Introduction

The organisations most likely to establish projects to work with domestic violence perpetrators in the UK are voluntary bodies and the Probation Service. The latter has been taking over from the former as the prime site of activity. As at May 1999, there were 27 projects listed by the National Practitioners' Network (now relaunched as Respect), of which 14 were primarily run within the criminal justice system (a doubling in five years if we take Scourfield, 1994, as a baseline) and a further two had extensive overlaps between the two sectors. The National Probation Directorate is about to pilot eight new groups in London and West Yorkshire alone. Even so, we are a very long way from comprehensive provision. In the USA, with a population only around four times as large as the UK (though admittedly the work has a longer history), there were already 500 programmes by 1990 (Thorne-Finch, 1992). The majority of British perpetrators do not have a group in their area.

Although major funding for expansion is most likely to come from the Home Office and to be targeted on Probation, questions have been raised as to whether a diversity of provision is not healthier. Respect is arguing to retain the creativity and diversity of the voluntary sector, as well as its role in working with men who have not been convicted, but who nevertheless form a majority of those perpetrating domestic violence. There is a fear that too much standardisation may stifle innovation and continued development (see also Healey and Smith, 1998). Also, there is every reason to believe that survivors and concerned professionals want there to be provision for men not subject to court action. The situation should not have to escalate before an abusive man is held responsible for his behaviour.

Models of practice

There is a convergence of thinking on the most appropriate way of working with domestic violence perpetrators in groups, with a broadly cognitive behavioural approach combined with challenging negative gender attitudes emerging as the favoured model of intervention (see Mullender, 1996, for an overview). A key influence on practice has been the long-standing work in Duluth (Pence and Paymar, 1990). Programmes concentrate on perpetrators accepting responsibility for their own violence, without blaming uncontrollable anger, alcohol (Kaufman Kantor and Straus, 1987), family background or their partners. This requires skilled groupworkers to challenge men directly, to use the group to encourage men to challenge one another, and to facilitate discussion about unacceptable attitudes and behaviour towards women.

Recent evidence has emerged from America that extending beyond behaviour change into altering gender attitudes may positively affect the success rates of perpetrators' groups (Gondolf, 1998d). Perpetrators learning to talk things through was statistically linked with the third of women who reported 'a great extent' of change in their partners. The longest groups had better results with men using discussion to avoid violence. This means that the essence of change efforts may well be groupwork experience and skill, as borne out also by men's own feedback in Scotland on the influence of group discussions (Dobash et al., 1996).

Areas of concern

Although growing in numbers and clearer about their chosen approach, perpetrators' programmes remain highly controversial. This is due to:

the lack of clear evidence of their success overall;

fears that, if they do not work, they may leave survivors in greater danger through having built up false hope;

the potential that they may compete for resources with direct services for women and children;

suspicions that workers might become collusive with the denial and minimisation that typify perpetrators' perspectives on their violence;

the potential for diluting the response of the criminal justice system.

It is essential that any new programme that may be established as a result of this feasibility study takes all these criticisms fully into account and ensures that each can be tackled within any scheme recommended for adoption. Equally, it is essential that any proposed group should work to agreed minimum standards. Respect (2000; see Appendix 1) provides minimum standards and good practice guidelines covering matters such as the size and length of groups, and the need for adequate training and supervision for workers.

The uncertainty which continues to surround perpetrators' programmes means, too, that it is important not to direct all our efforts or resources into this work. Wider preventative efforts, and particularly emergency services for women and children, require renewed commitment also (Tolman and Edleson, 1995).

Non-completion rates

In all the countries where work with domestic violence perpetrators is furthest advanced, completion rates are problematic (Mullender and Burton, forthcoming). In addition to those men judged unsuitable at initial assessment, rates for 'no shows' (failure to report to the group) and drop-outs at subsequent stages are uniformly high. As well as being dangerous for women (who may be more likely to trust a man who seems to be trying to change) and frustrating for practitioners, this leads to attrition in evaluation samples which makes results harder to judge. In Australia, non-completion rates can exceed 50 per cent (National Campaign Against Violence and Crime Unit, 1998) while, in an overview of American research, Gondolf (1997; 1998b) cites 40-60 per cent of men dropping out and overall completion rates as low as 10 per cent of initial referrals. In the UK, completion rates are reported as ranging from 30 to 90 per cent of those starting a group, but with substantial latitude in some instances in excusing absences (Mullender and Burton, forthcoming).

Low completion rates mean that effective monitoring and sanctions for non-attendance need to be in place. A written agreement with each man who joins the programme is the norm, so that he is clearly aware what is required. Sanctions in groups operating outside the criminal justice system have been reviewed by Mullender (1996). Exclusion from the group is, of course, an obvious possibility, but is somewhat self-defeating. Various alternatives traced, mainly in the American literature, include insisting on the man starting the programme again, requiring him to agree to place money and a signed confession at his partner's disposal for future use should she eventually decide to leave, and collecting donations for the local refuge in the form of fines imposed by the group. The most interesting trend within this is towards ensuring that the man's partner, and/or women more generally, should benefit from any sanction imposed on the man. This accords with the emphasis in the programme itself on his not being allowed to escape a challenge to accept full responsibility for his actions and for his partner's safety.

Mandatory versus voluntary attendance

The involvement of the National Probation Directorate in rolling out perpetrators' programmes under its Pathfinders initiative coincides with a move against funding voluntary sector perpetrator projects to work with probation clients. Since probation groups do not take non-convicted men, the days of groups that combine referrals from the courts and from elsewhere would appear to be numbered.

There will be challenges for the voluntary sector beyond surviving without probation funding. Mandated participation offers the benefits of 'integrated service delivery, monitoring of a perpetrator's activities and whereabouts, consistently applied penalties for non attendance, and increasing penalties for continued breaches' (National Campaign Against Violence and Crime Unit, 1998, p.17). Mandating should theoretically be better able to tackle issues of poor attendance and consequently improve the ability to monitor change and prioritise survivors' safety. There may also be gains in effectiveness; involvement with the courts at intake was one of only two variables in one American study found to be predictive of lower rates of violence at follow-up, the other being an absence of prior mental health treatment (Edleson and Syers, 1991). Group facilitators working with non-convicted men have to be able to tackle all these issues without the criminal justice service formally supporting them in challenging each man to comply with the rules of the group. Nor do they have much opportunity to engage men's trust or motivation through relationship building, given that they are working within a model that requires them to be confrontative from the very beginning.

Arguably, however, new kinds of mandates are developing, with a new mix likely to appear in voluntary sector groups. Growing links between perpetrator projects and social services departments could lead increasingly to service level agreements to accept men as part of care plans. CAFCASS may develop in a similar direction in relation to contact cases now that the Family Courts are being guided to take domestic violence into consideration. This, together with partner pressure, has been referred to as a 'community mandate' (Burton et al., 1998).

Survivors' safety and the inter-agency context

Particular emphasis is placed in practice and research on the safety of the perpetrator's partner and of any children involved. Best practice involves establishing a partner support service. Considerations about the female partners of perpetrators also need to be at the forefront of planning both the men's groupwork aspect of any new project, because of safety issues, and its evaluation, because partner reports are likely to be the best measure of the cessation of violence and associated problems. Partners need to be warned that referral to a group is not a guarantee of behaviour change. In order to make the safety of women and children the primary consideration in practice, Respect makes a number of detailed recommendations covering information sharing, confidentiality, service delivery and awareness raising. These are shown in full at Appendix 1.

Groups are also seen as needing adequate child protection policies in place, for example where men reveal actual or potential child abuse in the group. Close links with child welfare and child protection agencies are essential for safety reasons. There are also useful examples now of such agencies contributing parenting content to men's programmes (notably in Birmingham).

All of the above leads to an emphasis on operating perpetrators' groups within an inter-agency context. Inter-agency partnerships and forums (Hague and Malos, 1996) play a key role in making agencies more aware of one another's practice, sharing good practice, drawing up policy and practice guidelines, offering joint training, creating channels for referrals to perpetrators' programmes, increasing the accountability of these programmes and linking the work into a comprehensive approach to prevention. They also establish new services to fill emerging gaps in provision. Some commentators consider that links with advocacy and other services for women, and with wider work on public attitudes and assumptions, may have an impact on programme success (Gondolf, 1997). Some of the best developed work with perpetrators in the USA, for example in Minneapolis and Duluth, Minnesota, forms one part of a co-ordinated response to men, women and children which is widely regarded as setting a model of best practice.

Diversity of cultural and ethnic background

Diversity is an issue both in relation to men accessing perpetrator programmes and to the women and their children to whom a parallel support service may be offered. As far as the perpetrators are concerned, most of the relevant work to date has been conducted with African American men (Williams, 1994; 1999). It has suggested that predominantly white groups may be less able to motivate black men to change than ethnically sensitive groups which harness community values such as the importance of family. Not all communities in the UK have yet accepted the existence of domestic violence in their midst, or that men's behaviour within the family should be subject to social control. Considerations of family honour can also make individuals unwilling to expose their private concerns to others.

But cultural issues are an unjust smokescreen for perpetrators. Women from minority ethnic communities have the same right to safety as others and, indeed, may need more rather than less help when attempting to access services. Unsatisfactory and discriminatory, as well as overtly racist, attitudes (Mama, 1996) persist; a study in London (James-Hanman, 1995) suggests that existing provision still fails to meet language and cultural needs, leaving black and minority ethnic women isolated and unaware of their options and sometimes placed in additional danger, for example if interpreters are used from within their own family or community. Women who are losing all their family and community support are in particular need of both emergency and longer-term support from the relevant agencies (Hanmer, 1996). Rai and Thiara (1997) found that black organisations in the community are valued by minority ethnic survivors but that other sources of help are perceived as not understanding the black experience and sometimes as institutionally racist, particularly in predominantly white areas. For survivors who do not have British nationality, control and domination may have been exercised by the perpetrator in relation to citizenship, residence rights and required documentation. The inclusion of a ‘race’ module in a perpetrators’ programme can address the existence and impact of racism, contextualising this within both the family domain and wider society. Racism can be challenged without this being used as a smokescreen to obscure responsibility for violence.  Workers in any parallel service established for survivors in Coventry would require a high level of awareness on immigration and asylum matters as these relate to domestic violence.

Monitoring and evaluation

Perpetrators' programmes are notoriously difficult to evaluate. Methodological criticisms of reported studies have included subjectivity, lack of control groups, confusion about causal links, high drop out rates (those who complete may not be typical), and projects studied being too small and too recently completed to test their true effectiveness (see overview in Mullender and Burton, forthcoming). 'Best practice' in evaluation in this field is now considered to combine self-report, partner report and official data. Of these, partner reports have emerged as the most valid and reliable measure. It is also regarded as valuable to combine attention to any repeat of the violence with the question of whether the partner's quality of life has actually improved, i.e. whether she now feels safe and whether the perpetrator's attitude towards her has improved. All results need to be reported against a clearly stated baseline or baselines; for example, monitoring of those referred to the programme, those commencing or those completing it. In addition, length of follow-up, response rates, control groups and attrition rates all require attention in future studies. A comparison across groups would be particularly desirable. There are so many pitfalls that it is critically important for evaluations in this field to be undertaken by specialists.

Costings

There is little UK information generally available on the costs of establishing and running perpetrators' programmes. The National Practitioners' Network, now Respect, has stated for some years that unit costs should be calculated to include not only the minimum standard of work it advises with the perpetrator himself, but that for work with the man's (ex)partner. This may mean that intervention looks expensive, but it can usefully be set against the enormous cost of doing nothing. Leaving aside the human pain and distress involved, the financial cost of allowing domestic violence to continue uninterrupted has been estimated in one London borough alone as standing as high as £7.5 million per annum (Stanko et al., 1998).

Some relevant material on costings is available from the USA. Snow Jones (1999) compared the costs of four contrasting domestic violence perpetrators' programmes and found the estimated cost per individual session of attendance to be lower than for comparable group sessions in relation to mental ill-health or substance misuse. Costs ranged from $265 to $864 per overall programme attendance, the lowest being kept artificially low by using sessional workers to run the groups. There were no effectiveness measures incorporated into this study, and no comment as to how other elements of this programme were sustained.

Cost effectiveness always has to be measured against alternatives. If perpetrator programmes work, they might potentially prevent very heavy costs to the criminal justice system and to health (including mental health) and social services agencies in terms of services for women and children (Stanko et al., 1998). There are also immeasurable future costs, for example where children who live with violence do not fulfil their educational potential or where women are lost to the community as active citizens, workers and mothers, through being injured, emotionally disabled or killed. It is also important to remember that domestic violence has a long-term pattern; it typically goes on over many years, thus multiplying all the costs many-fold. Dobash et al. (1996) point out that an effective response, whatever the outlay, might bring a halt to this expenditure over time and thus eventually save money. Nevertheless, in the light of the undoubted cost of providing and evaluating perpetrators' programmes, it is not surprising that those who continue to run underfunded emergency and outreach services for women and children are concerned about questions of disproportion and why the National Practitioners' Network has long argued that there should never be competition for funding between work with men and work with women and children.

Perpetrator programmes may also offer 'added value'. Having her partner attend a group may give the woman a chance to leave, monitors her partner's behaviour and gives her at least a weekly respite, and offers contact with support services where she will be taken seriously and believed, perhaps for the first time. These spin-off benefits are not inconsiderable, even when the group is not rehabilitative in an individual case. Where there is social services or Family Court involvement, the opportunity to monitor and assess the perpetrator's behaviour in a group may be the key to an effective outcome where, hitherto, men's dangerousness has too often been an unmanaged risk (Farmer and Owen, 1995).

Costs to the perpetrator

In the American context, programme participants pay from a fifth to the whole of the cost of their own attendance in the form of fees, partially recouped in some cases from their insurance cover (Snow Jones, 1999). In a UK survey (Mullender and Burton, forthcomng), two voluntary projects were charging fees to self-referred men, on a sliding scale according to income and outgoings. There are other costs to the perpetrator, beyond financial considerations, and benefits, too, to him and his family. He has to give up time to attend the group, has his behaviour and whereabouts over the previous week questioned and monitored, is required to disclose his unacceptable behaviour in front of a collection of frequently unsympathetic and highly challenging strangers, is expected to work hard in the group and on 'homework' tasks in between, and may feel as if he is back in school. Conversely, Burton et al. (1998) list a number of gains which Stage One completers of the DVIP programme were able to cite in their own, their partners and their children's lives; 72 per cent, for example, listed greater security for the children and 59 per cent cited learning alternatives to violence.

Success rates

At the end of the day, do perpetrators' programmes work? Not surprisingly, this is a difficult question to answer. As explained above, few evaluations have been sufficiently rigorous. Then, too, we must exercise particular caution in reading claims for success. First, these may be given as a percentage of widely differing baselines: Tolman and Edleson (1995) give an example of a large-scale follow-up in which a two-thirds success rate could be claimed for programme completers but where this represented only 20 per cent of initial contacts with the agency. Secondly, without experimental studies (with randomised controls), we cannot know whether the same proportion of men might have achieved this result without assistance. This caution is borne out by one study (Feld and Straus, 1989) which did find cessation of violence in the general population without formal intervention. Thirdly, we have to say what we mean by success: is it less violence, no violence, no longer subjecting the partner to a life of fear, or changing the man's attitudes towards women more generally? Best practice in evaluation normally now looks for the second and third of these, with the most weight placed on whether the woman says she now feels safer and has a better quality of life.

Having said all this, a typical finding from the USA is that programmes do have a modest effect in reducing repeat offending (see review in Healey and Smith, 1998). Most studies report successes in reducing overt violence - in the region of 53 to 85 per cent, according to a review by Edleson and Tolman (1992). A more sobering finding comes from the Domestic Abuse Project in Minnesota where, although two-thirds of completers were reported non-violent at an 18-month follow-up, most persisted with controlling or threatening behaviour (Edleson and Syers, 1991). Some men also relapse over time. The longest evaluative project in the literature is that by Edward Gondolf in which a 53 per cent combined success rate was found across four types of programme after 30 months (Gondolf, 1998c). The rest used violence at some point during the 30 months. Only 21 per cent of men were reported by their partners to have been neither verbally nor physically abusive during that time. The most dangerous period was the first six months (nearly half of those who would reassault had first done so by then, and a third within the first three months). A dangerous 23 per cent of the men were repeatedly violent throughout the follow-up and they also inflicted the majority of the worst injuries. Statistically associated with likelihood of reassault were: completing less than three months of a programme, being frequently drunk, having antisocial tendencies, and showing evidence of severe psychological problems (though there may not be a simple causal link). Looking at quality of life, 69 per cent of the women felt 'better off', 84 per cent considered it 'very unlikely' that their partners would hit them and 83 per cent felt 'very safe' at 30 months. Only 12 per cent felt 'worse off'. It may be seen as a success that, by the 30 month follow-up, 60 per cent of the initial partners were no longer living with the perpetrator and a quarter of the total sample had no contact. However, at least a quarter of the latter men were in new relationships.

For those who seek to know what kind of programme works best, Gondolf's frustrating finding was that outcomes from all four programmes were similar. To some extent, the similarities between the interventions may have been greater than their differences. All were linked to the courts, all used cognitive behavioural techniques, all were gender-based in the analysis offered of partner violence, and all were integrated with partner services (Gondolf 1998a and 1998c).

There is very little research to cite from the UK. The most important study is the evaluation of the CHANGE and Lothian Domestic Violence Probation Project programmes in Scotland by Dobash et al. (1996). This found 67 per cent success in avoiding further violence for a year after the programmes, as measured by self-report and partner report, as against only 25 per cent success for men subject to other disposals. Most of the re-assaults in both categories occurred in the first three months: 30 per cent of men in perpetrator groups (very close to Gondolf's finding, see above, though in a far smaller study) and 62 per cent of the others. Violence had not been replaced in programme participants by other controlling behaviours, which had also significantly reduced, and, though female respondents' own quality of life was only slightly better, most reported a consistent improvement in the men's overall behaviour and attitudes and in their relationship together, as against a more typical deterioration amongst men receiving other disposals. Most men were very positive both about the group and about their own hopes for the future.

Conclusion

The fairest conclusion is probably that the jury is still out on whether perpetrators' programmes work (at best, they work for some men) and that it is better to have other reasons for setting them up. For example, it may well be considered immoral to do nothing to challenge violent men to change their behaviour. Doing nothing about violent men leaves abused women to take all the action necessary to guarantee their and their children's safety, which may be virtually impossible in the face of a persistent harasser. The cost of doing nothing is enormous and there is some evidence that perpetrator projects are not unduly costly in comparison with other forms of intervention. Other interventions do not, in any case, currently exist and at least, in establishing a group-based project, there is currently extensive evaluation taking place in the UK and USA, which can be used to inform continuing improvements. Above all, the general public is no longer tolerant of domestic violence, as once it was, and there is arguably a public duty to follow the best available current knowledge and at least do something. Having said that, this section has been permeated with words of caution which can be summed up by reminding ourselves that groups can do more harm than good if enormous care is not exercised in the way they are established and run. No woman must ever be allowed to believe that her partner's involvement in a group gives her a guarantee of safety and no decision about a groupwork project for perpetrators, at any level of policy or practice, should ever be taken without putting the safety of women and children first.