«Evaluation of Three Court-Mandated Family Violence Interventions: FVEP, EXPLORE, and EVOLVE Stephen M. Cox, Ph.D. Professor Pierre M. Rivolta, Ph.D. ...»
In 2013, Public Act 13-247 An Act Implementing Provisions of the State Budget mandated, under section 53(a), that an evaluation be conducted to “…assess the effectiveness of programs maintained by [CCSD] with respect to family violence...” (i.e., FVEP, EXPLORE, and EVOLVE). Moreover, the law specified that “such assessment […] consider findings from the Pew-MacArthur Results First Initiative’s cost-benefit analysis model [to] determine whether any program changes may be implemented to improve the cost-effectiveness of such programs.” The current research was conducted pursuant to the legislative requirements outlined in these recent public acts. Faculty from the Institute for the Study of Crime and Justice at Central Connecticut State University were contracted to evaluate the effectiveness of the Judicial Branch’s three court-mandated family violence interventions: FVEP, EXPLORE, and EVOLVE.
This report summarizes our evaluation and is organized into sections describing (1) the literature relevant to this research project, (2) the nature of the three family violence interventions, (3) the research methodology and data used to conduct this evaluation, (4) the outcome analyses, and (5) the conclusions of the evaluation and recommendations for future programming.
This report is available at http://www.housedems.ct.gov/dv/pubs/020612/JB_Report_on_FVEP_and_DVDockets.pdf Court-Mandated Family Violence Interventions Central Connecticut State University
The following sections present a brief survey of the literature relevant to this evaluation.
The review begins with an overview of family violence statistics (at the national and state levels) to situate the study in its context. This section is followed by a discussion of existing standards that frame family violence interventions. The review concludes with a summary of recent studies evaluating the effectiveness of family violence interventions.
National and State Statistics Regarding Family Violence
To this date, family violence remains a significant social problem in the United States.
National statistics illustrate the gravity of this issue. According to the Bureau of Justice Statistics (BJS), between 1998 and 2002 roughly 3.5 million violent crimes (about 11% of all reported and unreported violence) were committed against family members (BJS, 2005). In 60% of these cases, the victim was either a spouse or an offspring of the perpetrator; approximately 75% of victims were female (BJS, 2005). More recent data from the National Crime Victimization Survey (NCVS) show that around 907,000 individuals (or 3.6 per 1,000) were victims of intimate partner violence in 2010; in 80% of cases, the victim was female (Catalano, 2012).
Findings from the National Victimization Against Women Survey (NVAWS), conducted in 1995-1996, shows that almost 75% of the women who reported being raped or physically assaulted since the age of 18 were victims of a current or former husband, partner, boyfriend, or date, thus illustrating the intimate partner nature of violence against women (Tjaden & Thoennes, 2000). According to the BJS (2005), family violence cases were reported to the police in less than 66% of cases. Progress has been made, though, in reducing the prevalence of family and intimate partner violence over the years. For instance, NCVS data shows that the overall rate of intimate partner violence (age 12 and older) decreased 64% between 1994 and 2010, from 9.8 victimizations per 1,000 to 3.6 per 1,000 (Catalano, 2012).
As discussed earlier, the number of family violence incidents in Connecticut has remained stable over the past two decades, with about 19,000 to 21,000 incidents occurring each year. In 2011, 20,494 family violence incidents led to an arrest. In the wide majority of cases (17,782 or about 87%), the arrest involved a charge of assault, breach of peace or disorderly conduct. Eighteen of these cases were classified as homicides. These 20,494 incidents involved 42,982 individuals, including 18,132 victims, 16,644 arrested offenders, and 8,206 participants.
Approximately 20% of all incidents included a dual arrest. Regarding the gender of involved parties, about 75% of victims (13,471) were female and 25% (4,661) were male. In contrast, slightly over 75% of offenders (12,785) were male and less than 25% (3,859) were female.
Approximately 60% of all participants were between the ages of 21 and 44, and about 80% were under 45 years old. A weapon was involved in 74% of all incidents and a serious injury was inflicted in about 40% of cases. That year, Windham County had the highest prevalence of family violence crimes (744 incidents per 100,000 inhabitants) and Tolland County had the lowest prevalence (302 per 100,000 inhabitants). 6 It is important to note that these statistics represent cases known to the police. The Connecticut Coalition Against Domestic Violence See DESPP’s 2011 Family Violence Arrest Report, available at http://www.dpsdata.ct.gov/dps/ucr/data/2011/2011%20Family%20Violence%20Arrest%20Report.pdf
(CCADV) reports that 56,178 victims received services through their 18 member agencies in 2011, 7 thus emphasizing that many victims chose not to report their victimization to law enforcement authorities.
Standards Regarding Types of Batterers’ Interventions Programs A wide range of criminal justice interventions exist that are designed to deal with perpetrators of family violence (i.e., “batterers”). The types of batterers’ interventions vary based on the general theoretical perspective used to explain the etiology of family violence perpetration. Such perspectives include the feminist view, the intergenerational transmission of violence, and various psychological views (Corvo, Dutton, & Chen, 2008). Related interventions include feminist psycho-educational men’s groups, cognitive-behavioral men’s groups, anger management, and couples’ therapy. One clear problem with these many interventions is that only a handful have been “subjected to rigorous empirical test” (see Babcock, Greene, & Robie, 2004, p.1025).
Forty-four states and territories have enacted statutory standards regarding the various interventions that can be used to deal with family violence perpetrators (Maiuro & Eberle, 2008).
The scope and content of these standards tend to “vary according to the administrative bodies and the means of regulation” (Maiuro & Eberle, 2008, p. 135). Standards typically specify the various protocols used for screening and assessment; the content, modality, and length of programming; the training and education requisite to administer such programs; and whether sharing findings and performing evaluations are required (Maiuro & Eberle, 2008). The Washington State Institute for Public Policy (2013) followed this study and found that 44 states had legal guidelines for domestic violence treatment. Connecticut is one of those six states (along with Arkansas, Mississippi, New York, South Dakota, and Wyoming) that do not have statutory standards regarding their court-mandated interventions for perpetrators of family violence.
Some common trends can be discussed. For instance, according to the survey of state standards conducted by Maiuro & Eberle (2008), the primary theoretical and conceptual orientation of domestic violence treatment rests upon a combination of power and control and social psychological conceptualization of family violence. Regarding the length of treatment, the majority of the state standards require a minimum of 24 to 26 weeks (with a once a week session lasting 90 minutes). The primary choice of intervention is group therapy. Specifications of assessment requirements include some type of victim contact, a determination of formal risk or lethality, and a screening for substance abuse problems (including alcohol) (Maiuro & Eberle, 2008). In addition, the Washington State Institute for Public Policy (2013) survey found that 28 states require programs to center on power and control dynamics (this is commonly referred to as the “Duluth model” of domestic violence treatment), 12 states have broad curriculum mandates, and 4 states require intake and assessment of domestic violence cases but do not mandate specific types of treatment.
While there are definite advantages to having such standards, Maiuro and Eberle (2008, p. 148) point out that the wide majority of these standards “were composed without authoritative See http://www.ctcadv.org/information-about-domestic-violence/statistics/ for more information.
Court-Mandated Family Violence Interventions Central Connecticut State University references or, on occasion, [based on] very dated sources.” Other researchers also lament the lack of empirical grounding of such standards. For instance, Corvo and colleagues (2008, p. 111) argue that policies and practices regulating interventions with domestic violence offenders “have enjoyed immunity from the external, empirical accountability” that can now be found in evidence-based practices (EPBs).
Effectiveness (or Lack Thereof) of Batterers’ Interventions Programs
Court-mandated domestic violence programs began appearing in the early 1980s. Soon thereafter, evaluation studies measuring their effectiveness followed (see Feder & Wilson, 2005).
This early evaluation research showed positive results and indicated that programs were successful in reducing recidivism (i.e., acts of subsequent domestic violence) among the treated population. However, these results may have been overly optimistic given that studies comprising this early wave of evaluation research suffered from several methodological flaws (e.g., small sample sizes, focus on program completers instead of intent-to-treat population; lack of appropriate comparison groups; unreliable outcome measures) (see discussion in Davis & Taylor, 1999, and Feder & Wilson, 2005). The next wave of research (i.e., since mid-1990s) provided a different picture of the impact—or lack thereof—of batterers’ interventions programs (Corvo et al., 2008; Feder & Wilson, 2005; Rosenfeld, 1992).
A review of the recent literature reveals a preponderance of evidence against the effectiveness of batterers’ intervention programs. To date, very little support exists for the notion that court-mandated programs help reduce future incidence of domestic abuse; at best, metaanalyses of methodologically sound studies reveal a minimal impact of batterers’ intervention and small overall effect sizes (Arias, Arce, & Vilarino, 2013; Babcock et al., 2004; Feder & Wilson, 2005; Miller, Drake, & Nafziger, 2013; Davis & Taylor, 1999). In addition, research also shows “an inverse relationship between design rigor and likelihood of finding program effectiveness” (Feder & Wilson, 2005, p.243). In other words, the more rigorous the design the lesser the likelihood of finding program success.
Davis & Taylor (1999) conducted the first extensive synthesis of the literature on this subject. Their review painted a somewhat optimistic picture of the effectiveness of batterer intervention programs. Overall, the authors concluded the effect of treatment was substantial (h = 0.41) when focusing on six methodologically rigorous studies only (i.e., two true experiments and four quasi-experiments). Interestingly, Davis and Taylor found the average effect size of the two true experiments was almost identical to that of the four quasi-experiments. In addition, the authors did not find any difference in program effectiveness related to treatment modality or length of treatment intervention (Davis & Taylor, 1999).
Babcock and colleagues’ (2004) meta-analytic review included 22 studies—5 true experiments, and 17 quasi-experiments. These studies reported on the effect of a wide range of group-treatment modalities (Duluth model and CBT, among others) for male perpetrators of domestic violence. The review included research design and treatment type as main moderators.
The authors reported an overall small effect size of the batterer intervention on subsequent domestic violence reoffending. No differences in effect sizes were observed across treatment modalities—in other words, no one treatment type worked better than the other (specifically when comparing Duluth model vs. CBT interventions). Similarly, no difference in effect size Court-Mandated Family Violence Interventions Central Connecticut State University was observed when comparing recidivism outcomes (i.e., police records vs. victim reports).
Studies following a quasi-experimental design showed greater effect sizes than those following a true experimental design, although the differences in effect sizes were not significant. Overall results from this analysis pointed to a minimal impact of current group-based batterers’ interventions.
Feder & Wilson (2005, p.239) performed a meta-analysis to determine whether “courts can affect abusers’ behavior” through legally mandated interventions. Their review of evaluations purported to identify and assess “the most rigorous research on court-mandated batterer intervention programs” (p.242). As such, only studies utilizing an experimental or rigorous quasi-experimental design—that is, those studies that established baseline equivalence between the treatment and comparison groups—were included in the meta-analysis. This yielded 10 studies. It is worthy to note that evaluations also had to focus on a post-arrest court-mandated intervention specifically targeting domestic violence offenders; pre-trial diversion interventions were included as well. The authors provided effect sizes as a function of outcome type and of design types; overall, the results were mixed. Experimental studies showed a modest effect size when using official reports as the outcome measure, but no statistically significant effect when using victim reported outcomes. Quasi-experimental studies with a no-treatment comparison group showed an overall deleterious effect of treatment on outcomes, while quasi-experimental studies that used a dropout comparison group showed an overall positive effect of treatment (Feder and Wilson, 2005). These findings highlight the inverse relationship between the likelihood of treatment outcomes and stringency of research methodology previously discussed.