probability of that person being reconvicted and sent to prison.

If someone has a lengthy history of regular and persistent minor offending, their ROC*ROI should indicate them to be at low risk. The risk of conviction (ROC) would be high, but their risk of being sent to prison would not.

The Department has adopted ROC*ROI as its measure, rather than just risk of conviction alone, because this gives some indication of serious re-offending, the type of offending which Corrections is attempting to target under its offender management system.

 

Description of offender and offending history

Offenders risk of conviction

(i.e., ROC score only - what some people base their own judgements of reoffending on, often leading to overestimation of risk of serious reoffending)

Offenders risk of conviction AND risk of imprisonment

(i.e., ROC*ROI score which more accurately reflects the probability of serious reoffending)

Offender A is a 23-year-old female convicted of a breach of Community Work. She has five previous convictions, for wilful trespass, receiving and unlawfully getting into a motor vehicle.

0.94

0.38

Offender B is a 41-year-old male convicted of Indecent Assault on a Female under 12 years old. While this is his first conviction for a sexual offence, his offending occurred continuously over a substantial period of time. He has three previous convictions for violence.

0.94

0.88

Offender C is a 30-year-old male convicted of driving while disqualified. He has several previous convictions for property and drug-related offences. However, the last of these convictions was in 1990.

0.96

0.59

Offender D is a 22-year-old male convicted of theft ex person, possession of cannabis and a failure to answer police bail. He has 40 previous convictions, mainly property offences, but also including aggravated assault and several breaches and failure to answer bail charges. He has served at least four short prison sentences over the last four years.

0.99

0.95

 

Anomalous Results

A number of anomalous results have been reported with the use of ROC*ROI in the field. An anomalous result is when the result does not seem to match the other information presented for the offender and the offence. Several of these cases are explained below.

Driving while intoxicated

A number of people who have been convicted for repeat drunk driving offences appear as low or moderate risk using the ROC*ROI tool. Many people who are convicted for repeat drink-driving offences have few previous convictions for other offences, and tend not to have received prison sentences. Consequently, the way the ROC*ROI model operates they would be assessed as having high probability of being reconvicted, but a low probability of being sent to prison. Since the ROC*ROI was developed, stricter sentencing has been put in place for repeat drink driving which has not been taken into account.

Repeat driving offenders are priority targets for the MODS (Making Our Drivers Safe) programme (a programme run by the Community Probation Service for high-risk repeat driving offenders). Consequently, if imprisonment is being considered for a third or subsequent excess blood alcohol count, and the offender has a low or medium ROC*ROI, override would be normally considered by Corrections staff.

Men who offend sexually against children

Many child sex offenders have very low ROC*ROI scores. This reflects the fact that often this is a specialist form of offending, which occurs at a very low frequency with long gaps between offences. Sexual offending against children may also go undetected for long periods due to the nature of the offending and its effects on victims.

Some of these men may have accumulated lengthy histories of non-sexual offending - this will elevate their risk score, but it is not a good predictor of their further sexual offending. Because of the seriousness of this type of offending, and because of the potentially devastating effects on the victim, all men convicted of this type of offence are normally classified by Corrections as high risk, and targeted for intervention.

ROC*ROI has been proven accurate at predicting re-offending among people convicted of aggressive sexual offences (such as rape), and in a recent evaluation of Parole Board cases, the ROC*ROI measure proved to be a good predictor of serious sexual and/or violent offending for men who had been released from prison following completion of their sentence for serious aggressive sexual offences against adults.

Young offenders

There are cases of very young offenders who come into the criminal justice system, who show no official record of offending in the adult courts, but who may have extensive offending histories in the juvenile court. In these cases, the ROC*ROI measures in IOMS (Corrections' computerised offender database) can only be calculated on the criminal history data which is available, and this does not include Youth Court criminal histories. Professional override by Corrections is considered for offenders with extensive Youth Court criminal histories.

Overseas convictions

The seriousness of overseas convictions cannot be calculated through IOMS. Offenders with a history of serious and persistent offending overseas should be considered for override.