Effective Approaches to Risk Assessment in Social Work: An International Literature Review




Within the criminal justice system, there is a heightened climate of fear amongst the general public as a result of media and political interest in crime, most notably in relation to youth crime, although it is often unsubstantiated, as Wood (2006: 319) suggests: 'the current climate of fear of crime… is known to far outweigh the actual experience of it'. Risk in this regard relates not only to the possibility of becoming embroiled in the criminal justice system but also to becoming a victim of crime. Risk broadly refers to an individual's chance of being exposed to harm. Little et al. (2004) note that the word 'chance' is the defining feature: children with multiple risk factors (truanting, single parent family, low achievers, minimal parental supervision, etc.) have a greater chance than their peers of exhibiting anti-social behaviour, but only 4 in 10 children with such risk factors will actually behave in an anti-social manner. The extent to which that ratio in youth crime may be exacerbated by public opinion and political rhetoric (arguably themselves both risk factors) is too broad an issue for this review, but suffice to say here that from an actuarial point of view, the risk of people becoming involved in crime (as offender or victim) is more prevalent in youth and reduces considerably into adulthood. In respect of young people, risk is no longer about dangerousness - which according to Brearley (1982) is a feared negative outcome of a hazard in the eye of the observer - but is about 'undesirable conduct' (Rose, 2000) and anti-social behaviour. Malloch (2002, 13-14) cites McConville et al, 1991 and others when suggesting: 'The uncertainty and fallibility of risk assessment in terms of the wider forms of behaviour management which professionals are expected to respond to can lead agencies to focus on particular social groups (the socially excluded) who become the "usual suspects"'. Risk assessment, rather than reducing pressure on the criminal justice system, can thus create a net widening effect.

There is also a strong correlation between age and crime for both offenders and victims. Since the vast majority of adolescent offenders do not become adult offenders, it is argued that they in particular should be dealt with as vulnerable first and foremost rather than as culpable ( SWSI, 2005).

However, youth crime apart, the focus of risk assessment and management is increasingly on serious and violent offenders in the UK and US in particular. The MacLean Committee on Serious Violent and Sexual Offenders was charged in 1999 with making proposals on the sentencing and future management and treatment of serious sexual and violent offenders in Scotland who may present a continuing risk to the public. The definition of risk assessment in relation to violence adopted by the Committee was 'the process of evaluating individuals to characterise the likelihood they will commit acts of violence'. The Committee recognised the interplay between psychological factors and situational factors and viewed risk assessment as an ongoing process which required constant review and modification, primarily to ensure public safety.

To be constantly reviewed and updated, risk assessments need to use consistent criteria over time. The MacLean Committee advised that 'it is in the interests of accuracy and efficiency if risk assessments are carried out using a common language, and sharing common information' (Scottish Executive, 2000). To this end, the Committee recommended the setting up of an independent body, the Risk Management Authority ( RMA), the aim of which was to 'ensure the effective assessment, management and minimisation of risk of serious violent and sexual offenders, and become a national centre for expert advice on offender risk assessment and management' ( RMA, 2006a: 1).

This chapter focuses on the legislation and guidance relating to criminal justice matters, predominantly serious violent and sexual offenders and the plethora of risk assessment tools built up around them. It also examines the notion of defensible decision making and the various practices and risk assessment techniques of the different agencies. It concludes that there is currently a preoccupation with risk assessment per se, possibly at the expense of effective interventions with offenders.


The Criminal Justice Act (2003) and Sex Offenders Act (2003) target dangerous offenders in England and Wales who are assessed as posing a high probability of causing harm. The Criminal Justice (Scotland) Act (2003) allows for serious violent and sexual offenders to be subject to a Risk Assessment Order ( RAO) and subsequently an Order of Lifelong Restriction ( OLR). The findings of the RAO can be challenged by an offender, who may also commission his/her own additional risk assessment. This provision goes somewhat against the grain of 'service user' participation and infers adversarial rather than consensual practice. Indeed, Irvine (2006) suggests that the sex offender registration and assessment procedures in Scotland are flawed because of the lack of any statutory requirement on the part of an offender to participate in the process of risk assessment. Nevertheless, the development of recent risk assessment tools and procedures encourages the participation of the offender in his/her ongoing risk assessment and management. For example, Asset - a tool devised specifically for young people - actively engages the young person in the risk assessment process, as does the Dynamic Supervision Project risk assessment tool for sex offenders. The development of accredited programmes both in prison and in the community also ensures that the offender is involved in an ongoing self-assessment of risk and in joint decision making in relation to risk management.

Multi-agency Public Protection Arrangements ( MAPPAs) have been put in place as a result of the Criminal Justice Act (2003) in England and Wales for serious and violent offenders, not only those convicted or imprisoned but also those pre-conviction. MAPPA offenders are categorised in England and Wales by offence rather than current circumstances. Offences are static factors and Wood (2006) has argued that MAPPAs can only be effective if they include a consideration of more dynamic risk factors as well. MAPPAs currently focus on an assessment of mental health or disability needs, offending history and housing, and risk categorisation should robustly reflect these factors. Such offenders are managed on a case conference basis and the arrangements apply to those liable to post-release supervision, those released without post-release requirements and those on temporary (home leave) release.

MAPPAs were introduced in Scotland in September 2006 and the key players are Scottish Prison Service, local authority social work departments and the police. Whereas England and Wales included non-convicted offenders, in Scotland only convicted sexual or violent offenders, registered sex offenders, those acquitted on grounds of insanity and others convicted of an offence likely to cause serious harm to the public are included in these new management arrangements. In Scotland, there have been three levels of risk management identified, mainly underscored by 'defensible decision making', as in England and Wales. Offenders may be moved between the three levels of management depending on circumstances, input and outcomes and should be given the lowest intervention consistent with providing a defensible risk management plan (Scottish Executive, 2006b). These levels are:

Level 1 - ordinary risk management - low or medium risk, mainly managed by one agency;

Level 2 - local inter-agency risk management - needing active involvement of more than one agency, with MAPPA meetings chaired by someone at service manager or chief inspector level;

Level 3 - MAPP Panels - for the 'critical few' high or very high risk offenders, chaired at senior manager or chief superintendent level. These panels require close cooperation at a senior level because of the risk involved or because of potential media and public interest in a specific case.

Following the Management of Offenders Act (2005) in Scotland, from April 2007, all registered sex offenders are subject to MAPPA requirements. Given the paucity of information on longer-term risks from sex offenders, not least adolescent sex offenders, sex offender registers are crucial sources of information for managing both pre- and post-conviction offenders in Scotland. Whilst under-16s cannot be placed on a sex offender register, the police can carry out assessments on them and impose restrictions on their activities via other court orders. However, there are concerns that given the emphasis on convicted sex offenders in Scotland, first offenders may not be seen as a priority. For example, Irvine (2006) notes that some 36 per cent of serious sex offenders have no previous similar convictions, and are therefore likely to be assessed as low risk. Equally, a focus on risk - past, present or future - can often be at the expense of addressing needs or providing specific interventions to address behaviour.

There are many risk assessment tools on the market in criminal justice and youth justice, although arguably less so for young people or women. In the adult criminal justice system (over the age of 16 in Scotland), the most commonly used tools are the Risk Assessment Guidance and Framework ( RAGF), the Offender Group Reconviction Score ( OGRS), the Level of Service Inventory - Revised ( LSI-R), Matrix 2000 and Tayprep. For young offenders, OASys, YLS and Asset are often used. McIvor and Kemshall (2002a) suggest several essential and desirable criteria for effective risk assessment tools in Scotland:

Essential criteria:

  • at least one peer reviewed publication on validation of the tool;
  • validation against a relevant population to the target group, ideally in Scotland;
  • based on actuarial and empirical factors contained in the research literature;
  • able to differentiate accurately between high, medium and low risk;
  • has inter-assessor and inter-rater reliability

Desirable criteria:

  • user-friendly;
  • resource lean;
  • 'easy' to train staff in its appropriate use;
  • process of use is transparent and accountable.

These authors found that tools were more likely to be used by social work practitioners if they were easy to administer and had good predictive power, but their weaknesses were reported to be their complexity, their lack of objectivity and lack of accuracy in determining the specific risks posed by serious violent and sexual offenders. Whilst time constraints in completing risk assessment tools in community care and child protection are problematic (see Chapters 3 and 4 respectively), criminal justice staff are less likely to have to undertake a risk assessments as a matter of urgency. In an evaluation of OASys, the tool often used in England and Wales by probation and prison workers, Mair et al. (2006) found that the main complaint by probation officers was the time required to complete the assessment form, often up to two hours. Under a half of the respondents felt it helped in risk assessment, but it was deemed inappropriate for low risk offenders, arguably the mainstay of probation practice. An evaluation of Asset, the tool used across the Youth Justice Teams (Yots) in England and Wales (Baker et al., 2003) found that it could predict reconviction over a 12 month period with 67 per cent accuracy. Asset is the first common, structured assessment tool to be used across the youth justice system, and this strengthens the statistics gained in the aggregation of data both locally and nationally.

The MAPPA Guidance which was revised in 2006 by the Home Office noted that decision making in the assessment of risk was not infallible but should nevertheless be underpinned by 'defensible decision making':

In place of infallibility, we must put defensibility - making the most reasonable decisions and carrying them out professionally in a way which can be seen to be reasonable and professional (Home Office, 2006: 5).

The revised Guidance notes that such decision making should not be defensive but be based on 'rigour and risk management with robustness' (ibid: 6). Kemshall (2003) identifies the following criteria for defensible decision making:

  • all reasonable steps are taken;
  • reliable assessment methods are used;
  • information is collected and thoroughly evaluated;
  • decisions are recorded and carried through;
  • agency processes and procedures are followed;
  • practitioners and managers are investigative and proactive.

McIvor and Kemshall (2002a) note that the use of risk assessment tools is primarily to ensure 'defensible decisions', which is quoted in a Scottish Social Work Services Inspectorate manual (undated) as when 'a responsible body of co-professionals would have made the same decision'. Defensible decisions were noted by McIvor and Kemshall (2002a) as a driving force in certain workers' choice of risk assessment tool: hence the preoccupation with accreditation and validity of certain tools (McIvor and Kemshall, 2002a). The MAPPA guidance described above specifically mentions defensible decision making as a priority for resource allocation:

"The risk management structure is based on the principle that cases should be managed at the lowest level consistent with providing a defensible risk management plan. The level at which a case is managed is therefore dependent upon the nature of the risk and how it can be managed." (Scottish Executive, 2006b, para 47).


"Opinions were divided on whether it was feasible or desirable to implement a common approach to risk assessment across different disciplinary groups, though a greater degree of consistency was considered desirable." (McIvor and Kemshall, 2002b).

As the above quotation exemplifies, McIvor and Kemshall (2002b) suggest that risk assessment and management procedures are usually context and agency specific and are therefore less likely to be transferable to other situations or settings. According to the Expert Panel on Sex Offending (Scottish Executive, 2001), difficulties and confusion arise when different agencies use different tools at different times in the process of assessing risk for different purposes. Such assessments should be shared between agencies and risk assessments need to be ongoing and constantly updated using a structured clinical approach. MAPPAs have ensured that this is now the case. Not only does the duty to collaborate create a climate for consistency and information sharing across the key agencies, but the development of static and dynamic risk assessment tools for use across Scotland's local authorities and other agencies will ensure a common usage of risk assessment tools and continuity between prison-based and community-based interventions.

However, few tools are available for use specifically with certain heterogeneous groups, for example, women, young people and mentally disordered offenders and risk assessment procedures within and between local authorities in Scotland still vary in practice. Prison staff, for example, are more concerned with risk of violence within the prison, and also prefer validated tools, whilst the police, like social work, tend to use non-validated tools and ones with little proven scientific accuracy.

The Assessment, Intervention and Moving On ( AIM) risk assessment tool provides guidelines for practitioners from a wide range of agencies and disciplines, including the police, social services, probation, education and health). It purports to give a common language and shared approach to sexual offending (Print, Morrison & Henniker, 2000). Equally, in Scotland, the police in particular want greater uniformity and structure to risk assessment nationally, including the use of one specific tool throughout all forces (McIvor and Kemshall, 2002a). Inter-agency work was found in McIvor and Kemshall's study to be less likely to be effective if there was a failure to share information or because of the use of a variety of risk assessment approaches and interpretations. These authors also note that some agencies will go to arbitration to decide on risk where collaborating agencies disagree on the level of risk posed or its management, although this level of disagreement was rare. Some respondents in their study suggested that disagreements were actually productive, in that they resulted in more information being gathered and more opinions being formed.


The 'language of risk' (Horsefield, 2003: 376) masks the social and personal problems facing offenders and sees victims as the main consumers of risk assessments. Horsefield cites O'Malley (2001) who suggests that in the US at least, exclusion and separation has taken over from punishment in the name of risk management. Risk assessments may appear scientific, accurate and effective, but only really serve to give credence to an organisation in responding to crime, regulating its staff, and limiting its liability when things go wrong. Risk assessment has also been described as creating 'automated environments' which devalue personal relationships and the need for trust, and hence result in further exclusion of offenders (Hayles, 2006). Hayles argues that basing punishment on risk lengthens that punishment indeterminately - in other words, 'once a risk, always a risk'. She also argues that risk assessment tools 'label' for longer periods than is perhaps warranted by the type and severity of the offence. Reducing offending should be about offering alternative constructive activities and social bonds, rather than about reducing negative outcomes and attitudes via cognitive behavioural approaches (Barry, 2006; Hayles, 2006; Maruna, 2001).

Webster et al. (2006) note that the static risk factors usually associated with offending by young people are also prevalent amongst law-abiding populations. These authors equate so-called risk factors such as truanting, single parenthood, educational low-achievement and disruptive childhoods with poverty rather than criminality:

"the narrowing down of risk factors to the family, parenting, truancy and peer-groups, reflects more a process of political expediency… than any genuine attempt to understand the causes of criminality." (Webster et al., 2006: 12).

Webster et al. also draw attention to the 'false negatives' of young people who show all the predictive signs of offending but remain law-abiding and suggest that risk assessment may be a reactive process of containment rather than a proactive means of resolving structural constraints for many disadvantaged young people:

"Much of what happened in the lives of our informants could not have been predicted from earlier experiences… In isolating individual risk factors from their context in biography, place and social structure, such [risk assessment] devices offer ways of managing offenders rather than addressing the causes and cessation of individual offending." (ibid: 18).

Whilst in the recent past, there was an over-reliance on actuarial assessment based on static factors (e.g., previous criminal convictions) which inhibited rehabilitation in favour of management of 'groups' of offenders, introducing dynamic factors (e.g., employment status) into the equation now allows for more constructive work with (rather than to) offenders. Pitts (2001) quotes Farrington (2000: 7) in suggesting that the "risk factor prevention paradigm" is 'easy to understand and to communicate, and it is readily accepted by policy makers, practitioners and the general public'. However, it has its limitations in that it cannot readily separate out cause from effect: homelessness, unemployment and low income, for example, can be both cause and effect of criminal involvement. Brown (2005) also criticises the literature on static risk factors which are deemed 'modifiable' by the individual rather than by society, notably those targeted by Farrington (2002):

"Perhaps the most unnerving aspect of the whole approach is that as Farrington comments, 'prevention requires change within individuals' (p. 661). Since the vast majority of 'risk factors' appear to be beyond the individual's control (low family income, family breakdown, single parenthood, low IQ, etc.) this is puzzling. Presumably the idea is to change an individual's response to their life circumstances, not the life circumstances themselves. In other words, offending is conveniently individuated and is not a matter for social justice or social inclusion." (Brown, 2005: 101).


This chapter has looked at the criminal justice field, the field possibly most influenced by the media and the wider public in respect of dangerousness of offenders and the vulnerability of [potential] victims. It is thus not surprising that the main focus of risk assessment in criminal justice is in relation to violent and sexual offenders, around which a centre of excellence in criminal justice risk management has been set up and a myriad of legislation and guidance has been developed. The criminal justice system also arguably houses the most risk assessment tools, which has resulted in a preoccupation with how to measure risk factors perhaps at the expense of what exactly one is measuring, to what end, and whether other forms of intervention may not prove more effective in reducing longer-term offending. The focus currently in criminal justice seems to be on managing risk rather than alleviating other problems in offenders' lives that might influence their behaviour. Thus, supervising social workers may resort, under advice from the guidance, merely to making defensible decisions and offending becomes a matter of containment rather than resolution. Guidance apart, however, there is little consistency in practice in criminal justice although it is the field most advanced in terms of inter-agency collaboration. However, the culture of the various organisations (e.g., the police versus social work) is doubtless an influence on the way risk is perceived and managed within each agency, making the gains from collaboration less obvious to identify and more difficult to achieve.