This section focuses on the approaches underpinning intervention programs for perpetrators of domestic violence, and what is known about the effectiveness of these programs. The analysis draws on empirical studies, which have been undertaken largely in North America (the United States of America and Canada) and, to a lesser extent, Europe, with only a small number of studies conducted in Australia. The types of evaluation undertaken and the implications for further research are also discussed.
Intervention programs for domestic violence perpetrators can be broadly characterised as having a psycho-educational approach or a psychotherapeutic approach, or a combination of the two approaches. These intervention programs may be either court-mandated or non-mandated (i.e., involving voluntary participation).
Psycho-educational approaches emerged from the Duluth model (Duluth Domestic Abuse Intervention Project, DAIP), which has become one of the most widely adopted model since the 1980s. The Duluth model is based on an interagency approach that promotes intersectoral collaboration and links to the criminal justice system. It applies a feminist perspective with an educational approach that involves consciousness raising and confronting offenders’ beliefs about masculinity and gendered power relations (see Barner & Carney, 2011; Day, Chung, O’Leary & Carson, 2009; Pence & Paymar, 1993). The Duluth Model combines punishment (for the offender’s violent behaviour) with education to facilitate change in offenders’ attitudes and behaviour.
Psychotherapeutic approaches are derived from the disciplines of psychiatry and psychology, and intervention programs include those that have a psychodynamic, humanist or cognitive-behavioural perspective. Cognitive-behavioural therapy (CBT) has emerged as the most widely used psychotherapeutic intervention approach for a wide range of behavioural problems. CBT is focused on identifying and correcting erroneous or distorted cognitions and skill training, and may involve anger management.
It is noteworthy that, over time, there has been a merging of psycho-educational and psychotherapeutic approaches, most commonly with the amalgamation of CBT and the Duluth Model or feminist (gendered) perspective. As a result, intervention programs for domestic violence perpetrators are increasingly difficult to differentiate based on their theoretical approach (Eckhardt et al., 2006).
In addition, programs that articulate the same intervention approach are not necessarily comparable because the same approach is operationalised in different ways. In the national review of integrated programs for domestic violence perpetrators conducted by Chung et al. (2004), it was found that many organisations identified the Duluth Model as the underpinning for their programs. However, examination of detailed program documentation and observations of program deliver revealed different operationalisation of the Duluth model in practice (Day et al., 2009).
Intervention programs differ not only in approach but also on a range of dimensions, including program duration and intensity, conditions of participation (e.g., court-mandated or voluntary, acceptance of responsibility), policies such as those relating to payment, absence, monitoring and assessment, linkage to the criminal justice system and other related agencies, session structure (e.g., structured or open), facilitator qualifications and experience, contact with the victim, referral sources, funding sources, target group characteristics, and program integrity (e.g., Cissner & Puffett, 2006; Price & Rosenbaum, 2009; Smith, 2009). It is therefore a challenge to assess the relative effectiveness of intervention programs and identify the active components of a program – that is, program components that contribute to improved perpetrator and victim outcomes (e.g., Eckhardt et al., 2006).
While the majority of intervention programs are delivered in a group setting, individual and couples counselling are also used for domestic violence perpetrators.
Group-based interventions are widely accepted as a more appropriate format for changing the attitudes and behaviour of domestic violence perpetrators than individual or couples counselling (e.g., Labriola et al., 2008). A group setting reduces the social isolation experienced by many perpetrators and provides greater opportunities for role-play, feedback and confrontation of existing beliefs (Labriola et al., 2008; Saunders, 2008). However, a group format is limited in its capacity to respond to individual needs, and group dynamics may reduce the willingness to self-disclose and engage among perpetrators who are less motivated to change (e.g., Murphy & Eckhardt, 2005; Murphy & Meis, 2008).
There is a growing view and evidence base that individual or individualised interventions increase the likelihood of program completion, which in turn reduces recidivism (e.g., Murphy & Meis, 2008). Murphy and Meis (2008) argued that individualised programs can be more effective than group-based programs in reducing recidivism because they can be tailored to the motivation and readiness to change for each perpetrator and have greater capacity to set and monitor individual change targets. Individualised programs also increase the likelihood that comorbid conditions (e.g., mental health and substance abuse problems) will be identified and addressed, and minimise potential reinforcement of problematic attitudes and behaviour through negative peer influences.
Notably, the Duluth Model has been increasingly criticised for its confrontational nature, which may decrease program compliance and completion (Bowen, 2010; Day et al., 2009; Sartin et al., 2006). It has also been argued that the focus on structural factors – for example, gendered power relations – as the primary causes of domestic violence can promote a one-size-fits-all approach and intervention programs that are less client-centred (e.g., see Day et al., 2009).In practice, current programs based on the Duluth Model often include individual assessments or sessions.
3.3.2Program evaluation and effectiveness
Intervention versus no intervention
Evaluations of individual intervention programs for men who use violence against women have produced largely mixed evidence about the effectiveness of such programs in reducing recidivism.
Recent meta-analyses of program evaluations generally suggest that intervention programs contribute to a small reduction in recidivism (e.g., Eckhardt et al., 2006; Saunders, 2008). In contrast, meta-analysis of court-mandated interventions for domestic violence offenders (Feder, Wilson & Austin, 2008) indicated no clear positive impact of court-mandated interventions. The effectiveness of intervention programs has been found to vary as a function of characteristics associated with the evaluation methodology as well as the perpetrators (e.g., Saunders, 2008).
Specifically, intervention programs were found to lead to greater reductions in recidivism when official records rather than victim reports of reoffence are used as the outcome measure, and when the evaluation employed a quasi-experimental rather than an experimental design with random assignment and matched control groups (e.g., Eckhardt et al., 2006; Labriola et al., 2008).
In addition, intervention programs were more likely to be completed by, and produce attitudinal and behavioural changes, leading to reduced recidivism for, perpetrators who:
are more motivated or ready to change
have no or fewer previous contacts with the criminal justice system
have no comorbid (e.g., substance use or mental health) conditions
are Caucasian, employed, married, older, and have a higher level of education (e.g., see Edleson, 2008; Jewell & Wormith, 2010; Kingsnorth, 2007; Olver, Stockdale & Wormith, 2011; Sartin et al., 2006).
There is some evidence to show the programs based on a well-established theoretical approach are more likely to be effective than programs (for a review, see Blatch & Delaney, 2009),
These findings suggest that while domestic violence perpetrators are not limited to particular socio-demographic categories, they do not respond to the intervention programs to the same extent. They also support the growing view that, to maximise the effectiveness of intervention programs, programs should be matched to offender characteristics (e.g., Day et al., 2009; Jewell & Wormith, 2010). Intervention programs, however, do not generally address differences in socio-demographic characteristics. A study conducted in South Australia to examine the impact of a perpetrator program in the context of a small rural community (Jamieson & Wendt, 2008) pointed to the need for rural perpetrator programs to address concerns regarding anonymity and confidentiality, and community attitudes that inhibit help-seeking among perpetrators (e.g., through community education). It also suggested the need for a holistic approach to intervention that is sensitive to community norms in a rural context.
Socio-demographic variables such as employment (hence, income) and marital status are also known as stake-in-conformity variables, because people who have social bonds through marriage, employment and stable housing have more to lose when they do not conform to non-violent social norms (e.g., Aldarondo, 2009). This view is consistent with findings demonstrating that perpetrators who are married and employed, and have residential stability (i.e., have a higher stake in conformity) are more likely to complete programs and remain non-violent post-program (e.g., Aldarondo, 2009; Ashcroft et al., 2003).
To date, there are few published evaluation studies of domestic violence offender programs in Australia. While there are some evidence on the effectiveness of Australian programs, the majority of studies did not include a control group.
Early Australian evaluation studies include the Urbis (then Urbis Keys Young) evaluations of the ACT Interagency Family Violence Intervention Program (2000), the ACT Family Violence Intervention Program: Phase II (2001), and the NSW Pilot Program for Perpetrators of Domestic Violence (2004). The evaluation of the NSW Pilot Program found that men who completed the program were less likely to reoffend than those who were withdrawn from the program, with first-time offenders being less likely to reoffend than those with previous domestic violence convictions. The vast majority of men also found the program to be helpful. It is important to note, however, that a significant group of men (albeit, a small proportion of program participants) did come into contact with the police after commencing the program.
A recent evaluation of the NSW Domestic Violence Intervention Court Model (Rodwell & Smith, 2008) found that there was no consistent impact of the program on recidivism across different sites. Nevertheless, victims were found to be generally satisfied with the various processes involved, and other stakeholders generally agreed that the program was successful. No control groups were included in the evaluation. These findings suggest that measures beyond recidivism may hold meaning, but ultimately a lack of recidivism would indicate program failure.
The evaluation of Lifeworks, which provides intervention programs for perpetrators of domestic violence, was conducted by Monash University (Brown & Hampson, 2009). Specifically, this evaluation compared the benefits of Lifeworks’ Men’s Behavioural Change Program (based on the Duluth Model) and its Couple Counselling service. Evaluation findings supported the effectiveness of the Men’s Behavioural Change Program in ceasing or reducing physical violence among program participants. However, it was unclear whether the group-based program was more effective than Couples Counselling because the clients wanted choice and preferred the type of intervention they selected. The researchers concluded that it is important for a service provider to offer a range of services to match client’s needs and preferences. Little is known about the effectiveness of domestic violence perpetrator intervention programs for Aboriginal and Torres Strait Islanders. A review of selected domestic and family violence intervention programs by Loxton, Hosking, Stewartwilliams, Brookes and Byles (2008) found that interventions programs for Aboriginal people are more likely to be successful when they have been developed in consultation with the community, involve interagency collaboration, and are multi-disciplinary and holistic. The need for culturally competent programs for Aboriginal and Torres Strait Islander perpetrators is highlighted by Day, Jones, Nakata and McDermott (2011).
Relative effectiveness of interventions
While offenders who complete an intervention program may be less likely to reoffend, there is no clear evidence that one type of intervention program is more effective than another in reducing recidivism (e.g., Eckhardt et al., 2006; Saunders, 2008). This finding is not surprising given that intervention programs are evolving on an ongoing basis, often based on an amalgamation of approaches, and differ not only in theoretical approach but also on a wide range of program characteristics (e.g., Cissner & Puffett 2006; Smith 2009). Given that domestic violence intervention programs have not had a focus on matched interventions, much is still unknown in terms of ‘what works for whom’.
Saunders (1996) found men with antisocial tendencies obtained better outcomes from cognitive-behavioural counselling, whereas those with more depressive tendencies achieved better outcomes in psychodynamic counselling. Holtzworth-Munroe, Meehan, Herron, Rehman & Stuart (2000) developed four categories of domestic violence offenders: generally violent aggressor, family-only, low level antisocial, and borderline or dysphoric. While domestic violence offenders are not reliably associated with specific personality disorders (e.g., Edleson 2008; Gondolf 2004), other individual characteristics such as criminal history, comorbidity, employment and education have been shown to be related to program completion and recidivism rates.
In a recent study, Jewell and Wormith (2010) found that older offenders were more likely to complete cognitive-behavioural programs and other unspecified programs than feminist psycho-educational programs. In particular, it was found that people with lower levels of education were more likely to drop out of feminist psycho-educational programs. Offenders were more likely to complete feminist psycho-educational programs when program participation was court-mandated. Older offenders were also more likely to complete shorter programs.
These findings highlight the heterogeneous nature of domestic violence offenders and point to the need to match the intensity of the intervention to the characteristics of perpetrators (e.g., Day et al., 2009; Jewell & Wormith, 2010; Murphy & Meis, 2008). Given that high risk offenders tend to have more previous criminal offences, be less motivated to change, have comorbid conditions, and are more likely to drop out, having intervention programs that target high risk offenders and reliable assessment tools to identify high risk offenders are likely to improve outcomes.
The need for an individualised or matched intervention approach is also supported by evidence that program compliance and effectiveness in reducing recidivism are higher when:
the intervention is matched to the motivational readiness to change of the client (e.g., through motivational interviewing strategies)
mental health and substance abuse issues are addressed concurrently (e.g., Day et al., 2009; Eckhardt et al., 2006; Edleson 2008; Olver et al., 2009; Sartin et al., 2006).
A recent study was also conducted by Coulter and VandeWeerd (2009) to assess a multi-level approach to domestic violence perpetrator intervention. The study involved a three-level approach to intervention programs, with program intensity based on perpetrator characteristics (e.g., previous violent convictions, history of mental health issues, relationship to victim(s), and nature and severity of most recent offence). The results showed that program completion rates were much higher than most existing programs, and recidivism was substantially lower among program completers than non-completers. These findings applied to all program levels. It is worth noting, however, that there were no matched control groups in the study.
In line with these findings, domestic violence researchers are increasingly pointing to the importance of applying the risk, need and responsivity (RNR) principles, which have been more widely used in sexual assault and other offender programs. It has been found that sexual offender intervention programs that adhered more closely to the RNR principles were more effective in reducing recidivism (Murphy & McGrath 2008; Olver et al., 2011).
The risk principle refers to the match between the intensity of treatment to the risk level of the offender, and points to the use of valid assessment tools. Based on the need principle, effective [?] treatment programs should address offender’s psychological, social and emotional functioning linked to the development and continuation of criminal behaviour (i.e. criminogenic needs such as attitudes supportive of crime, delinquent peers, substance abuse, unemployment). The responsivity principle postulates that effective treatment should be cognitive behavioural in nature (general responsivity) and tailored to the learning style, cognitive capabilities, motivations, personality and cultural background of the offender (specific responsivity).
In order to increase program completion rates through a better matching of intervention strategies to perpetrator characteristics, Day et al. (2009) and others have suggested that the Transtheoretical Model of Change (Casey et al., 2005) could be applied to the design of domestic violence perpetrator intervention programs. In other words, intervention strategies could be matched to the perpetrator’s particular stage of readiness to change (i.e., pre-contemplation, contemplation, preparation, action, maintenance, termination). For example, at the contemplation stage, consciousness-raising strategies may be used to prepare offenders for subsequent participation in the intervention program; at the preparation stage, the focus may be on making a commitment to change; and at the action stage, the strategies may be to facilitate the development of coping skills.
Other factors influencing program effectiveness
Other program factors that have been associated with program effectiveness:
There is some evidence that court-monitoring of program attendance can improve program completion, which is associated with reduced recidivism (Gondolf, 2004). However, others have found that monitoring did not make a difference (e.g. Labriola et al., 2008).
Edleson (2008) argued that programs that are part of a coordinated effort within the criminal justice system are more likely to achieve better outcomes (reduced dropout and recidivism rates) when:
offenders receive intervention within two to two and a half weeks of their arrest
offenders are monitored on an ongoing basis in relation to their treatment compliance monitoring of compliance with treatment requirements
courts responded quickly to those who do not comply.