Keep up-to-date with drugs and crime

The latest research, policy, practice and opinion on our criminal justice and drug & alcohol treatment systems
Predicting sexual offending
Comparing two predictors of sexual recidivism: the Risk Matrix 2000 and the OASys Sexual Reoffending Predictor

Share This Post

Risk assessment for sex offenders

Earlier this week (26 January 2020), HMPPS published a new report in the MoJ Analytical series: Comparing two predictors of sexual recidivism: the Risk Matrix 2000 and the OASys Sexual Reoffending Predictor. The study, authored by Philip Howard & Helen Wakeling, aimed to inform decisions about the risk assessment of men convicted of sexual offences. It did this by testing the predictive validity for sexual reoffending – the ability to separate lower- and higher-risk individuals – of two actuarial risk instruments cited in the report’s title.

The study

Previous research has found OSP to be a slightly better predictor of proven contact sexual reoffending than a simulated version of RM2000/s. This study re-examined this issue using full RM2000/s scores and minor revisions to OSP’s scoring. It also compared a further OSP scale, developed for indecent image reoffending, with RM2000/s. Finally, the study examined whether length of time without offending in the community (on release from prison) affected contact sexual reoffending rates and whether this should be considered in determining individuals’ risk ratings. It examines these issues using a sample of men released from custody in the 2000s.

A prison sample of 2,728 men who had been convicted of at least one sexual offence and who had been risk assessed for participation in an offending behaviour programme, was utilised, and rates of proven reoffending were examined.

RM2000/s, OSP/Contact (OPS/C) and OSP/Indecent images (OSP/I) risk categories were produced for each individual in the sample, and the predictive validity of the tools were examined. The authors set out the limitations of the study which study include the fact that the sample is historic, and that most had contact sexual offences as their index offence.

Predicting sexual reoffending

Actuarial risk assessment tools – which arrive at an overall risk level using empirically-supported risk factors and mathematical rules, rather than professional judgement – have been shown to be good predictors of proven reoffending. The actuarial tool Risk Matrix 2000/s (RM2000/s) was developed using a large sample of adult males convicted of sexual offences in the United Kingdom (UK) and cross-validated with further UK samples. HMPPS and the Police Service currently use it across England and Wales to assess the sexual recidivism risk of such men.
Although RM2000/s is well evidenced and validated, a new actuarial tool called the OASys Sexual Reoffending Predictor (OSP) was developed by Howard and Barnett (2015) in order to address some concerns with RM2000/s. In its initial form (now OSP/C), OSP focused on those offences which tend to cause the most serious harm to the victims (i.e., contact sexual offences), rather than all sexual recidivism. It now also has an additional scale, OSP/I, which predicts the risk of indecent images of children recidivism.

In addition to its focus on specific offences, OSP was designed to be less resource intensive than RM2000/s, and to be used with all men who have committed sexual offending (RM2000/s cannot be used on those whose sexual offending was confined to childhood, i.e., all offending took place before the age of 16).

The table below, reproduced from the study, examines the distributions and contact sexual and indecent image reoffending failure rates by risk category, at the five-year point using life table methods.


The study found that the OSP tool was more accurate, particularly for offences relating to indecent images:

  • OSP/C was a slightly better predictor of proven contact sexual reoffending than RM2000/s.
  • By far the best predictor of proven indecent images reoffending was OSP/I.
  • Analysis, based on a small sample, indicated that those contact sexual offenders who had been in the community offence-free for a five-year period had a lower risk of reoffending beyond the five-year point. These results, therefore, lend some support to the recent guidance proposing the reduction of risk by one category for every five years an individual convicted of a sexual offence has been in the community offence-free.


The authors conclude that the results provide further support for the use of OSP over RM2000/s in assessing the risk of proven contact sexual reoffending in men with a sexual offending history. OSP has a number of advantages over RM2000/s: it focuses on specific types of sexual reoffending, it has a simpler scoring process, there is no age restriction on its use and it has better predictive validity for its two outcomes.


Thanks to Jen Theodore for permission to use the header image previously published on Unsplash.

Share This Post

Related posts

On Probation
Procedural justice in probation

HMPPS developed and tested a measure of procedural justice to see how fair people on probation thought their supervision was.

How do you find work as a released sex-offender?

‘It’s ok if you were in for robbery or murder, but sex offending, that’s a no no’: a qualitative analysis of the experiences of 12 men with sexual convictions seeking employment.

One Response

  1. If anyone using a pre-programmed algorithm to establish likelihood of sexual re-offending, add to it a 15 minute telephone conversation of probation officer , who is just ticking boxes of algorithm and decides of the future of the young first time offender, who offended only because he was under influence of psychogenic drugs….
    Why don’t we use a crystal ball, or tarrot cards, or even horoscope!
    Just to re-cap my worry: a clearly psychic probation officer can establish during the 15 min telephone conversation with great help of OSP/C and report on person that have never met or seen as oppose to character reports of 5 humans! who know young offender, meeting him daily and in every social situation, going on holiday together, interacting with all age groups and sexes?
    There is no program to know better a human than another human from mutual interaction.
    Are you all in favour of OSP/C, OSP, OSPI, RAM2000 off your head? Who makes those tools? And another question: who actually let them circulate in the system? You and your tools are frankly making me sick- the worse thing you can do is: to predict a human behaviour with help of some programs! We are all unique and even the smartest geek can not make the program that fits all! The system has a big flaws and those have to be exposed and corrected.
    – not giving a chance to first time offenders- just because you have power to make decision which is at your worse
    – consequently you destroy a young life which had no chance to live! The jobs, the studies, the mortgage , just about any form one has to fill in for bureaucracy reasons, includes that little box asking you whether you have a record / being offender!
    Let me ask you a question: if your family member was assessed by 15 min on the phone and OSP/C algorithm and the whole family knew that it is total untrue, fiasco, what would you do? Would you still agree that such a systems are really helpful in covering up incompetence of Probation Officers and help with harsh sentencing of “out of character offenders”, who offended only under influence of first time tried drugs?

    Lets start from you, or your family members, friends, cousins etc how many of you were trying forbidden things in your teens? Many of you were lucky because were not caught. And so are you now impaired adults? No you are not , because you were not prosecuted, you have learned from brush with law and you lived your lives, studied, worked, with clean slate. What future has 18-19 years old starting already with tarnish slate? What about psychological state of his mind to carry lifelong burden which could not be turned back?
    So before applying wrong sentencing, stop and think back….have empathy and understanding, where required. Don’t use a personal negative experiences or situations blown out in social media’s to cloud your professionalism and fair judgement.

    Please feel free to contact me, I have for you a very good and unfortunate personal experience.
    I have no idea who invented rules of law that make an intelligent brain stand still from stupidity applied.
    Now that I had a taste of UK jurisdiction, I am frankly not surprised if people handle law by their own hands…

Leave a Reply

Your email address will not be published. Required fields are marked *

Probation posts sponsored by Unilink


Excellence through innovation

Unilink, Europe’s provider of Offender/Probation Management Software


Get every blog post by email for free