A summary of evidence relating to offender risk assessment, risk of reoffending and risk of serious harm.
What influences criminal behaviour?
Criminal behaviour is influenced by a range of individual, social and environmental factors. People tend to interpret others’ behaviour as because of the sort of person they are. We often fail to see situational, environmental or social influences. Much decision making in criminal justice needs to be informed by an assessment of whether someone poses a risk to the public. For example, is there a risk that they might break the law again and might that be for a serious offence? Considerable effort has been made over the years to develop reliable, unbiased estimates of the risk of further offending.
The Offender Assessment System (OASys) was introduced in 2001 and built on the existing ‘What Works’ evidence base. It combines the best of actuarial methods of prediction with structured professional judgement to provide standardised assessments of offenders’ risks and needs, helping to link these risks and needs to individualised sentence plans and risk management plans1.
There are two main types of risk:
- likelihood of future re-offending and reconviction - the probability that someone will offend, be arrested, and reconvicted within two years
- risk of serious harm - if reconvicted, the probability that the offence will be one of ‘serious harm’
The Criminal Justice Act 2003 defines serious harm as: ‘death or serious personal injury, whether physical or psychological’.
OASys defines serious harm as ‘an event which is life threatening and/or traumatic and from which recovery, whether physical or psychological, can be expected to be difficult or impossible’.
The risk of harm posed by offenders to others has two key dimensions:
- the relative likelihood that an offence will occur
- the relative impact or harm of the offence - what exactly might happen, to what or whom, under what circumstances, and why
Some crimes like shoplifting, have relatively little impact or harm. But, statistically, they are the most common. Others like homicide are rare but cause greatest damage.
The level of serious harm is defined by the likelihood of it happening:
Low: current evidence does not indicate a likelihood of causing serious harm
Medium: there are identifiable indicators of serious harm. The offender has the potential to cause such harm. But they are unlikely to do so unless there is a change in circumstances. For example, failure to take medication, loss of accommodation, relationship breakdown, drug or alcohol misuse
High: there are identifiable indicators of serious harm. The potential event could happen at any time and the impact would be serious
Very high: there is an imminent risk of serious harm. The potential event is more likely than not to happen as soon as the opportunity arises and the impact would be serious. ’Opportunity’ can include the removal or overcoming of controls, and changes in circumstances.
Who is at risk?
It is important to identify the person or groups of people who are specifically at risk. Risk is categorised as risk to:
- the public: either generally or a specific group, for example, the elderly, vulnerable adults, women or an ethnic minority group
- a known adult, such as a previous victim or partner
- prisoners: within a custodial setting
- children: either specific children or children in general who may be vulnerable to harm of various kinds - this includes violent or sexual behaviour, emotional harm or neglect or because they are in custody
- staff: anyone working with the offender, whether from Probation, the Prison Service, police or other agency
- self: the possibility that the offender will commit suicide or self-harm.
What are risk and protective factors?
Risk factors are circumstances or characteristics that make criminal behaviour more likely. For example, poor temper control or an anti-social peer group. Protective factors make anti-social behaviour less likely. For example, supportive family or secure employment. Risk factors may be mitigated by protective factors. People who reoffend most often have risk factors across many areas. To reduce reoffending, we need to help people manage or overcome risk factors and develop or strengthen protective factors.
Risk assessment can also help target services on the right people
Rehabilitative intervention is most effective when it is proportionate to the likelihood of someone’s reoffending. We should focus more intensive help on those with the highest likelihood of reconviction. We should avoid directing limited resources to those unlikely to reoffend.
Alongside likelihood of reoffending, we sometimes want to estimate the risk of other future events, like absconding from prison.
We can measure likelihood of reoffending in several ways:
Actuarial or clinical? Actuarial assessments combine information about individuals in a structured way. This is weighted by the strength of their link with future behaviour, based on large scale research evidence. In HMPPS these are predictor tools such as the Offender Group Reconviction Sore (OGRS) or the Risk of Serious Recidivism (RSR). Clinical assessments rely on a practitioner’s judgement and experience to estimate the imminence of the offending behaviour. Some risk assessment tools fall in one or other category – others use elements of both.
Static or dynamic? Static risk factors are fixed or past elements that will not change. These include age, gender, current offence type or childhood experiences. Dynamic factors are associated with the changing likelihood of reoffending over time. These are factors that we can support service users with to help reduce the risks. Research shows there are nine issues commonly associated with offending behaviour:
- unstable accommodation
- a lack of employment
- no positive recreation activities
- poor personal relationships
- alcohol misuse
- drug misuse
- impulsivity and poor emotional control
- anti-social peers
- attitudes that support crime
These dynamic risk factors are also sometimes called criminogenic needs. Both clinical and actuarial assessments can draw on static and dynamic elements. For example, a history of truancy and a current problem with binge drinking.
Using risk assessments
The choice of actuarial or clinical risk assessment and static or dynamic is determined by the context and purpose of the risk assessment and will be included in the design of any risk assessment tools. An actuarial tool using only static data, can bring a swift, reliable estimate of the likelihood of reconviction. It also demands minimal staff resources. But this tells us little about that person’s strengths or areas of need that we can work on. For sentence planning or risk management, we need a further assessment using dynamic information to identify:
- areas for attention
- strengths to build on and
- priorities to agree
Clinical predictions of the likelihood of reoffending are less objective and less accurate. However clinical understanding of the relevance of risk factors is important for effective sentence planning and risk management.
Risk of what?
It is important that we are clear about the outcome that we are estimating.
Do we need to know whether the person is likely to commit any new offence over a particular period? Typically, we look at one year or two years following the start of a community sentence or release from prison.
Or do we want to estimate the risk that the potential reoffence might be for a violent crime or for any crime likely to cause serious harm?
The core risk assessment tools that HMPPS currently uses are:
Offender Group Reconviction Scale (OGRS3)
Percentage likelihood of committing any offence within 2 years leading to reconviction (proven reoffending). An OGRS3 score of 50% or more means that an offender is more likely than not to commit a proven re-offence within 2 years. OGRS scores can be used to target those resources designed to reduce reoffending. Accredited offending behaviour programmes often require particular OGRS scores as part of their eligibility criteria. There is good evidence that when these programmes are delivered to participants with too low or too high a risk they are much less likely to benefit.
OASys Violence Predictor (OVP)
Percentage likelihood of committing any violent proven re-offence within 2 years. This includes minor violent offences like common assault, harassment and criminal damage and more serious violent offences. An OVP score of 30%+ is the criterion for accredited programmes that address violent offending behaviour. The more intensive programmes specify an OVP score of 60% or above.
Risk of Serious Recidivism (RSR)
Percentage likelihood of committing a seriously harmful reoffence within two years. This is defined as an offence where the victim is killed or suffers trauma from which it will be hard or impossible to recover. RSR can form part of the decision to allocate individuals to National Probational Service (NPS) or Community Rehabilitation Company (CRC). Serious reoffending is rare. Fewer than 2% of the caseload commit a serious reoffence within two years of a release from prison or the start of a community order.
Risk of Serious Harm (RoSH)
A structured professional judgement assessment based on completion of OASys. Practitioners consider risk and protective factors alongside immediate situational and relational factors. People are allocated to a Risk of Serious Harm category from ‘low’ to ‘very high’. The RoSH assessment identifies whether the risk is to known adults, children, staff and/or the public. It also identifies whether the risk identified should be regarded as imminent. An individual who is rated as presenting a ‘high’ or ‘very high’ risk of serious harm is managed by the NPS. The core risk and need assessment tools are integrated into OASys. OGRS and RSR can be calculated outside of OASys assessment as they rely on a limited number of items available from the person’s records. OVP and RoSH are normally generated as part of the OASys process. They both draw on the standardised assessment of the more dynamic features of the person’s life that OASys provides. There are further tools for particular contexts. These include the Spousal Assault Risk Assessment Guide (SARA) for risk of partner violence, and the RM2000 assessment of the risk of further sexual offending.
What makes a good risk assessment tool?
It is important that the risk assessment tools we use are theoretically sound and provide reliable and valid estimates. Criteria for approving risk assessment tools is supported by advice from MoJ’s Correctional Services Accreditation and Advice Panel (CSAAP). A sufficiently robust risk tool will demonstrate the following components:
- clear description of the tool(s) and its fit with the overall approach to assessment
- sound theoretical underpinning and credible rationale
- evidence that the tool does what it aims to do
- commitment to ongoing research and validation to ensure ongoing fitness for purpose
- assessors using the tools are competent in its use
- use of the tool is implemented as intended
Managing risk and building hope – what next for assessment?
Our priority is to reduce reoffending and protect the public. Until recently, we have focused on the risk presented by individuals but in future will need to understand more about areas of strength and factors that support people to desist from crime. Focusing on negative labelling and stigmatisation following conviction can hinder desistance. Desistance is how people with a previous pattern of offending abstain from crime. An effective risk assessment system can help us do both.
We will continue to evaluate and learn from evidence, and to develop more effective risk assessment tools, to help individuals to reduce their reoffending and to lead better lives.
Prison, probation and rehabilitation: Public Protection Manual Guidance for prison and probation staff on managing offenders and protecting the public from harm.
The Risk Assessment Tools Evaluation Directory (RATED) provides a summary of the empirical evidence on each assessment tool included in the directory (The Risk Management Authority, Scotland).
- Research and analysis on the Offender Assessment System A compendium of research and analysis on the Offender Assessment System (OASys); studies completed between 2009 and 2013 (MoJ 2015)
This page summarises the available evidence base and is informed by independent academic peer review. It does not represent Ministry of Justice or Government policy.