We are testing a new beta website for gov.scot go to new site

21st Century Social Work: Reducing Re-offending: Key Practice Skills


21 st Century Social Work: Reducing Re-offending: Key Practice Skills


This literature review was commissioned by the Scottish Executive's Social Work Services Inspectorate in order to support the work of the 21 st Century Social Work Review Group. Discussions in relation to the future arrangements for criminal justice social work raised issues about what skills were needed to supervise offenders effectively in the community. Did it, for example, require the skills of prison officers to ensure compliance and 'manage' offenders; or the skills of police officers to rigorously monitor the activities of offenders; what was it that criminal justice social workers did? Given that the current focusing of attention on reducing re-offending, it seemed worthwhile to start not with what we know, or think we know, about the remit and skills of the individual staff groups, but to consider instead the research evidence about the ingredients of effective interventions to reduce offending and the skills required to deliver this. We hope that this will allow us to look afresh at what this might tell us about which disciplines might best encompass the requisite skills for reducing re-offending in the community.

To this end, we agreed to begin with the intended outcomes of work with offenders and seek to work our way back to the skills required to promote this outcome, rather than starting with current statements about or understandings of the 'skill sets' of those professionals currently involved in such interventions. Thus, we make no reference to the existing Standards in Social Work Education and their delineation of the knowledge, values and skills required for competent and confident professional practice, 1 nor to the curricula of probation studies courses in other jurisdictions.

The structure of this report is as follows. In section 2, we begin by providing an analysis of the intended outcomes of community-based criminal justice interventions with offenders, noting that in recent years the primary intended outcome of such interventions in Scotland has become reducing re-offending; an objective that is essentially concerned with the achievement of positive change in the lives of offenders. While the narrow focus implied by this objective is on a change in individual behaviour, it is necessary to locate this focus within a broader understanding of that change process within its personal and social context. To this end, in section 3 we set the context of change efforts by reviewing evidence about the profiles of offenders under supervision in terms of their needs, deeds and characteristics; this provides a picture of the starting point for change. In section 4, we review the evidence about the nature of the process of desistance from offending, given that developing a clear understanding of this change process is critical to thinking about how best to support it. In section 5, we examine evidence from the research literatures about 'what works' to reduce re-offending. However, in order to address criticisms about the narrow scope of this research, we also explore the psychotherapy and counselling literatures for evidence about how positive personal change in general is best supported by professional helpers. Finally, we look again at the desistance literature to find out what offenders themselves say about how desistance is best supported. In section 6, we use this accumulated evidence to suggest four key skill sets required to support reducing re-offending (building relationships that support change; assessing risks, needs and strengths; research-based planning and delivery of change programmes; and managing change) before presenting the conclusions of our review.


A detailed, critical and historically contextualised account of the intended outcomes of community based criminal justice interventions in Scotland is beyond the scope of this paper. 2 For our purposes, it is sufficient to note that the National Objectives and Standards for Social Work Services in the Criminal Justice System (hereafter 'NOS') were published in 1991 3 and that section 12 of that document provides a list of the initial objectives for these services. The first objective delineated in the NOS was 'to enable a reduction in the incidence of custody… where it is used for lack of a suitable, available community based social work disposal'. 4 However, a concurrent focus on reducing re-offending, informed from the outset by emerging research evidence, 5 was nonetheless seen as being critical to the enhanced credibility of community penalties on which reduction in the use of custody was thought to depend. 6

The objectives of the NOS have been recently updated in 'Criminal Justice Social Work Services National Priorities for 2001-2002 and onwards'. 7 Though this document signals no dramatic changes in the philosophy or policy direction of the NOS, it places a higher priority on public protection and developing systems for managing the risks posed by 'dangerous' offenders. Thus, the first of the three priorities mentioned in the 'National Agenda' is to make a 'Contribution to increased community safety and public protection'. 8 That said, the second priority echoes the original NOS in aspiring to 'Reduce the use of unnecessary custody by providing effective community disposals'. 9 Unlike in England and Wales, community disposals in Scotland are not cast as 'punishment in the community'; rather, the emphasis is on rehabilitation. Thus, the third priority is to 'Promote the social inclusion of offenders through rehabilitation, so reducing the level of offending'. 10 In line with the philosophy underpinning the Social Work (Scotland) 1968 Act (under which offender services became part of generic social work departments), this recognises both the intrinsic worth of promoting the social inclusion of offenders and its instrumentality in reducing offending.

It is helpful to conceptualise these three intended outcomes (public protection through reduced re-offending, reducing the unnecessary use of custody and the social inclusion of rehabilitated offenders) as the three points of a triangle (see Figure 1 below). Although protecting the public by reducing re-offending has moved to the apex of the triangle as the super-ordinate purpose, Scottish policy and practice continues to recognise the interdependence of the three priorities. This is why, for example, the Executive's recently published document, 'Supporting Safer, Stronger Communities: Scotland's Criminal Justice Plan', 11 recognises that short-term prison sentences which fail to rehabilitate prisoners are counter-productive; indeed, in often making prisoners' situations and problems worse, they increase social exclusion and with it the likelihood of re-offending. Put another way, they fail to protect the public.

Figure 1: The inter-dependence of the intended outcomes of CJSW

Figure 1: The inter-dependence of the intended outcomes of CJSW


In order to think clearly about how the outcomes delineated above might be most effectively pursued, it is first necessary to think about the starting point for any effort to support or bring about change. To this end, in this section we provide a brief review of some evidence about the needs, deeds and characteristics of those subject to probation and CJSW supervision.

According to the most recent criminal justice social work statistics, 12 a total of 7,359 community service orders (including 2,393 probation orders with a requirement of unpaid work) were made in Scotland in 2002-03. This represents an average incidence of 20.4 per 10,000 population. Community Service Orders (CSOs) were most common amongst young offenders, with 77.7 orders per 10,000 population for 18 to 20 year olds and 59.5 orders per 10,000 population for 21 to 25 year olds. Ninety percent of CSOs were given to male offenders, 88 percent to white offenders, and 63 percent to offenders who were unemployed or not seeking employment. The average length of a CSO was 155 hours.

Probation statistics show a similar pattern. A total of 7,417 probation orders were made in Scotland in 2002-03. This corresponds to 20.5 per 10,000 population. Average incidence was highest amongst 18 to 20 year olds (76.8 per 10,000 population), but was also common amongst 21 to 25 year olds (56.3) and 16 to 17 year olds (53.7). Male offenders accounted for 81 percent of probation orders and white offenders for 88 percent. The proportion of probation orders made in respect of offenders who were unemployed or not seeking work was 76 percent. Courts were most likely to make probation orders lasting between 12 and 17 months long (51%), followed by orders lasting 18 to 23 months (20%) and two years or more (20%). The minimum of six months and the maximum of three years were used comparatively rarely.

As regards post release supervision, a total of 457 parole licences, 262 non parole licences, 58 extended sentences, 169 supervised release orders, 112 life licences and 57 other forms of supervision was reported in 2002-03. A total of 1,849 referrals for diversion from prosecution was also recorded, resulting in 1,723 assessments and 1,015 new cases.

More detailed information on the needs, deeds and demographic characteristics of offenders subject to social work supervision can be found in recent studies of probationers and parolees. One of the best known such studies is Mair and May's (1997) Offenders on Probation. 13 Mair and May's sample of 1,986 offenders was generally representative of all persons on probation in England and Wales; for example, 82 percent of their sample were male and just over 40 percent fell into the 16 to 24 age range. In terms of ethnic origin, 93 percent were white and, of the rest, most defined themselves as black (5%). Only one in five probationers was employed or self-employed (similar to the 1 in 4 figure in Scotland reported above) and, of those who were employed, 79 percent were in manual occupations. About a third of female probationers were living alone with dependent children and state benefits were the main source of household income for two-thirds of all respondents. About 40 percent had had problems relating to debt.

Also of significance, a sizeable minority of the sample had spent time in some form of local authority care as a child; probationers were 10 times more likely to have been in care than the general population. Of the probationers in the study, 19 percent had lived in a children's home and 14 percent in a borstal or young offenders unit. A similar proportion had been bought up in a one-parent family. Forty-two percent of the sample had left school before the age of 16, while 49 percent left at age 16. Only 12 percent of male probationers had qualifications at 'O' level or equivalent. Forty-nine percent of the sample had health problems or disabilities and 30 percent said that this restricted the work that they could do. Forty-eight percent of offenders admitted some form of illicit drug use in the past 12 months: 42 percent had taken cannabis, 24 percent had taken amphetamines, followed by temazepam (15%), LSD (14%) and ecstasy (12%). Offenders in the 21 to 35 year age group were most likely to have used crack (6% compared to 5% of the overall sample), heroin (12% vs. 8%) and methadone (10% vs. 8%), while women were less likely to have taken drugs than men (34% and 50% respectively). Ten percent of those in the overall sample were identified as having a problem relating to their alcohol consumption.

In terms of their current offence, male probationers were most likely to have been convicted of burglary (23%), violence (19%), drink driving (14%), other driving offences (14%), other theft/handling offences (13%) and theft of, and from, a motor vehicle (12%). Women were most likely to have been convicted of other theft/handling offences (35%), fraud, forgery and deception (29%) and violence (16%). Four-fifths of the sample had previous criminal convictions, with males more likely than females to have become involved in crime before the age of 18 (76% and 59% respectively).

Scottish research studies reveal similar patterns. McIvor and Barry's (1998a) study involved a sample of 155 probationers drawn from four study areas across Scotland. 14 Eighty-two percent were male and 48 percent were between 16 and 20 years of age. Eighty-six percent had one or more previous conviction (the average number of previous convictions was 8.1) and 30 percent had served at least one previous custodial sentence. Most received their current probation order for offences involving dishonesty (68%). Offenders in their throughcare sample 15 tended to be older - the majority were aged 21 or over - and their criminal histories were more significant. Seventy-nine percent had previous convictions and 55 percent had previously served a custodial sentence. Only two of the 60 ex-prisoners were women.

Problems relating to family relationships, drug and alcohol abuse featured prominently in probationers' Social Enquiry Reports (SERs) and workers most often attributed offending to alcohol or drug abuse. That said, there were some important differences identified between different types of offender. Compared with male probationers, for example, women's offending was more likely to be described as financially motivated or a response to emotional stress, while young offenders were more often described as impulsive or opportunistic and as influenced by negative peer pressure. The reasons put forward by social workers when recommending probation to the courts were: the potential offered by probation to address or monitor offending (or behaviour associated with offending, including drug and/or alcohol abuse), the potential to provide help with practical problems such as employment or education, support of a more general kind (e.g. help with relationship problems, social skills etc.), and to build upon the offender's motivation to change.

A more recent analysis of the criminogenic needs of offenders sentenced to community disposals and custodial disposals can be found in Harper et al. (2004). 16 Drawing on recent data from the Offender Assessment System (OASys) in England and Wales, Harper and her colleagues reported that, on average, offenders are assessed as having four criminogenic needs. 17 Although offenders in custody generally have a greater number of needs than offenders in the community, the similarities between the two populations are striking. Their findings are reproduced in the table below (Table 1). 18

Table 1: Factors associated with offending

Percentage of offenders assessed as having a problem

Section of OASys

Community sentences

Custodial sentences

1&2 Offending information*



3 Accommodation



4 Education, Training and Employment



5 Financial management and income



6 Relationships



7 Lifestyle and associates



8 Drug misuse



9 Alcohol misuse



10 Emotional well-being



11 Thinking and behaviour



12 Attitudes**



No. of criminogenic needs



No. of criminogenic needs excluding sections 1&2



* Offending information includes the current offence and criminal history.

** The percentages with attitudes needs are likely to rise when an amendment is made to the OASys scoring system, effective from early 2005.

Source: Harper, G. et al. (2004: X)

Of course, whilst it is important to acknowledge the common characteristics of the offender population as a whole, consideration must also be given to variations in the needs, deeds and characteristics of different 'types' of offenders and of offenders subject to different court disposals. The literature on women who offend, for example, shows that while male and female offenders share a set of universal needs, there are also key differences in terms of behavioural issues, domestic expectations and risk factors. 19 The dangers of over-generalising the needs and deeds of different offender groups are highlighted by the findings of one recent Scottish study, which focussed on persistent young offenders in one locality: 20

'…"persistent" offenders in our sample were more likely [than other offenders] to have begun offending at an earlier age, had committed a greater number of offences and wider range of offence types, had more pronounced, poly-substance misuse problems, came from families with several (and more serious) risk factors, associated with persistent (as opposed to merely delinquent) peers, had lower levels of educational attainment and higher levels of impulsiveness, aggression, truancy, absconding and social exclusion… [That said,] There were some important variations within the sample and, more importantly, significant exceptions to the rule. For example, a sizeable minority of the young people appeared to experience a sudden onset of frequent and/or serious offending over the age of 14; came from relatively affluent or stable family backgrounds; and had no social work involvement prior to their offending behaviour. This not only raises important questions about our ability to identify persistent offenders in advance, but also about how to target effective youth justice interventions. The central theme from the local study is one of significant diversity even within predictable commonalities. In some respects this is the inevitable challenge that faces local managers and practitioners in responding to youth crime and persistent offending in an evidence-based manner; they have to interrogate and interpret necessarily generalised research messages about the needs, deeds and characteristics of the populations with whom they are seeking to effect change in the light of often imperfect local data. The range, severity and diversity of the difficulties experienced by the young people even in our small local sample should caution us about the search for generalisations about "remedies"'. 21

In summary, then, offenders subject to community supervision have very high levels of need. This means that, in addition to addressing their client's offending behaviour, practitioners often have to deal with problems relating to poor parenting, abuse, neglect, and damaged relationships, criminal and anti-social peers, low educational attainment, substance abuse or dependency, high levels of impulsiveness and aggressiveness, poverty, poor housing and/or homelessness. 22 Irrespective of how these difficulties are related to offending behaviour, they have important implications for supervision and the skills required to prompt and support reductions in re-offending.


Given that desistance from offending is the outcome to which CJSW services are directed, the fact that the desistance research has had, until recently, a muted impact on policy and practice might seem like something of a paradox. Part of the reason for this apparent paradox is that, until recently, the desistance literature has tended to address 'the wider social processes by which people themselves come to stop offending'. 23 Thus, it has not been necessarily or indeed primarily a literature about criminal justice interventions. Indeed, Maruna points out that some of the research focuses on 'spontaneous desistance', which is achieved without 'assistance' from or through the criminal justice system. 24 However, he also argues that, in theory and in practice, the boundaries between 'natural' processes of desistance and rehabilitative interventions blur. Processes of change towards desistance may share similarities whether they are spontaneous or professionally assisted. Knowledge about such processes therefore becomes critical to our understandings not just of 'what works' in terms of interventions but also of how and why ex-offenders come to change their behaviours. This section therefore aims to build an understanding of the human processes and social contexts within which rehabilitative interventions are (or should be) 'embedded'. It begins with a brief review of related theoretical perspectives before exploring some of the relationships between desistance, age and gender and motivation and attitudes.

Theoretical Perspectives

Maruna identifies three broad theoretical perspectives in the desistance literature: maturational reform, social bonds theory and narrative theory. 25 Maturational reform theories have the longest history and are based on the established links between age and certain criminal behaviours, particularly street crime. Social bonds theories suggest that ties to family, employment or educational programmes in early adulthood explain changes in criminal behaviour across the life course. Where these ties exist, they create a stake in conformity, a reason to 'go straight'. Where they are absent, people who offend have less to lose from continuing to offend. Narrative theories have emerged from more qualitative research which stresses the significance of subjective changes in the person's sense of self and identity, reflected in changing motivations, greater concern for others and more consideration of the future.

Maruna argues that none of the more general desistance theories has offered much specific assistance to practitioners as to what they should actually do to encourage change. However, the same problem of seeking to find ways to interpret and use research in driving policy and practice arises in connection with the better-known 'what works' research concerning effective rehabilitative programmes (reviewed at 5.2 below), since:

'…such research tells us little about individual differences among client experiences in the process… Every individual encounters and interprets unique social interactions within a program setting… every intervention consists of thousands of different micro-mechanisms of change… By concentrating almost exclusively on the question of "what works", offender rehabilitation research has largely ignored questions about how rehabilitation works, why it works with some clients and why it fails with others'. 26

Maruna argues that desistance research can and should redress these deficits in the 'what works' research by identifying processes of reform and helping in the design of interventions that can enhance or complement spontaneous change efforts. Such research could assist practitioners with the challenge of helping people progress towards the point of being ready and able to make changes, perhaps by accelerating processes that appear to have slowed or stalled for a variety of reasons. Recognising the limitations of each form of research (desistance and rehabilitation) on its own, Maruna proposes a marriage of the two; with the desistance research's focus on the success stories of those that desist offering an 'individual-level view' that, in partnership with the rehabilitation literature's identification of general practices that seem successful, can better inform understandings of the change processes involved.

The desistance research studies reviewed below begin to explore some key aspects of this 'individual-level view' and, in so doing, appear to have much to offer the current study of key practice skills to support reducing re-offending.

Age, Gender and Desistance

In a recent Scottish study, 27 Jamieson, McIvor and Murray explored desistance and persistence amongst three groups of young people aged 14-15 (the peak age for recruitment into offending for boys), 18-19 (the peak age of offending) and 22-25 (the age by which many would be expected to grow out of crime). They paid particular attention to gender differences in their study, which was based on interviews with a total of 75 'desisters' (43 male and 32 female) and 109 young people (59 male and 50 female) who were still offending or had done so recently.

The researchers discovered some significant age related differences concerning desistance. In the youngest age group, desistance for both boys and girls was associated with the real or potential consequences of offending and with growing recognition that offending was pointless or wrong. Young people in the middle age group similarly related their changing behaviour to increasing maturity. This was often linked to the transition to adulthood and related events like securing a job or a place at college or university, or entering into a relationship with a partner or leaving home. For the oldest group, 'desistance was encouraged by the assumption of family responsibilities, especially among young women, or by a conscious lifestyle change'. 28

Although age and, in particular, the transitions associated with it, seemed to be a more important determinant of desistance than gender, McIvor et al did note some gender differences. The young women in their sample tended to offer moral as opposed to utilitarian rationales for stopping offending and were more likely to emphasise the importance of relational aspects of the process. 29 Some young women linked their decisions to desist to the assumption of parental responsibilities. In general, young men focussed more on personal choice and agency. Amongst persisters, girls and young women were more often keen to be seen as desisters, perhaps reflecting societal disapproval of female offending. McIvor et al speculate that:

'Assigning the offending to the past rather than acknowledging it as a current or future reality may enable young women to better cope with the tensions that may arise when, on the one hand, society encourages gender equality and, on the other, continues to double condemn young women who step beyond their traditional gender roles'. 30

In a more recent article, Rumgay has suggested that women's desistance from crime is best understood as a process initiated by the perception of an opportunity to claim a pro-social identity during a period of readiness to reform, which is subsequently sustained by the deployment of strategies of resilience and survival in conditions of adversity. 31

The development of this 'readiness to reform' seems to be slower, in general, for young men. Graham and Bowling's earlier study 32 of young people aged 14-25 noted a clear association between the life transition from adolescence to adulthood and desistance from offending among young women. Young men, in contrast, were less likely to achieve independence and those that did leave home, formed partnerships and had children, were no more likely to desist than those that did not. Failure to desist among young men seemed to be best explained by three sets of risk factors: a high frequency of prior offending, continued contact with delinquent peers and heavy drinking and controlled drug use. Graham and Bowling speculate that life transitions,

'only provide opportunities for change to occur; its realisation is mediated by individual contingencies. Males may be less inclined [than females] to grasp or be able to take advantage of such opportunities…' 33

More recent studies have revised this conclusion to some extent; suggesting that similar processes of change do indeed occur for (some) males but that they seem to take longer to 'kick-in'; the assumption of responsibilities (for example, in and through intimate relationships and employment) does make a difference but this difference is more notable in men aged 25 and over. 34 Thus, it seems that young men take longer to perceive and to grasp the opportunities for change that these life transitions provide.

In Graham and Bowling's study, only two factors seemed to be positively associated with desistance for males in the 16-25 age range: firstly, their perception that their school work was above average, and, secondly, continuing to live at home. It may be that continuing to live at home is associated with desistance because of relatively positive relationships with parents and, as a result, spending less time with delinquent peers.

Attitudes, Motivation and Desistance

Returning to McIvor, Murray and Jamieson's study, it is also important to note that 'persisters' in the youngest age group were more optimistic about their ability to desist whereas:

'for older respondents, who may have become more entrenched in patterns of offending and drug use, desistance was rarely considered to be an immediate or achievable goal'. 35

The significance of this finding is underlined by Burnett's study of efforts to desist amongst 130 adult property offenders released from custody. 36 She noted that whilst most, when interviewed pre-release, wanted to 'go straight' (eight out of 10), six out of 10 subsequently reported re-offending post-release. Burnett noted that, for many, the intention to be law-abiding was provisional in the sense that it did not represent a confident prediction; only one in four reported that they would definitely be able to desist. Importantly, Burnett discovered that those who were most confident and optimistic about desisting had greatest success in doing so. For the others, the 'provisional nature of intentions reflected social difficulties and personal problems that the men faced'. 37

On the basis of her interviews, Burnett delineates three categories of persisters, though she notes that these categories are neither fixed nor mutually exclusive. 'Hedonists' were attracted by the feelings of well-being gained through criminal involvement, whether in terms of the 'buzz' at the time, the emotional high afterwards or the place of the financial rewards of crime in funding lifestyles sometimes associated with alcohol and drugs. The 'earners' varied in their enthusiasm for crime, but regarded it as a viable money making enterprise. The 'survivors' were generally dependent on substances and unhappily committed to persistent property offending to fund their substance misuse.

The desisters also fell into three broad categories. The 'non-starters' adamantly denied that they were 'real criminals' and, in fact, had fewer previous convictions than the others. For the 'avoiders', keeping out of prison was the key issue. They appeared to have decided that the costs of crime outweighed the benefits. The 'converts', however, were:

'the most resolute and certain among the desisters. They had found new interests that were all-preoccupying and overturned their value system: a partner, a child, a good job, a new vocation. These were attainments that they were not prepared to jeopardize or which over-rode any interest in or need for property crime'. 38

Burnett notes that simply classifying the men as persisters or desisters 'misrepresents the switching, vacillating nature of desisting from offending'. 39 Most were ambivalent towards crime and, in consequence, desisting seemed like a protracted 'back and forth' or 'zigzag' process.

Bringing these studies together, the research on factors associated with desistance is neatly summarised in Farrall's account, 40 which stresses the significance of the relationships between what we might term 'objective' changes in the offender's life and his or her 'subjective' assessment of the value or significance of these changes:

'… the desistance literature has pointed to a range of factors associated with the ending of active involvement in offending. Most of these factors are related to acquiring "something" (most commonly employment, a life partner or a family) which the desister values in some way and which initiates a re-evaluation of his or her own life…' 41

In some senses, this characterisation of desistance marries the three theoretical perspectives above; desistance resides somewhere in the interfaces between developing personal maturity, changing social bonds associated with certain life transitions, and the individual subjective narrative constructions which offenders build around these key events and changes. It is not just the events and changes that matter; it is what these events and changes mean to the people involved.

Desistance, choice and control

Maruna's recent study 42 explores the subjective dimensions of change by comparing the narrative 'scripts' of 20 persisters and 30 desisters who shared similar criminogenic traits and backgrounds and who lived in similarly criminogenic environments. In the 'condemnation script' that emerged from the persisters,

'The condemned person is the narrator (although he or she reserves plenty of blame for society as well). Active offenders… largely saw their life scripts as having been written for them a long time ago'. 43

By contrast, the accounts of the desisters revealed a different narrative:

'The redemption script begins by establishing the goodness and conventionality of the narrator - a victim of society who gets involved with crime and drugs to achieve some sort of power over otherwise bleak circumstances. This deviance eventually becomes its own trap, however, as the narrator becomes ensnared in the vicious cycle of crime and imprisonment. Yet, with the help of some outside force, someone who "believed in" the ex-offender, the narrator is able to accomplish what he or she was "always meant to do". Newly empowered, he or she now seeks "give something back" to society as a display of gratitude'. 44

The desisters and the persisters shared the same sense of fatalism in their retrospective accounts of the development of their criminal careers, thus minimising their personal accountability for their pasts. Maruna reads the desisters' retention of fatalistic accounts of their criminal pasts as evidence of the conventionality of their values and aspirations and of the need to believe in the essential goodness of the 'real me'. However, in their accounts of achieving change and in their discussions of their future prospects, Maruna's findings may suggest that desisters have to 'discover' their ability to make choices and to manage their own lives in order to resist and overcome the social pressures that bear down upon them.

The practical implications of these findings are explored in a number of recent research studies on the relationships between intervention and desistance. In particular, the research offers important insights into the ways in which practitioners can assist offenders to construct and navigate a pathway to desistance. Before we review those studies, however, we look first to the related literatures about 'what works' in criminal justice interventions.


In this section, we move on to consider the evidence about how professional interventions might best support the reduction of re-offending. The first part of the section provides an account of the evidence about the kinds of interventions and approaches that seem to work most effectively to reduce re-offending. However, in the light of emerging evidence about the need to look beyond programmes in order to develop effective practice more generally, in the second part of the section we review the relevance and usefulness of the broad literature about the effectiveness of psychotherapy and counselling in supporting personal change. The third part of the section looks at some particular approaches and techniques that seem to draw effectively on these two literatures. In the last part of this section, we return to the broader literature about how the process of desistance can be best supported drawing its evidence from what offenders themselves say about the role that interventions can play in that process.

5.1 Supporting the reduction of re-offending

After a period of considerable pessimism about the prospects for rehabilitative interventions with offenders, the late 1980s and 1990s produced renewed optimism about 'treatment' or rehabilitation, under the cumulative influence of a series of literature reviews 45 supported in turn by the application of new statistical techniques in 'meta-analysis'. A decade ago, McGuire and Priestley, 46 reviewing Andrews' work 47 and the findings of other meta-analyses, provided a seminal summary of six key principles for the design and assembly of effective programmes 48 -

  1. Risk Classification: the level of service provided should match the level of risk assessed; higher-risk individuals should receive higher levels of intervention - lower risk, lower levels of intervention. This principle should also stress minimal intrusiveness.
  2. Criminogenic Needs: only some needs contribute to or are supportive of offending - the focus of intervention should be on addressing offending by alleviating those needs that are criminogenic. This principle underlies direct work on offending behaviour. The most promising targets of intervention (according to Andrews 1995) are 'anti-social' attitudes, habitual patterns of thought and feeling associated with criminality, personal control issues and peer associations, developing social skills, and promoting family support.
  3. Responsivity: learning styles vary - approaches that match worker and client learning styles work best. In general, it seems that offenders require active and participative rather than didactic (lecturing), unstructured or experiential methods. In general, the most effective programmes seem to use behavioural or social learning principles, and to include a cognitive component challenging attitudes, values and beliefs.
  4. Community Base: programmes in the community fare better than those in institutions, perhaps because they facilitate real-life learning in the home environment. The effects of institutional programmes tend to 'wash out' in the 'real world'.
  5. Treatment Modality: effective interventions recognise the variety of offenders' problems, they employ a skills-oriented approach, and they use methods drawn from behavioural, cognitive, or cognitive-behavioural sources.
  6. Programme Integrity: effective intervention connects the methods used to the aims stated; is carried out by appropriately trained and supported staff; is adequately resourced; and plans monitoring and evaluation from the outset. The plan is implemented with integrity, and those involved in its initiation are somehow involved throughout.

Attempts to engineer these principles into practice have, to date, focussed in most countries on developing systems of accreditation, to ensure that programmes are designed in a way that matches the level and content of the intervention to the level of risk and need. They have also endeavoured to ensure that programmes are also delivered to a high standard to ensure that they have the best chance of achieving the intended outcome. Criteria for accreditation are build around the key principles for effectiveness, setting out specific requirements to ensure that programmes are theoretically well grounded, designed for a clearly identified target group, use a range of appropriate methods and are properly integrated with other aspects of supervision.

Some commentators have argued that the emphasis on the role of accredited programmes (typically delivered to groups of offenders) in developing effective probation practice has, until recently, perhaps supplanted and marginalized more traditional concerns in social work with offenders around the quality of relationships involved in supporting change processes. 49 It is striking for example to contrast the questions of technical effectiveness that have dominated 'what works' discourse with the views of experienced probation staff who were asked by researchers in the late 1980s to describe the key skills required for professional practice. As well as skills relating directly to assessment and intervention, the workers stressed skills in mediating the interests of courts and probationers, in being able to make human contact with offenders, and in relating to offenders (which included effective listening as well as acceptance, respect and honesty). 50

Ten years on from McGuire and Priestley's original statement of 'what works', it is interesting to see some of these neglected skills re-emerging in revisions and refinements of the principles of effective practice. One authoritative recent review suggests that current knowledge about the design and delivery of effective programmes can now be summarised as follows:

  • using human service strategies based on "personality and social learning" theories and on evidence about factors which increase the risk of offending;
  • using community-based settings or, if in custody, making services as community-oriented as possible;
  • using risk levels and criminogenic needs, assessed by properly validated methods to inform targeting and allocation of services;
  • using multi-modal approaches which match services to learning styles, motivation and aptitude;
  • adapting services to difference and diversity, and recognising participants' strengths;
  • monitoring continuity of services and care, including relapse prevention;
  • giving staff clear guidance on principles and on where they can use discretion;
  • monitoring and maintaining programme integrity, i.e. that services are delivered as intended;
  • developing staff skills, including the capacity to maintain "high-quality interpersonal relationships";
  • ensuring good knowledgeable management;
  • adapting services to local context, client groups and services. 51

The author of this review, Peter Raynor, notes that this list is strikingly similar to the earlier formulation. However, he highlights the increasing attention that is being paid to the need for staff to use interpersonal skills, to exercise some discretion in their interventions, to take diversity amongst participants into account, and to look at how the broader service context can best support effective practice. 52

In seeking to address issues of diversity, there is a paucity of evaluations of accredited offending behaviour programmes designed specifically for women. 53 However, a number of writers have attempted to determine the common characteristics of effective interventions. Reviewing the literature on community-based programmes for young female offenders, for example, Batchelor and Burman 54 identify the following elements: a comprehensive and holistic approach aimed at addressing young women's multiple needs in a continuum of care; gender-specific programme models and services that address the specialised needs of young women who offend (paying particular attention to, for example, abuse issues, relationship skills, self-esteem and self-efficacy, self-harm and substance misuse); resources that utilise the skills and experiences of young women themselves; and positive relationships between young women and staff. 55 The attributes of sustainable projects for female offenders include: a resistance to the erosion of gender-specificity; an evolutionary and flexible organisation; a holistic approach to service delivery; a democratic model of policy-formation to enhance staff morale and project success; and a principled approach to probity in human relationships. 56

This evidence echoes Raynor's comments about the significance of practitioner using interpersonal skills and being able to exercise discretion; practitioners need the ability to be flexible and innovative in response to complex and varied needs. The early attempts to apply the 'what works' principles to probation practice in England and Wales perhaps underestimated the significance of interpersonal engagement, ownership and flexibility in delivering effective practice. 57 Neglect of these factors may account for some of the difficulties experienced south of the border in translating the successes of demonstration projects to general practice. 58 Though broad generalisations about effective programmes were undeniably historically necessary in challenging the pessimism that 'Nothing Works', they too readily lent themselves to a managerialised and homogenizing approach to intervention that has predictably struggled to cope with the heterogeneity of offenders to which practitioners must respond on a case-by-case basis. Even at their best, 'what works' studies tend only to address questions about which types of rehabilitative programme seem to work better than others in which contexts and with which particular target groups. While these are important questions, they conceal a flawed underlying assumption; that it is the qualities of the programme that are at the core of the pursuit of effectiveness.

Raynor's recent review suggests that the preoccupation with group programmes arises from their more standardised application which, in turn, allows for more systematic evaluation than the complex and varied nature of general practice. 59 That said, some recent studies have begun to explore the contribution of particular practice skills. Raynor refers in particular to a recent article by Dowden and Andrews 60 based on a meta-analysis examining the contribution of certain key skills (which they term 'core correctional practices' or CCPs) to the effectiveness of interventions with offenders. Dowden and Andrews conclude that their analysis provides strong preliminary evidence about the effectiveness of five particular skills in this regard: effective use of authority, anti-criminal modelling and reinforcement, problem solving, use of community resources, and quality of interpersonal relationships. In general, where these CCPs were present, the mean effect sizes of the programmes were found to be higher. Even more significant effect sizes were achieved where CCPs were present alongside other principles of programme effectiveness and where staff skills and programme design complemented each other. However, Raynor highlights one of Dowden and Andrews' more sobering findings; these CCPs were rarely present amongst the programmes included in the meta-analysis, indicating a worrying neglect of the development and use of appropriate staff skills in work with offenders.

In order to address this neglect, the next section provides an account of some relevant evidence from the psychotherapeutic and counselling literatures.

5.2 Supporting personal change: Lessons from psychotherapy

There are a number of reasons why these literatures about supporting broader processes of change are relevant in thinking about interventions with offenders. The process of giving up crime is not dissimilar from that of desisting from addictive behaviours, such as alcoholism, gambling, and eating disorders. Persistent offending, leading to repeat convictions and punishment, arguably qualifies as addictive behaviour because:

'besides diminished volitional control, what qualifies high-rate behaviours as addicting is that they persist despite harmful consequences. The person is willing to pay what seems too high a price in order to continue them'. 61

More generally, there is a high correlation between offending behaviour and many of the psychosocial problems that are frequently dealt with by psychotherapists, such as depression and low self-esteem. Also, the 'spontaneous remission' of these behaviours is comparable to the pattern of 'growing out of crime': in the majority of cases the problem behaviour eventually ceases even if there is no intervention, but interventions may hasten the process.

As in the corrections field, there has been a drive in the mental health field towards greater accountability and the identification and application of evidence-based practice. There is therefore an accumulating body of evidence concerned with effectiveness and best practice in psychotherapy and in the linked disciplines of counselling and clinical psychology. The conclusions drawn by the most comprehensive and reliable reviews are encouraging. An overview, of the outcome of psychotherapy, in the latest edition of the authoritative Handbook of Psychotherapy and Behavior Change concludes that around 75 percent show some benefit, 15-20 percent show no change, and about 5-10 percent deteriorate during therapy. 62

The latter statistic highlights the need for careful assessment and judicious use of interventions, but the results overall point to a high possibility of beneficial outcomes.

What works in psychotherapy?

As in research on what works in criminal justice interventions, much of the psychotherapy research is concerned with identifying the efficacy of specific methods for particular problems as well as and the relative effectiveness of the various components and principles. In carrying out investigations, distinctions are made between:

  • methods (or interventions or approaches, for example, cognitive behavioural methods)
  • process variables (for example, therapist style, the client-therapist relationship) and
  • extra-therapeutic factors (for example, client resources and chance factors).

Research investigating these has led to a long-term debate on whether any one method of intervention or technical approach is more effective than another in bringing about change, or whether the positive outcomes can be attributed to some common factors (the 'specificity versus commonality' debate). A recurring finding in much of the research is that no method or intervention is any more effective than the rest, and, instead, it is common aspects of each intervention that are responsible for bringing about change. 63, 64

Of course, acknowledging that there are common ingredients that account for much of the positive outcome should not be extended to a claim that all treatments are equal. There are many examples of controlled experimental studies that do show the superiority of some methods over others in treating particular problems. For example, exposure techniques have been shown to be the most effective for anxiety disorders, while behavioural approaches are associated with better results than other methods for treating panic, phobias and compulsions. 65 A major review of high standard comparative evaluations for the American Psychological Association found that the majority of empirically supported treatments (ESTs) were cognitive-behavioural. 66 However, the most balanced representations of the research findings allow that some methods and approaches are more effective than others for specific types of problem, but still reach the overall conclusion that there are common core conditions that are equally or more critical to outcome. 67

What are these 'common factors' or 'core conditions'? There are alternative conceptualisations in answer to this question, 68 but the following are consistently identified as common elements in successful interventions that lead to behaviour change or reductions in problem behaviours:

  • accurate empathy, respect, or warmth, and therapeutic genuineness (sometimes referred to as therapist factors and at other times described as relationship variables)
  • establishing a 'therapeutic relationship' or 'working alliance' (mutual understanding and agreement about the nature and purpose of treatment)
  • an approach that is person-centred, or collaborative and client-driven (taking the client's perspective and using the client's concepts)

Experts argue that this common factors model is widely applicable in the helping professions, including medicine, family therapy, individual therapy, thus cutting across arguments between clinicians operating within specific theoretical frameworks. 69

It should be noted that the most crucial variables of all - chance factors, external factors and client factors - have not been added to the above list because they are obviously 'extra' to whatever is provided by the interventions and brought to the process by the therapist. Meta-analytical reviews to identify the relative weight of relevant factors in producing effective outcomes 'bring home' the extent to which such 'extra-therapeutic' factors, as well as common or core factors, contribute to outcome. The percentage of success that is attributable to the various factors is as follows:

40 percent: client and extra-therapeutic factors (for example, ego strength, social support)

30 percent: therapeutic relationship

15 percent: expectancy and placebo effects

15 percent: techniques unique to specific therapies 70

Though such analyses do not in any way contradict findings that some treatments work better than others for certain conditions, these percentage differences underscore the relatively small contribution of specific approaches. If successful outcomes are mostly attributable to therapist and relationship variables in combination with the client's circumstances and own resources (extra-therapeutic factors) then careful attention must be given in practice to each of these aspects.

Therapist, relationship and client variables

The first element in the list above specifies therapist qualities conveyed to the client. The literature variously refers to such therapist qualities as 'therapist variables' and 'relationship variables', reflecting the difficulty in separating intra-personal qualities from inter-personal expression. Clearly, while therapists are the agents of their own professional skills and interpersonal expressions, there are limits to how far therapist qualities and skills can be treated as separable from how such variables are perceived by the client and are modified in response by client variables. Therapist, client and relationship variables interact and the relevance of each can only be fully understood in relation to each other.

Therapist factors

In a far-reaching review, Patterson summarised the findings of nine major review articles on relationship variables, and concluded that:

'There are few things in the field of psychology for which the evidence is so strong. The evidence for the necessity, if not sufficiency, of the therapist conditions of accurate empathy, respect, or warmth, and therapeutic genuineness is incontrovertible'. 71

Such findings confirm earlier research by Truax and Carkhuff in the 1960s, 72 which was influential in traditional probation practice. Studies have identified variability in the extent to which therapists are able to bring these and other appropriate qualities into their work, with some therapists appearing to cause harm rather than helping. A more recent authoritative review points out that: 'The therapist factor, as a contributor to outcome, is looming large in the assessment of outcomes. Some therapists appear to be unusually effective'. 73

In addition to the characteristics of empathy, warmth, genuineness, etc., various social influence attributes are of relevance, including trustworthiness, expertness and attractiveness. 74 For all such characteristics, it is how they are perceived by the client that is of relevance. 75 Perceived trustworthiness, expertness, and attractiveness have been found to be correlated with client satisfaction 76 and changes in client self-concept. 77 Conversely, premature therapy terminators viewed their therapists as less trustworthy, less expert and less attractive. 78

The relationship and 'working alliance'

In its guideline for treatment choice in psychotherapy and counselling, the Department of Health for England and Wales identifies as a general principle that 'Effectiveness in all types of therapy depends on the patient and the therapist forming a good working relationship'. 79 It gives this principle a higher weighting for reliability than any of its other listed principles. 80 In keeping with this, Lambert and Ogles, in their review of the literature, observe that:

'relationship factors are probably crucial even in the more technical therapies that generally ignore relationship factors and emphasize the importance of technique in their theory of change. This is not to say that techniques are irrelevant but that their power for change is limited when compared with personal influence.' 81

Some of the more specific studies of relationship factors in psychotherapy are organised around the related concepts of 'therapeutic alliance' and 'working alliance'. When a working alliance has been established, the therapist and client share:

  • agreement on overall goals;
  • agreement on the tasks that will lead to these goals;
  • a bond of mutual respect and trust. 82

Various tools have been developed for measuring the strength of alliance. 83 Horvath and colleagues identify both interpersonal skills and intra-personal skills as relevant to the working alliance. 84 Relating this to work with offenders, Marshall and Serran identify trust in the professional as 'crucial in generating change in clients with mental health problems', observing that offenders frequently have lost such trust following previous experiences in which they have shared information about themselves, leading to negative reports about them. 85 The working alliance is not a quality of the therapist, of course, but is 'a set of processes that are dependent on both therapist and client'. 86

Client factors

The interaction of client factors in outcomes is a research domain in its own right. Motivation to work towards change is crucial. It is when we get into the question of client factors that a key difference between working with psychotherapy clients and working with offenders becomes most evident: those seen by psychotherapists are more typically voluntary clients and therefore their motivation to co-operate can be assumed, whereas offenders are likely to be in contact under legislative duress. In practice, there are undoubtedly many exceptions to this 'rule'.

Another important client variable, that is complementary to motivation, is 'expectancy'; that is, whether the client expects treatment to 'work' or to help them in achieving their goals. Clinical theorists have emphasised the importance of facilitating and utilising positive expectancies in psychotherapy. 87 Patients are more likely to engage in therapy if they expect it to lead to the desired outcome, and their level of optimism about the positive outcome can predict the extent of recovery. 88 In relation to desistance from offending, Burnett's findings 89 (reviewed above) about the association of prisoners' optimism and confidence about 'going straight' with lower rates of offending post-release suggest a similar expectancy effect.

5.3 Effective approaches and principles

The clear implication of the preceding discussion is that it may be advantageous in work with offenders to apply some of the techniques and approaches, as well as underlying models and principles, that have been successfully applied in psychotherapy and counselling. This is already happening to some degree, particularly in the adoption of motivational interviewing in criminal justice settings. It is not possible to be comprehensive in the present short review, but is worth highlighting the following because of their obvious applicability to work with offenders: motivational interviewing; person-centred methods; and the application of the principle of responsivity. 90 In this section, we also discuss 'pro-social modelling', an approach which has developed within social work rather than psychotherapy but which has much in common with these other approaches. Indeed, there is considerable overlap between each of these approaches and the skills and principles involved, although each has attracted its own proponents and become a sub-field of investigation.

Motivational interviewing

The approach of motivational interviewing (MI) is now so much a part of work with offenders in the UK 91 that it might not always be appreciated that this is an import from the mental health field. However, work in the mental health field with addictive behaviours is characterised by poor attendance rates and high 'dropout' rates 92 and 'clinicians have come to see eliciting client motivation as part of their job'. 93 The work of Mary McMurran and colleagues in particular has been helpful in developing motivational approaches for work with offenders. 94

The concept of motivational interviewing was introduced by William Miller and Stephen Rollnick, who remain the leading authorities on the subject. 95 They define MI as 'a directive, client-centered counselling style for eliciting behavior change by helping clients to explore and resolve ambivalence'. 96

Note that the adjective 'directive' here refers to the approach being focused on the problem and goal-directed: it does not denote that the therapist should direct the client in an authoritarian or expert manner. On the contrary, a 'quiet and eliciting' counselling style is essential to MI, and the spirit of the approach is distinguished from the techniques. The rationale for the approach is that client ambivalence about the problem behaviour, and a lack of motivation to change it, is commonly at the root of treatment failure. It is this ambivalence or lack of resolve that is the principle obstacle to be overcome.

MI makes use of reflective listening (similar to accurate empathy as operationalised by Carl Rogers) in helping clients to focus on their problems and think about possible solutions. The approach actively discourages:

  • direct persuasion and advice;
  • arguing that the person needs to change;
  • diagnostic labelling;
  • a punitive or coercive manner;
  • a didactic, expert stance which leaves the client in a passive role;
  • behaving in a punitive or coercive manner.

All of these would be seen as contrary to motivational interviewing and violating its spirit. Instead, MI emphasises the personal responsibility of the client to decide whether they have a problem that needs to be addressed. It is also person-centred (see below) in that a premium is placed on the agency and perspectives of the client, with the counsellor's role being that of facilitating the client's articulation of 'both sides of the ambivalence impasse'. 97 In other words, counsellors are encouraged to use a negotiation method where the clients themselves identify and analyse the pros and cons of their addictive behaviour. By enabling this articulation of what is at stake, the counsellor aims to 'tip the balance' between the perceived positive and negative consequences of the targeted behaviour. Other techniques of MI include: eliciting self-motivational statements; providing choice with a variety of alternative change strategies; goal setting; role playing; and modelling.

When motivational interviewing works, it 'involves a sudden shift in how the person perceives the pros and cons of the behavior' and 'is as though one or more cons have suddenly become dramatically more salient, taking on a higher value weight'. 98 MI perfectly complements another import from the mental health field that has been adopted in work with offenders: Prochaska and DiClemente's model of change. 99 MI aims to move the client from the earlier stages of 'pre-contemplation' and 'contemplation' in this model of change to the more advanced stages of 'preparation', 'action' and 'maintenance'.

Non-directive, person-centred methods

A review of outcome research in psychotherapy concluded that a major operational variable in successful outcomes is the intentional utilisation of the client's frame of reference. The investigators produced a guide 100 for working with 'impossible cases', otherwise referred to as 'heartsink' cases and 'therapy veterans'. These are the long-term patients who do not respond to therapy and whose problems continue or get worse. The main message of this guide is that the key to making progress is for the therapist to identify and use the client's frame of reference and worldview, and to bring into play the client's own informal 'theory of change'. An underlying premise is that clients themselves know what needs to be done to address their problem, and that therapists should draw out these insights. In achieving this, clients are to be treated as co-therapists, rather than as cases to be managed with a pre-formulated plan. The client's motivation should be honoured, and the client needs to feel that the therapist sees them as someone with potential and worth. Such an approach taps into what is broadly referred to as a 'person-centred approach' or 'person-centred therapy' (PCT).

The person-centred approach, first introduced by Carl Rogers in 1940, is by no means new to probation services, but, in England and Wales at least, became discredited as part of the move away from social work and towards more punitive and programme-based approaches. Person-centred approaches by definition are focused on clients' own perspectives and concerns, enabling relational engagement as the basis of assisting them and influencing change. 101 Clients are related to as individuals rather than as representatives of a type or a label. A common misunderstanding is that therapists practising person-centred therapy (PCT) have to like their clients and do not question or challenge them. However, experts point out that a clear distinction is made between what a person experiences and how they behave: the former is 'unconditionally valued' but not the latter, and a therapist, without being critical, may convey to a client their own feelings in response to how a client has behaved.

Person-centred approaches can be either directive or non-directive, although purists would argue that person-centred therapy (PCT) is essentially non-directive because it involves engaging with the experiences of the client and becoming a 'fellow traveller' 102 or being 'beside' the client. 103 This is in contrast to motivational interviewing which is explicitly 'directive', in the sense of being focused and goal-oriented, albeit still person-centred. A non-directive approach may be more effective with reluctant clients, involuntary clients. This is particularly relevant to work with offenders who, as noted by Marshall and Serran, are typically a 'difficult population to work with at least partly because they are often defensive and oppositional'. 104 Offenders are also likely to have ambivalent feelings about whether they should change or whether they are able to achieve change. 105 As such, they are likely to be less co-operative, and less responsive to directive approaches. PCT emphasises that therapists should find ways of challenging clients that do not involve criticising them or undermining their sense of self-worth. Again, a distinction is made between being 'beside' the client and being 'on the side of the client'. 106 As in the criminal justice field, person-centred therapists have experience of working with some particularly difficult or fragile clients, including those who 'dissociate' or separate themselves from others, perhaps following abuse in childhood or traumatic adult experiences. One recent development in the field that has been influential has been the provision of 'pre-therapy' for 'contact-impaired' clients. 107

Responsivity: adapting approach to client

Therapists need different skills for different clients and settings and many experts recommend a flexibility of approach that is tailored to client problems and characteristics. This point resonates with the 'principle of responsivity', 108 (defined in section 5.1) The concept of 'responsivity' allows a different angle for addressing similar issues to those in the preceding paragraphs about motivational interviewing and person-centred approaches. As Norcross has expressed it 'seasoned therapists have learned to respond flexibly to patient qualities and to alter their relational stance'. 109

Generally, research indicates that highly resistant clients respond better if there is minimal therapist directiveness, whereas therapist directness and explicit guidance is appropriate for those with low resistance. For example, Patterson and Forgatch found that client non-compliance increased when therapists used directive behaviours such as attempts to teach and confront. 110 Miller and Rollnick advocate a respective and supportive stance for use in motivational interviewing. 111 Avoiding a challenging or over-directive manner takes account of the fact that many clients have ambivalent feeling about changing when they commence contact. When clients are hostile or resistant, a way through is more likely to be achieved by listening and reflecting on what they say rather than being assertive or pushing expertise or focusing on difficult issues too early. In contrast, Marshall and Serran found that, in work with sex offenders, empathy, warmth, rewardingness and directiveness significantly influenced behaviour changes among sex offender clients, though they emphasise that 'directiveness' involved suggesting rather than telling the client what to do. 112

Pro-social modelling

Perhaps the best-known model of intervention focussed on the supervisory relationship, rather than on the features of a given intervention programme, is that developed in Australia by Chris Trotter. 113 The central principles of Trotter's pro-social modelling approach include:

  • Role clarification: involving frequent and open discussions about roles, purposes, expectations, the use of authority, negotiable and non-negotiable aspects of intervention and confidentiality
  • Pro-social modelling and reinforcement: involving the identification, reward and modelling of behaviours to be promoted and the identification, discouragement and confrontation of behaviours to be changed
  • Problem solving: involving the survey, ranking and exploration of problems, goal setting and contracting, the development of strategies and ongoing monitoring
  • Relationship: involving the worker being open and honest, empathic, able to challenge and not minimise rationalisations, non-blaming, optimistic, able to articulate the client's and family members' feelings and problems, using appropriate self-disclosure and humour.

Trotter's empirical research 114 tested the hypotheses (formed on the basis of earlier research) that clients of probation officers who made use of these principles would be more likely to experience reductions in their problems and would be less likely to offend. Trotter trained 12 probation officers in the approach and followed up 104 of their clients. He compared the outcomes for this experimental group with outcomes for a control group of 157 probation clients. Clients in the experimental group were subsequently significantly more likely to report that their problems were reduced and their re-offence rates were also significantly lower than those in the control group. Among the principles, the use of pro-social modelling was most consistently, strongly and significantly correlated with lower offence and imprisonment rates. The model was most effective with young, high-risk, violent and drug-using offenders.

Despite the familiarity of the principles described above, Trotter's work is important for three reasons. Firstly, although it would be possible to conceive of pro-social modelling as a form of individualised programme, it is perhaps better described as a style of or approach to practice, focussed on certain key skills. He demonstrates therefore that we can conceive of styles and approaches and not merely specific programmes as being evidence-based and effective. Secondly, Trotter's research directs attention to workers' qualities as well as being about the characteristics of specific programmes. In this regard, Trotter has also produced evidence to suggest that among staff working in community corrections in Australia, those with a social work background were more likely than those with other occupational backgrounds and qualifications to learn and make use of pro-social modelling and, in turn, to produce lower rates of reconviction. 115 In line with Rex's findings (see below), 116 Trotter suggests that this might be about possession of the social work skills and qualities required to achieve genuinely collaborative problem solving. The third reason for the importance of Trotter's model 117 is that, perhaps by accident, through its focus on effective relationships and processes, it represents work at the interface of the rehabilitation and desistance literatures and attests to the value of exploring this interface.

5.4 Supporting Desistance

To conclude our review of the evidence about how change can be best supported, it makes sense to revisit the desistance research, this time focussing on those studies that have directly examined desistance amongst those who have been the subject of community-based interventions. Evidently, listening to what those previously involved in offending (and in supervision) tell us about assisted desistance seems essential in seeking better ways of working to promote the reduction of re-offending through community penalties.

Desistance and intervention

Rex's research addresses the features of probation relationships and how they come to exert positive influence. 118 Her research involved interviews with 21 probation officers and 60 of their probationers. One quarter of the sample was women. While her research methodology did not enable her to determine whether the offenders that she interviewed had, in fact, desisted, they provided considerable insights into the role of probation in their reported processes of change.

Those that attributed changes in their behaviour to probation supervision described it as active and participative. Probationers conveyed the sense of being engaged through negotiation in a partnership. Given their recognition both of the need to sustain a decision to desist and of the possibility of relapse, probationers seemed more willing to 'embark' on desistance where they felt committed to and engaged in the supervisory relationship. In turn, '[t]his engagement seemed to be generated by the commitment, both personal and professional, shown by workers'. 119 The 'mechanism' by which some probationers come to accept probation officers as role models, Rex suggests, may rely on 'the sense of obligation which the probation officers' support and encouragement seem to generate in probationers'. 120 She found that as many as half of the probationers she interviewed revealed feelings of personal loyalty and accountability towards their supervisors.

In their experiences of supervision, probationers could discern and appreciate efforts to improve their reasoning and decision-making. However, attempts to exert influence through cognitive approaches had to 'carry conviction in their eyes if they were to be effective'. 121 This conviction depended on the personal and professional commitment from workers discussed above. Furthermore, attempts to address cognitive skills seemed likely to be insufficient alone. Probationers also valued guidance concerning their personal and social problems at least as often. Rex summarises this aspect of work as strengthening social ties. Significantly in this context, younger men trying to establish independence also sought practical help, whereas women and other male probationers were keen to receive problem-solving advice so that they themselves could resolve such difficulties.

Reinforcing pro-social behaviour was another prominent feature of probationers' accounts of positive supervision. Probationers could identify advice in this regard as evidence of concern for them as people, and 'were motivated by what they saw as a display of interest in their well-being'. 122 Notably in this context, such encouragement seemed especially important for younger people involved in recidivist offending. Previous research in Scotland by Ditton and Ford 123 has similarly suggested that persistent young offenders might need to be 'won over' by persistent workers to change their behaviour. In this regard, Rex found some evidence that probationers were more willing to accept guidance than probation officers were to be directive, so long as the former could understand the latter's direction as evidence of concern expressed within an engaging relationship.

The Social Context of Desistance

In looking beyond the practitioner/probationer relationship at other factors in the desistance process, the most recent and perhaps most wide-scale study of probation and desistance is particularly pertinent to this review. 124 Farrall explored the progress or lack of progress towards desistance achieved by a group of 199 probationers. Over half of the sample evidenced progress towards desistance. Farrall found that desistance could be attributed to specific interventions by the probation officer in only a few cases, although help with finding work and mending damaged family relationships appeared particularly important. Desistance seemed to relate more clearly to the probationers' motivations and to the social and personal contexts in which various obstacles to desistance were addressed.

Farrall is surely right in going on to argue that interventions themselves and evaluations of them must pay greater heed to the community, social and personal contexts in which they are situated. After all, 'social circumstances and relationships with others are both the object of the intervention and the medium through which… change can be achieved'. 125 Necessarily, this requires that interventions be focussed not solely on the individual person and his or her perceived 'deficits'. As Farrall notes, the problem with interventions based on such shaky criminological foundations is that while they can build human capital, for example, in terms of enhanced cognitive skills or improved employability, they cannot generate that social capital which resides in the relationships through which we achieve participation and inclusion in society. Vitally, it is social capital that is necessary to encourage desistance. It is not enough to build capacities for change where change depends on opportunities to exercise capacities:

'…the process of desistance is one that is produced through an interplay between individual choices, and a range of wider social forces, institutional and societal practices which are beyond the control of the individual'. 126

For Farrall, this necessitates a re-thinking both of 'what works' and of practice. He suggests that practice should be focussed not on 'offence-related factors' but on 'desistance-related factors'. An offence focus must, of course, be necessary and appropriate given that, within any justice context, it is offending which occasions and justifies state intervention. However, being only or overly offence-focussed might in some senses tend to accentuate precisely those aspects of an offender's history, behaviour and attitudes which intervention aims to diminish. It may also tend towards identifying the problem as one of individual 'malfunctioning'. Being desistance-focussed, by contrast, implies a focus on the purpose and aspiration of the intervention rather than on the 'problem' that precipitates it. It also tends towards recognising the broader social contexts and conditions required to support change. Thus, where being offence-focussed encourages practice to be retrospective and individualised, being desistance-focussed allows practice to become prospective and contextualised.

The nature of the difference between the two approaches is well captured by one of the probationers in Farrall's study, in response to a question about what would prevent him from re-offending:

'Something to do with self progression. Something to show people what they are capable of doing. I thought that was what [my Officer] should be about. It's finding people's abilities and nourishing and making them work for those things. Not very consistent with going back on what they have done wrong and trying to work out why… For instance, you might be good at writing - push that forward, progress that, rather than saying "Well look, why did you kick that bloke's head in? Do you think we should go back into anger management courses?" when all you want to do is be a writer... I know that you have to look back to a certain extent to make sure that you don't end up like that [again]. The whole order seems to be about going back and back and back. There doesn't seem to be much 'forward'"'. 127

5.5 Conclusions: Supporting Desistance

This section of the report has reviewed a very broad range of material about how personal change in general and desistance from offending in particular might be best supported. From the review of the psychotherapeutic literature it emerges clearly that while there are variations in the efficacy of different therapies or treatments, these are, on balance, less significant than the common core conditions for effective interventions. It seems that contextual factors, including client and therapist factors and working relationships are more significant predictors of positive outcomes than particular methods. In the light of this evidence, the limited successes of attempts to apply the design principles of effective programmes to reduce re-offending begin to make more sense and the need for a broader understanding of effective practice becomes obvious. The literature on 'assisted desistance' provides a range of emerging evidence about the contextual and relational factors that contribute to effective intervention. From the last two sections of this report, pulling together our review of the evidence about the process of desistance and about how it might be best supported, we can conclude that:

  • Desistance is a process which is commonly characterised by ambivalence and vacillation. It is not an event. This suggests the need for motivational work to prompt, support and sustain change efforts.
  • Desistance may be provoked by life events, depending on the meaning of those events for the offender.
  • Desistance may be provoked by someone 'believing in' the offender. This underlines the importance of workers sustaining an optimistic and persistent approach through periods of lapse and relapse.
  • Although the development of better cognitive skills may be a part of the process, desistance probably involves a broader change in narrative identities (or self-stories). This suggests the need for interventions which support narrative reconstruction.
  • Desistance is an active process in which agency is first discovered and then exercised. Supervision processes should respect this agency by seeking to maximise involvement and participation. The 'discovery of agency' may also imply a prospective focus for practice, drawing on solution-focussed interventions that capitalise on strengths, resilience and protective factors.
  • Desistance requires social capital (opportunities) as well as human capital (capacities). This suggests an advocacy role for practitioners seeking to support change and underlines the need to target systems beyond the individual offender.
  • Desistance is about 'redemption' or restoration and often involves finding purpose through 'generative activities'. This implies the need, at an appropriate point in the process, to support the development of a more positive identity by accessing opportunities to make a positive contribution to local communities.


The broad range of evidence reviewed above suggests at least four key skill sets that it is necessary and desirable for criminal justice practitioners to develop in order that they can positively support desistance and thus to bring about a reduction in re-offending. In this concluding section of the report, we offer a brief and preliminary outline of what each of these skill sets might involve.

Skill Set 1: Building relationships that support change

It is clear from the psychotherapy and counselling literatures that the relationship between the therapist and the client is a critical factor in effective interventions. It is the basis for learning about, and gaining the co-operation of the client, and for matching and modifying interventions in interaction with client variables. Building effective relationships is, in turn, underpinned by the practitioner's ability to develop and use strong communication, counselling and inter-personal skills. Indeed, these skills are critical to each of the skill sets discussed below.

The 'core conditions' of effective interventions relate to the ability of practitioners to convey accurate empathy, respect, warmth and 'therapeutic genuineness'; to establish a working alliance based on mutual understanding and agreement about the nature and purpose of the treatment; and to develop an approach that is person-centred or collaborative. The literature on the working alliance in particular stresses agreement on goals, agreement that the tasks involved will lead to the achievement of these goals, and the existence of a bond of mutual respect and trust. The core correctional practices (or CCPs) identified in the 'what works' literature similarly suggest that key features of effective practice with offenders include the quality of the interpersonal relationship, the effective use of authority, anti-criminal (or pro-social) modelling and reinforcement, problem solving, and use of community resources.

The desistance literature more generally recognises that desistance from crime is characterised by ambivalence and that, therefore, the ability to foster and sustain motivation is critical to effective work with offenders. This underlines the significance of the skills associated with motivational interviewing in particular. Desistance is also an active process and one in which agency (that is, the ability to exercise choice and manage one's own life) is 'discovered'. This necessitates approaches to supervision that are active and participative and that seek to maximise involvement and collaboration.

The desistance literature also highlights the need to establish relationships within which attempts to positively influence the offender carry moral legitimacy (from the offender's perspective). This again underlines the need for the worker's authority to be exercised in a manner that is clear, explicit and fair. It also points to the importance of offering practical help to offenders since this a vital expression of concern for them as people, as well as demonstrating an awareness of their reality. That concern lends credibility to attempts to influence behaviour. A wider body of research attests to the fact that people tend to internalise and thus sustain changes in their behaviour when their compliance with attempts to influence them is based on normative rather than calculative grounds. In other words, the slower process of winning people over by example, by persistence and by persuasion is, in the longer term, more effective than the offer of rewards or threats.

It is clear therefore that paying adequate attention to the relational aspects of practice with offenders, and to the skills through which effective relationships are developed, is a necessary (but not a sufficient) condition of effective practice. Little can be achieved within any method of intervention unless practitioners can establish the right kinds of relationships with offenders, despite the challenges that the legal context provides in this regard. Working to establish consent (albeit constrained) to the process of rehabilitative intervention is, of course, also a requirement of ethical practice.

Skills Set 2: Assessing risks, needs and strengths

The extensive literature about the development, use and limitations of risk and needs assessment is beyond the scope of this review. 128 That said, in discussing the principles of effective programmes, we have noted that in the 1990s risk of re-offending increasingly came to be seen as pivotal in determining what level or intensity of service should be provided in individual cases. As 'the risk principle' became increasingly well known, researchers worked to develop risk assessment tools that could supplement professional judgement and, perhaps, provide for greater consistency. However, such technical questions (about what came to be described as 'treatment dosage' issues) have latterly been supplemented with (if not eclipsed by) questions about risk of serious harm. This shift in emphasis reflects penal policy on both sides of the border which has increasingly come to emphasise priorities and issues around community safety and public protection. Both issues, re-offending and serious harm, have also driven the development of new techniques of risk assessment.

Robinson provides a recent and succinct summary of approaches to risk assessment in working with offenders. 129 Given the focus on delivering 'defensible' risk assessments (that is, those that can stand up to scrutiny when things go wrong), issues of quality, consistency, reliability and accuracy have come to the fore in risk assessment practice and, in consequence, one of the key debates has been about the relative merits of 'clinical' and 'actuarial' methods. Whereas clinical approaches (or 'first generation' risk assessment) rely on the professional knowledge, skills and experience of individual practitioners, actuarial approaches (or 'second generation' risk assessment) derive from statistical calculations of probability correlating specific risk factors with reconviction data. As a result of the limitations of both approaches, 'third generation' approaches to risk assessment have been developed which include attention to 'dynamic' (or changeable) risk factors. By highlighting specific criminogenic needs, third generation risk/needs assessment tools aim to individualise risk assessment in order to guide practice, but without compromising the predictive validity of such tools. However, as Robinson suggests, such third generation instruments generate their own set of issues and problems. These include the demands which their increased complexity makes on workers' time, sometimes leading to 'completion fatigue'; related dilemmas in balancing comprehensiveness and predictive accuracy with usefulness and brevity; the re-introduction of elements of professional judgement and related issues of consistency and bias; and concerns about the ability of such tools (derived largely from research involving white males) to accommodate and respect diversity and difference vis-à-vis gender and ethnicity in assessing risk.

A further issue, at present, is that it appears that current risk assessment instruments have limited usefulness with respect to arguably the most difficult area of risk assessment; assessing the risk of serious harm. 130 This is concerned not with the probability of an event occurring, but with the anticipated severity of its impact. With this in mind, and recognising the risks of discriminatory biases in all forms of risk assessment, Tuddenham argues for and outlines a model of 'reflexive risk assessment' within which practitioners recognise that their knowledge is emergent, tenuous and open to revision; that definitions and discussions about risk are contestable and culturally relative; and, that risk assessment functions within certain policy contexts and within a society increasingly pre-occupied with risk and its prediction. Reflexive risk assessors need to retain awareness of their own potential for denial of risk; they need to retain the ability to 'ask the unaskable, think the unthinkable, imagine the unimaginable'. 131 They must encourage and enable offenders 'to say difficult things by asking the right questions; and then hear what the offender wants to tells them, giving the latter confidence that the worker can tolerate the answers'. 132 With Tuddenham's insights in mind, it is obvious that the quality of the relationship between the worker and the offender will be a critical factor in determining the effectiveness of the assessment; in the absence of sufficient trust between worker and offender, it is highly unlikely that the information on which accurate risk assessment depends can be gathered.

This relational aspect of risk assessment and the complexities of and necessity for reflexivity in the process imply the need for a very high degree of professional skill (and specialist training). They also suggest a need to look beyond the qualities of the assessment tools being deployed and to look at the ways in which they are used. One recent study in Manitoba in Canada has explored how probation officers use risk-needs assessments to formulate their case plans and how they manage their cases. 133 The results showed that the development of intervention plans was based more on what the court mandated then what the offender assessment indicated. As a result, addressing the offender's criminogenic needs in supervision was not as common as expected. Probation officers did engage in behaviours that have been associated with positive behavioural change (for example, pro-social modelling) but more effective practice could have been expected.

More generally, though well-designed instruments are useful in assessment work, the desistance literature also underlines the important part that the qualities and skills of the worker plays in developing the relationships within which information is gathered and analysed. Certainly, one of the recurring messages from the desistance research reviewed above is that practice must be thoroughly individualised. Both the age and gender related differences in both persistence and desistance and the significance of the subjective meanings of events and changes for those involved attest to the need for practice that sensitively and thoughtfully individualises the change process. This need is unsurprising since, as we have already argued in section 3, although there are certain commonalities, for example, among young people involved in persistent offending, the categorisation of their characteristics, needs and deeds in large-scale studies tends to conceal their differences. 134 It follows that employing styles of assessment, case management and direct practice that value and exploit that diversity seems necessary. Where risk and needs assessment instruments can properly underpin case management by informing thorough and properly argued professional analyses, and where they include resources for engaging people in the process, they can assist in enhancing the quality, credibility and consistency of individualised assessments. However, the desistance literature implies the need to review risk factors in the light of pro-desistance factors including the positive qualities of the individual and the available resources for supporting change within their social networks.

Skill Set 3: Research-based planning and delivery of interventions

The planning of effective interventions should follow from effective assessment practice. Essentially planning (or design) involves the development and continuous review of strategies for change. If assessment requires the development of clear understandings both of the reasons for the offending behaviour (criminogenic needs) and of the available resources within and around the offender to address it (desistance factors), then planning should rest on the development of credible and testable theories of change. In other words, the question becomes: on the basis of the best available research evidence, what do we (the practitioner and the offender) think might best promote the reduction of re-offending change in this situation? The planning process thus articulates the core rationale of the intervention: why do we think that doing what we propose to do will bring about the results that we want to achieve? Arguably, this is the logical step that is most commonly neglected in practice. 135 One reason for this neglect may be that the centralising aspects of 'what works' have, via processes of accreditation, imposed a generalised and homogenous theory of change at the strategic level, rather than encouraging the development of heterogeneous theories of change, allowing for more subtle and nuanced applications, at the level of individual cases. The evidence reviewed above suggests that, though well intentioned, this method of developing effectiveness is incapable of adequately responding to the individual complexities that practitioners face. 136

The development of individualised theories of or strategies for change, properly conceived and conducted, represents a considerable challenge for practitioners. The skills required include not just the ability to seek, secure and evaluate information from a wide variety of people and sources but also the ability to bring a wide range of theory and research to bear on the information gleaned. Furthermore, it is necessary, as far as is possible, to engage in this kind of theory/practice integration in partnership with the offender. This partnership is critical because however good the 'theory of change' may be, its implementation in practice will fall flat unless the offender recognises its relevance, appropriateness and viability.

Given the range of risk and needs factors discussed in section 3 above, it is clear that strategies for reducing reoffending are likely to involve multi-systemic, multi-modal interventions; that is, interventions that work in a variety of ways to address a variety of issues. Thus a truly multi-systemic intervention might involve, for example, individual work (whether in a group setting or one-to-one) to develop problem-solving and cognitive skills and to address other personal problems, family work to develop positive relationships capable of supporting desistance, work to encourage either changes within an 'anti-social' peer group or to facilitate withdrawal from the group, advocacy work to access resources to address disadvantages located within the local environment, and work to challenge social structures and attitudes that impede the inclusion of ex-offenders. Evidently, the degree to which practitioners focus on working in and through each system should depend on individualised (and criminologically informed) 137 assessments of risks, needs and strengths and on practical judgements concerning where the most effective degree of positive change can be achieved. However, the emphasis on the significance of social capital in the process of desistance makes clear that interventions plans are likely to include strategies for developing social opportunities as well as individual motivation and capacities.

Whatever the type and level of the intervention, at a practical level change planning also requires the ability, in partnership with the offender to set specific targets for the work. These targets should be such as to allow the practitioner and the offender to know whether or not the enactment of the plan is delivering the intended outcomes. The review process can then be based on clear evidence that informs thoughtful analysis concerning whether the theory of change is holding good and, where it is not, it should allow the practitioner and the offender to explore whether this is because the theory is flawed or because of other factors. This iterative process then permits the continuous revision of assessment, theories of change and action plans in pursuit of the desired outcomes.

Skill Set 4: Managing change

A further skill set required of practitioners concerns the management of the change process. If interventions are likely to be multi-modal and multi-systemic, and may involve several personnel within the agency and outside it, then the practical difficulties of maintaining sufficient integrity across the different aspects of the supervision process are likely to be considerable.

Contemporary commentators in England and Wales have noted that the shift from generic practice (where the caseworker manages and delivers all aspects of the intervention) towards more specialist practice (where case management and programme delivery functions are typically split) has been one of the most significant changes in probation work in the last decade. Robinson and Dignan, for example, 138 review the shift from generic casework to case management to a fragmented style of offender management in which staff increasingly occupy specialist roles and offenders encounter a variety of staff in the process of supervision. Reflecting on existing research, they argue that offenders are not best served by a system in which they are conceived as 'portable entities', and in which staff are obliged to engage in a 'pass-the-parcel' style of supervision.

The research evidence that we have reviewed above, in particular its consistent and compelling message about the importance of the relational aspects of effective practice, would tend to support Robinson and Dignan's conclusion that the task of managing interventions so as to promote and sustain desistance is not an administrative one; it makes better sense to conceive of the case manager's role as being 'therapeutic' 139. In reviewing the implications of research on models of case management for effective probation practice, Holt identifies four over-lapping features of case management: 140

  • Consistency is a vital ingredient of seamless service delivery. It also allows the worker to promote and reinforce effective learning (perhaps from structured programmes) by providing opportunities to exercise new skills; to put theory into practice. Consistency also provides an essential element of the positive working relationships that, as we have seen above, are critical in order to support and enhance motivation to change.
  • Continuity across all aspects of the intervention and over time is necessary if the intervention is to be meaningful and productive for the offender. The case manager needs to ensure that the offender experiences supervision as an integrated holistic process; a key part of achieving this integration is likely to be the provision of one stable and supportive relationship throughout the duration of the supervision experience.
  • Consolidation of the learning is achieved when the case manager allows the offender to reflect upon the learning achieved in the different aspects of supervision. This involves enabling the offender to make connections across all aspects of the process; to join up the learning. However, consolidating the learning also requires accessing opportunities for community reintegration, where the offender's strengths can be employed and confirmed.
  • Commitment of the case manager to the offender and to the supervision process is essential in promoting desistance. This commitment creates stability in the delivery of the intervention and provides a 'holding context for change'.

A fifth feature needs to be added to this model of case management given the criminal justice context; that is, the management of compliance. In recent years, a stronger focus on the enforcement of community penalties has emerged, particularly (but not exclusively) where community penalties have been recast as 'punishment in the community'. This recasting of purpose has increased the existing need for effective enforcement in order that courts regard community penalties as credible disposals. Though the language of 'enforcement' implies an emphasis on ensuring the meaningfulness and inevitability of sanctions in the event of non-compliance, Tony Bottoms has argued convincingly that attempts to encourage or require compliance in the criminal justice system must creatively mix constraint-based mechanisms (those that somehow restrict the offender), instrumental mechanisms (related to incentives and disincentives) and normative mechanisms (related to beliefs, attachments and perceptions of legitimacy). 141 What is clear from our review is that, through the establishment of effective relationships, the case manager's role in supporting compliance is likely to be particularly crucial to the development of these normative mechanisms. It is only within relationships of the kind discussed above that the formal authority conferred on the worker by the court is likely to be regarded as legitimate by the offender. This legitimacy is likely to be a crucial factor in preventing breach by persuading offenders to comply.

However, the success of case management at the individual level depends on the existence of the local strategic partnerships and pathways that allow the case manager to access and coordinate the required services and resources. Even the best designed and supervised research based individual case plan will fail if the case manager cannot access the services and resources required to implement it. 142 Similarly, the best developed approach to securing compliance will fail unless organisational arrangements exist that underpin the worker's legitimate authority by delivering swift and proportionate responses that reward compliance and deal effectively with non-compliance.

Conclusions: What works and who works

Though much more could be said about the case manager's roles, tasks and skills in assessment, planning, delivery, monitoring and evaluation of change processes, this final point, about the importance of effective relationships is perhaps a suitable place to end our review. Figure 2 below indicates how effective relationships have emerged in this review as being at the crux of effective practice. The supervision process begins with the establishment of relationships and the effectiveness of every subsequent part of the process will depend in part on the quality of the relationship, though good relationships alone will not be enough to bring about change. In other words, although we have described the ability to build and utilise relationships as a discrete skill set in its own right (including communication, counselling and inter-personal skills), in fact it underpins each of the other skill sets and each aspect of the supervision process. This has been the recurring theme of this literature review. Whether we look to the latest versions of the 'what works' research, to the psychotherapy literature or to the desistance research, similar messages emerge. The accumulated weight of evidence, coming from studies that start with quite different assumptions and using very different methodological approaches, drives us towards recognition that practice skills in general and relationship skills in particular are at least as critical in reducing re-offending as programme content.

Figure 2: The supervision process

Figure 2: The supervision process

In this regard there is common ground between both research literatures and practice perspectives in mental health and criminal justice because workers in both fields are involved in a complex 'human endeavor' which is 'messy' and cannot be 'reduced to a technical mechanistic enterprise'. 143 The division between 'who works' and 'what works' is a valuable distinction because it ensures that all elements are given necessary attention while recognising that they are intertwined. Ultimately it is unhelpful to regard programme, practitioner and offender as separate dimensions. There is every indication that future investigations, in both psychotherapy and criminal justice interventions, will give more emphasis to the multi-dimensionality of contributions to outcome, 144 and it is this multi-dimensionality that should be kept in view in developing evidence based practice in working with offenders.

Clearly, if these individualised interventions to support desistance need to be multi-dimensional, then so too are the skills required to deliver them. Crucially, the business of reducing re-offending by supporting change involves a range of skills that goes far beyond those involved in reducing re-offending by imposing control, monitoring or enforcement, important though these measures are. What is required is a complex mix of skills which require significant personal qualities as well as a high degree of training across a range of therapeutic, academic and management disciplines in order that one worker is able to draw together approaches that address various areas of an offender's life so as to coherently and consistently support the change process. It seems to us that, within the criminal justice context at least, social work almost exclusively encapsulates this broad skill base with its holistic attention to the full spectrum of an individual's needs in his or her social context. The development of effective services to reduce re-offending in Scotland therefore requires political and professional investment in equipping the relevant frontline staff with the key skills required for effective practice and in creating the contexts for practice that provide them with realistic opportunities to exercise these skills.