Changes in Intimate Partner Violence Among Women Mandated to Community Services

Abstract

Increasingly, female victims of intimate partner violence (IPV) are charged with IPV perpetration and mandated by courts or child protective services to receive domestic violence services. A critical need exists for evidence-based interventions targeting the needs of this unique population, but such research is scarce. To address this gap, we evaluate a novel intervention developed by two community-based agencies and delivered to 70 female IPV victims who are primary caregivers for children and mandated to services. Using a quasi-experimental design, IPV perpetration and victimization data are collected at three time points: baseline, program completion, and 3-month follow-up. Analyses use binary logistic regression to control for clustering and to incorporate time-varying covariates. Results show statistically significant reductions in both IPV victimization and perpetration at program completion. This study also provides evidence for the feasibility of recruiting and retaining this vulnerable group of IPV victims in research studies conducted in community-based settings.

Keywords

intimate partner violence justice involved child welfare mandated intervention evaluation binary logistic regression

Intimate partner violence (IPV) is a serious problem affecting many families in the United States. More than one in three women experience lifetime rape, physical violence, or stalking by an intimate partner, and about one in four women experienced severe physical IPV (e.g., hit with a fist, beaten, or thrown against a wall; Black et al., 2011). Research has shown that as a consequence of these violent experiences, female IPV victims often struggle with serious physical and mental health problems (Campbell, 2002), as well as harm to their families, financial situations, and social relationships (Briere & Jordan, 2004).

The harm caused by IPV often extends beyond the perpetrator and victim, with children in the household the most frequent “collateral damage” of IPV. Upward of 17 million U.S. children are affected by IPV annually (Holden, 1998). Children’s exposure to IPV is associated with many significant negative outcomes in critical developmental areas (e.g., academic, behavioral, and social; Evans, Davies, & DiLillo, 2008; Kitzmann, Gaylord, Holt, & Kenny, 2003). If left unaddressed, childhood exposure to IPV can lead to behavioral and psychological problems during adolescence (e.g., antisocial behavior, relationship aggression; Ireland & Smith, 2009) and adulthood (e.g., depression, alcohol dependence, and next generation child maltreatment; Roustit et al., 2009; Russell, Springer, & Greenfield, 2010). In addition, an important overlap exists between child maltreatment and IPV perpetrated against women (Edleson, 1999). Children living in families with IPV are at an increased risk of various forms of maltreatment, particularly neglect and physical abuse (Hamby, Finkelhor, Turner, & Ormrod, 2010; Zolotor, Theodor, Coyne-Beasley, & Runyan, 2007).

Unfortunately, little of the extant research has examined IPV interventions to help female IPV victims and their children (Rizo, Macy, Ermentrout, & Johns, 2011; Stover, Meadows, & Kaufman, 2009). Moreover, the few studies that have investigated the subpopulation of IPV victims with children are marked by substantial methodological limitations because few have used rigorous study methods, such as systematic quasi-experimental designs or randomized controlled trials (RCT; Sullivan, 2011; Wathen & McMillian, 2003). As such, the evidence base about how best to help women and children who struggle with IPV in their families is woefully inadequate.

Despite the evidence gap around domestic violence services, community-based agencies throughout the United States offer various safety-focused interventions to IPV victims and their children (Macy, Giattina, Sangster, Crosby, & Montijo, 2009; Riger et al., 2002). The demand for services is astounding: During one 24-hr period in 2011, domestic violence agencies across the country provided services to more than 67,399 victims (National Network to End Domestic Violence, 2012). These services are based on the empowerment philosophy, which entails that providers (a) collaboratively involve victims in service planning; (b) emphasize victims’ strengths and resources; and (c) provide information and teach skills to enhance victims’ self-efficacy (Busch & Valentine, 2000). Typically, domestic violence services are structured for victims who voluntarily seek help. Thus, service providers do not usually expect to work with IPV victims who are mandated to attend services.

Contrary to usual expectations, two agencies in Raleigh, North Carolina, InterAct and SAFEchild, noted a dramatic increase in the numbers of IPV victims who were directed to attend agency services by either child protective services (CPS) or the county court system. In light of this growing population of victimized women and their children, the two agencies (SAFEchild is a child abuse agency and InterAct is a domestic violence services agency) joined in a collaborative effort to create and implement a program specifically designed for female IPV victims who had been mandated to their services and who were primary caregivers of young children.

As stated, little existing research has examined how best to help women and children who struggle with IPV in their families. Even less research has examined services for IPV victims mandated by CPS or the courts to attend community-based domestic violence services (Rizo et al., 2011). This knowledge gap presents a considerable challenge to providers working to tailor programs to the unique circumstances of these IPV victims and their children. To address this gap, our research team partnered with these two community agencies to investigate their novel, community-developed program.

Why Are Some IPV Victims Mandated to Attend Services?

CPS Mandates

Children in families with IPV are at an increased risk of maltreatment. A growing number of studies have shown that many households that are involved with CPS also experience IPV within the family (Casanueva, Martin, & Runyan, 2009; Kohl, Edleson, English, & Barth, 2005); therefore, child maltreatment researchers have appealed for CPS systems and workers to address IPV (e.g., Hamby et al., 2010; Hazen, Connolley, Kelleher, Barth, & Jandsverk, 2004). In addition, the seriousness of exposing a child to IPV has been recognized by 22 states and Puerto Rico, which have enacted statutes that make committing IPV in the presence of a child a separate crime in addition to other IPV-related charges (U.S. Department of Health and Human Services, 2009). Therefore, to better ensure children’s health, safety, and well-being, CPS workers are giving increasing attention to IPV in families (Schechter & Edleson, 1999).

Typically, CPS workers address and prevent child maltreatment by providing education, information, and support through home-based services focused on the family. In addition, CPS workers direct families to other community agencies to address particularly severe and specific problems, such as IPV. Thus, many IPV victims involved with CPS might seek help from domestic violence agencies at the direction of CPS workers.

Court Mandates

Proarrest statutes for IPV perpetration were put into effect on a national scale in the 1980s, and since that time, arrests of women for IPV have raised in an average of 25–35% throughout U.S. communities, with women representing 8–20% of all IPV arrestees (Rajan & McCloskey, 2007). Although some women are primary IPV perpetrators (Rajan & McCloskey, 2007; Simmons, Lehmann, & Collier-Tenison, 2008), a growing body of evidence has indicated many women arrested for perpetration also report IPV victimization (Henning, Renauer, & Holdford, 2006; Simmons et al., 2008). Consistent with this evidence, other researchers have shown women arrested for IPV perpetration have reported they often felt fearful or powerless and used violence against male partners to protect themselves, to defend their children, or to retaliate for prior abuse (Stuart, Moore, Hellmuth, Ramsey, & Kahler, 2006; Swan & Snow, 2006). In addition, some women might be arrested and become involved with the justice system even though they have not perpetrated IPV. For these women, justice system involvement is often the result of their accepting blame for IPV in an effort to avoid retaliation from a male perpetrator or the consequence of stringent dual arrest policies implemented in some areas (Busch & Rosenberg, 2004; Miller, 2001). Following the arrest and adjudication of these female IPV victims, instead of sentencing these women to jail, they are likely to be court mandated to attend community-based services (Miller, 2001; Simmons, Lehmann, & Dia, 2010).

The Mothers Overcoming Violence Through Education and Empowerment Program

To address the needs of the growing number of women mandated to receive services, two community-based agencies collaborated on the creation and implementation of the Mothers Overcoming Violence through Education and Empowerment (MOVE) program that was designed specifically for this vulnerable and understudied population. Notably, the providers at InterAct and SAFEchild collaboratively (a) identified the need for this program, (b) conceptualized the program philosophy, (c) developed the program, and (d) created the program’s intervention strategies.

Given the lack of research and services for this subpopulation of IPV victims, providers at these two community-based, nonprofit agencies developed MOVE based on their experiences working with families struggling with IPV, their professional expertise, and existing parenting curricula developed for families experiencing IPV (Turner, Gilbert, Hendricks, & Demaree, 2006). Guided by these resources, the agencies implemented MOVE as a 13-session IPV safety and parenting program that was delivered as weekly sessions in conjunction with therapeutic support-group services for participants’ children.

To be eligible for MOVE, a woman had to meet the following criteria: (a) mandated to receive domestic violence services by the county CPS or court; (b) self-identified as the mother (biological, adoptive, foster, or stepmother) or other primary caregiver to a minor child or children; and (c) assessed as an IPV victim and not a primary IPV abuser. Screening for eligibility criteria was determined during the program intake through a detailed assessment. The assessment was conducted by a master’s level social worker (MSW) with professional experience in the area of family violence. The intake assessment included appraisals of the women’s histories of violence, specifically issues of IPV.

The program used a group modality in which six to nine participants met one evening per week. The group meetings were coled by two social workers (one MSW and one bachelor’s level) with experience in group treatment and family violence. The agency providers had purposefully selected the group format for delivering the MOVE intervention to (a) provide an opportunity for peer support and mutual learning, (b) provide participants with a safe place and the opportunity to practice new knowledge and skills, (c) normalize participants’ experiences with IPV, and (d) diminish social isolation. Over the 13 sessions, the group leaders delivered content on parenting, IPV, and safety. The theoretical framework of MOVE was informed by Bandura’s (1991) social cognitive theory and the empowerment philosophy used in the delivery of IPV safety services for victims. As implemented, the program emphasized participants’ strengths by encouraging the women to serve as models for their peers and by encouraging self-assessment of supports, skills, and resources. Consistent with MOVE’s social cognition framework, the program aimed to ensure that the participants possessed the three requirements for change: behavioral capability, expectations, and self-efficacy. To facilitate participants’ learning, the service providers also used intervention strategies based on social cognitive theory, including modeling and reinforcements.

All program meetings were held at the child abuse prevention agency. Program amenities included (a) child care for young children (birth to 4 years) and therapeutic support group services for older children (5–13 years); (b) dinner for mothers and children; (c) transportation to and from the program, as needed; and (d) security services (i.e., an off duty female police officer) to ensure safety of staff, mothers, and children. The program was provided to mothers and their children at no cost. By offering the program free of charge and including the amenities, the providers aimed to model a positive, safe environment for families that would encourage program engagement and participation. Additional information on important aspects of both the MOVE mothers’ and children’s program content and delivery is available in other publications (Ermentrout, Rizo, & Macy, in press; Macy, Ermentrout & Rizo, 2012).

Research Aims

This research was conducted in collaboration with the two community agencies that developed and implemented the program, InterAct and SAFEchild. This work was guided by intervention science recommendations for increased collaborations between investigators and providers toward developing programs that are both feasible and sustainable for use by community-based agencies (Pachankis & Goldfried, 2007). In addition, the team’s work was guided by a growing consensus among intervention researchers on the steps for development of an evidence-based practice: (a) conceptualizing an intervention and designing a treatment manual, (b) conducting feasibility testing, (c) conducting pilot tests, (d) conducting randomized efficacy trials, (e) conducting effectiveness trials, and (f) adapting the intervention for dissemination across diverse settings and communities (Fraser & Galinsky, 2010). Our research team collaborated with the two agencies delivering MOVE on a multiyear study with the goal of completing the first three steps for evidence-based practice development. Our overall study aims were (a) to establish a protocol for implementing MOVE in a community-based setting; (b) to determine if MOVE was feasible to deliver and acceptable to participants; and (c) to garner pilot data to inform future research. Our team was particularly concerned with the practicality of conducting intervention research with this vulnerable group of women (i.e., CPS and court-involved IPV victims) in a community-based setting. Our research on the program development and feasibility has been published elsewhere (Ermentrout et al., in press; Macy, Rizo, & Ermentrout, in press). This article presents the findings from our pilot test of MOVE.

Our team collected data from program participants at three time points: program entry, program completion, and 3 months post program completion. To assess the range of participants’ possible changes over the course of their program involvement, we collected data on experiences with IPV victimization and perpetration, mental health status (e.g., depression, posttraumatic stress), parenting beliefs and practices, and demographic information. In this study, we were interested in exploring two specific research questions. First, relative to program entry, does participation in MOVE effect the likelihood of participants reporting IPV victimization and perpetration at program completion and at the 3-month follow-up? We hypothesized that MOVE participants would be less likely to report IPV victimization and perpetration at program completion and 3-month follow-up relative to program entry. Our second research question focused on whether participant characteristics (e.g., age, education, employment, number of children, relationship status with abusive partner) were related to participants’ experiences of IPV victimization and perpetration. Given the scarcity of intervention research conducted with CPS and court-involved IPV victims, we did not have specific a priori, directional hypotheses for the second research question.

Method

Participants

Following CPS or court referral, all women who enrolled in the MOVE program between January 2009 and July 2011 were invited to participate in the evaluation study. During this time, MOVE was delivered to 10 cohorts of women as a group intervention. Of the 89 women invited, 73 (82%) agreed to research participation; however, 3 of these women did not complete the program. Data presented are based on the 70 (78.6%) women who completed the program. The MOVE service providers defined completion as attendance at no less than 10 of the 13 program sessions. The service providers determined this completion criteria based on their prior experiences providing group interventions to IPV perpetrators and victims and based on the length of existing parenting curricula developed for families experiencing IPV (Turner et al., 2006). An analysis of differences between women who did and did not complete MOVE based on demographics, IPV victimization, and IPV perpetration data collected at baseline showed that there were significant differences between the two groups only on the number of children living with the participants (t = 2.98, df = 69). Women who completed the program had an average of 1.60 (SD = 1.12) children living with them, while the three women who did not complete MOVE had 2.00 children living with them (SD = 0.00).

Procedures

All study methods were reviewed by the following groups to ensure that the research would be acceptable, feasible, and beneficial: (a) staff at InterAct and SAFEchild who developed and implemented MOVE; (b) the study’s community advisory board, which was comprised of service providers and professionals in the areas of child protection and domestic violence; (c) a group of MOVE program graduates who completed the program before January 2009; and (d) the institutional review board of the research team’s university. In light of the unique safety issues and vulnerabilities of the study participants, our team also obtained a National Institute of Health Certification of Confidentiality before beginning the research.

Potential participants were presented with information about the research study by one of the MOVE program facilitators. During the intake interview, the facilitator screened for study eligibility criteria and provided eligible women with a brochure and overview of the research study. Women who were interested in study participation were directed to an on-site private office where they met individually with a research team member who provided an in-depth study description. Each woman was informed that study participation included the completion of questionnaires at three time points (i.e., entry, program completion, and 3-month follow-up) and that questionnaires could be either self-administered or read aloud by a research team member.

To minimize coercion or undue influence during the consent process, all participants were reminded both verbally and in written study information that they were voluntary participants who had the right to decline to participate and the right to refuse to respond to any question. Participants were also reminded verbally and in writing that their decision regarding whether to participate would not affect the services they were currently receiving or may receive in the future from InterAct or SAFEChild.

To reduce the burden of study participation, transportation and child care were offered to participants who scheduled a meeting with a research team member to complete the questionnaires. Participants who opted to self-administer the questionnaires were provided with postage and mailing materials to return completed questionnaires to the research team. To acknowledge their time and contribution to the study, participants received a gift card to a discount department store each time they completed a questionnaire. After providing informed consent, participants completed the program entry (baseline) self-report questionnaire.

Retention

Multiple strategies were used to increase retention of study participants, including (a) making multiple contacts with a participant using the contact method she had identified as her preferred and safest contact; (b) providing research supports (e.g., transportation, child care, and postage); and (c) providing multiple ways for completing the study packets (e.g., in person, at home, over the phone). These strategies yielded a study retention rate of 100% (n = 70) at program completion, and 85.7% (n = 60) at the 3-month follow-up. Analyses examining differences between the women who did and did not complete the 3-month follow-up data collection based on demographics, IPV victimization, and IPV perpetration data collected at baseline showed that there were no significant differences between the two groups.

MOVE Program Delivery

One SAFEchild staff member who had primary responsibility for developing, staffing, and delivering the program trained all MOVE service providers (n = 7) in program delivery. For 8 of the 10 MOVE cohorts, this primary MOVE program manager/service leader delivered and led the MOVE mothers’ groups along with a cofacilitator. In two instances, the primary service leader was unavailable to deliver MOVE because of scheduling. In these instances, another MOVE service leader was recruited by SAFEchild to be the lead group facilitator. This secondary MOVE service leader was trained by the primary service leader and had coled six groups with the primary leader before leading a group on her own (along with another SAFE child trained cofacilitator).

Before implementing this pilot phase of the MOVE research in which we collected outcome data, our team completed feasibility research on the program (Ermentrout et al., in press; Macy et al., 2012). The feasibility phase of the research lasted for approximately 1 year in which MOVE was delivered to four cohorts. Our team also used this feasibility research phase to standardize MOVE, as well as to develop a pilot MOVE manual that included fidelity instruments (Fraser & Galinsky, 2010). MOVE providers then used the manual and fidelity instruments developed in the feasibility research phase during this pilot research to ensure that MOVE was delivered in a consistent manner across all groups and providers.

Measures

Outcome Measures

The Revised Conflict Tactics scales (CTS2; Straus, Hamby, McCoy, & Sugarman, 1996) were used to assess each participant’s experiences of IPV victimization and perpetration at three periods: (a) during the 12 months prior to her program entry, (b) the 3 months of program participation, and (c) at the 3-month postcompletion follow-up. CTS2 is a widely used measure of partner violence that has been shown to possess adequate reliability and validity (Straus et al., 1996).

Measures of victimization and perpetration were assessed using four CTS2 subscales: Psychological Aggression (8 items); Physical Assault (12 items); Sexual Coercion (5 items); and Injury (6 items). Based on feedback from the group of MOVE graduates, the current study dropped 2 items from the Sexual Coercion subscale (i.e., “My partner/I insisted on sex when I/my partner did not want to [but did not use physical force]” and “My partner/I insisted I/my partner have oral or anal sex [but did not use physical force]”). The MOVE graduates strongly agreed that the 2 items could be off putting to research participants. Prevalence scores were calculated for each of the subscales according to CTS2 guidelines (Straus et al., 1996). Specifically, each subscale score was calculated as a binary variable (1 = participant reported one or more of the acts in the respective subscale; 0 = participant did not report any instance of the acts in the respective subscale).

Covariates

The analysis used covariates to account for important participant characteristics including age (in years), education, income, number of children at home, employment, race, referral source, and relationship to abusive partner. All covariates were assessed at program entry with one exception; the covariate for relationship to abusive partner was assessed at each time point. Education was assessed using two indicator variables: (a) graduated or attended college/technical school (coded 1) and otherwise (coded 0); and (b) completed or attended graduate degree program (coded 1) and otherwise (coded 0). The reference group for education was high school/general equivalency diploma (GED), or below, which was an aggregation of responses indicating those who completed Grades 6 through 9, Grades 9 through 11, or high school/GED. Income was assessed using two indicator variables: (a) government assistance (coded 1) and other (coded 0); the other category included responses options for no source of income, used income from another’s employments, and pieced together income from multiple sources. The reference group for income was personal employment income. Number of children in the home was assessed with two variables: (a) number of children (any age) currently living with the participant and (b) number of children 5 years or younger currently living with the participant. Race used one indicator variable African American (coded 1) with all other responses categorized as other (coded 0); the reference group for race was White and other racial groups. Employment was assessed with two indicator variables: (a) part time (coded 1) or other (coded 0); and (b) unemployed or full-time homemaker (coded 1) or other (coded 0). The reference group for employment was full time. Referral source was assessed using two indicator variables: (a) CPS (coded 1; 0 = otherwise) and (b) court referral (coded 1; 0 = otherwise); the reference group for referral sources was both CPS and court. Whether the participant was in a current relationship with the abusive partner was assessed at each of the three data collection time points using yes/no responses (yes = 1; no = 0).

Analytic Plan

This pilot investigation of the MOVE program used a quasi-experimental design best represented by the following:

O_{1} X O_{2} O_{3},

where O ₁ represents the program entry data collection, X indicates the MOVE intervention, O ₂ represents the data collection at program completion that is 3 months post the program entry, and O ₃ represents data collection at the 3-month follow-up. Consistent efforts were made to implement the intervention and collect data from all participants within this time frame. Based on the design, we created a time variable in months such that Time = 0 if the data were collected at O ₁, Time = 3 if the data were collected at O ₂, and Time = 6 if the data were collected at O ₃. Using this design, the participants’ program entry data serve as their own control. That is, we compared a participant’s outcome at a postintervention point (i.e., O ₂ or O ₃) with the same outcome at a program entry (i.e., O ₁).

First, however, we assessed for data clustering to determine an appropriate analytic approach. Data clustering is present whenever data are nested (e.g., data from various time points nested within participants). The presence of clustering effects violates independent-observation assumption embedded in most multivariate models. For this study, because all outcome variables are binary, we used the following equation (Snijders & Bosker, 1999) to estimate the intraclass correlation coefficient (ICC):

I C C = \frac{ψ}{ψ + π^{2} / 3},

where ψ is the variance of the random effect u _0j from the following multilevel logistic regression:

l n (\frac{P_{i j}}{1 - P_{i j}}) = γ_{00} + u_{0 j}, u_{0 j} \~ N (0, ψ), a n d π = 3 .1416 .

The study data posed two potential types of clustering. The first type was clustering of the IPV outcomes at different time points by study participants, and the second type was clustering of participants by intervention groups. Results of the ICC evaluation showed that the first type of clustering was present in almost all study variables, and the second type was absent. Given these findings, we concluded it was important to use a statistical method that could efficiently control for temporal clustering effects.

Due to the small sample size and a large number of important covariates, we applied a binary logistic regression with a robust estimator of standard errors. We selected this approach for three reasons. This method (a) effectively controls for clustering, so that the significance test is accurate and efficient (Guo, 2005); (b) runs a low risk of having a model that fails to converge (Stata, 2007); and (c) effectively incorporates time-varying covariates (e.g., the MOVE participants’ relationship with their abusive partners at each data collection point) into the analysis to make the investigation about the outcome change and its correlates dynamic (Guo, 2005).

The logistic regression model for this evaluation is expressed as follows:

l n (\frac{P}{1 - p}) = X β,

where P is the probability of Y = 1 for the IPV victimization or perpetration outcome under study, β is the vector of estimated regression coefficients showing effects of the independent variables, and X is the matrix of independent variables including the key study variables and control variables. Specifically, the X matrix included: a time variable in months to represent the three data collection time points, a dichotomous variable to indicate a postintervention point (1 = outcome is collected at either O₂ or O₃ ; 0 = outcome is collected at O₁ ), and all covariates detailed earlier. Thus, the coefficient of postintervention served the primary interest of the evaluation. Whether the coefficient for the postintervention variable was statistically significant in the hypothesized direction helped to answer our first research question (i.e., whether participants were less likely to report IPV victimization and perpetration at any program completion time point relative to program entry). In other words, we were investigating the data to determine if statistically significant reductions occurred in the number of participants reporting IPV victimization and perpetration from program entry to program completion. Taking the exponent of each element in β (i.e., exp(β)), we obtained the odds ratio for each independent variable, which is the final statistic presented in Tables 1 and 2 and used to address our second research question. To correct for clustering effects, we used the Huber and White correction to obtain robust estimation of standard errors in the logistic regression analysis (Huber & Ronchetti, 2009). All analyses were conducted using Stata version 10.

Table 1.

Binary Logistic Regression Models Examining Changes in Intimate Partner Violence (IPV) Victimization From Mothers Overcoming Violence through Education and Empowerment (MOVE) Program Entry to Program Completion and Program Follow-Up.

Variable	Psychological Abuse	Physical Abuse	Injury
Variable	Odds Ratio	Odds Ratio	Odds Ratio
Number of months since baseline	0.965	1.168	1.034
Postintervention^a (preintervention)	0.162**	0.035***	0.141**
Age at baseline	1.051	0.989	0.994
Education at baseline (HS/GED or below)
College/technical school or some	5.739*	3.738*	3.687*
Graduate or some	3.671	0.581	0.613
Baseline income (personal employment)
Government assistance	1.212	0.616	0.926
Other	1.460	1.941	4.506*
Number of children live with at baseline	0.972	0.844	0.994
Number of children aged 5 or under at baseline	1.460	1.571	1.552
Race/ethnicity (Non-African American)
African American	1.543	1.640	1.144
Baseline employment (full time)
Part time	1.471	0.918	0.512
Unemployed/homemaker	1.502	1.343	0.773
Referral sources (both CPS and court)
Child protective services	1.707	0.992	1.159
Court	1.168	2.094	0.897
Relationship with abusive partner—time varying (No)
Yes	1.844	0.543	0.535
Pseudo R ²	.161	.276	.214
Number of observations	167	167	167
Number of participants	68	68	68
Wald χ² (df)	54.08 (15)***	53.46 (15)***	27.42 (15)***

Note. HS = high school; GED = general equivalency diploma.

Reference groups are shown in parentheses for indicator variables.

^a “Postintervention” includes both program completion and program follow-up data collection points.

*p < .05. **p < .01. ***p < .001, one-tailed test for “Postintervention” and two-tailed test for all other covariates.

Table 2.

Binary Logistic Regression Models Examining Changes in Intimate Partner Violence (IPV) Perpetration From Mothers Overcoming Violence through Education and Empowerment (MOVE) Program Entry to Program Completion and Program Follow-Up.

Variable	Psychological Abuse	Physical Abuse	Injury
Variable	Odds Ratio	Odds Ratio	Odds Ratio
Number of months since baseline	1.037	1.166	1.088
Postintervention^a(preintervention)	0.108***	0.064***	0.174*
Age at baseline	1.037	0.969	1.035
Education at baseline (HS/GED or below)
College/technical school or some	3.878	1.636	7.994**
Graduate or some	1.510	0.151*	0.581
Baseline income (personal employment)
Government assistance	0.902	3.534	2.201
Other	1.204	5.651*	6.626*
Number of children live with at baseline	0.951	0.738	0.769
Number of children aged 5 or under at baseline	1.192	1.233	2.963**
Race/ethnicity (Non-African American)
African American	1.854	4.414**	6.674**
Baseline employment (full time)
Part time	2.269	0.625	0.828
Unemployed/homemaker	1.130	0.429	0.370
Referral sources (both CPS and court)
Child protective services	1.362	0.450	2.126
Court	1.043	1.617	3.320*
Relationship with abusive partner—time varying (No)
Yes	2.653	0.759	1.583
Pseudo R ²	.171	.266	.282
Number of observations	167	167	167
Number of participants	68	68	68
Wald χ² (df)	54.45 (15)***	52.27 (15)***	50.22 (15)***

Note. HS = high school; GED = general equivalency diploma.

Reference groups are shown in parentheses for indicator variables.

^a “Postintervention” includes both program completion and program follow-up data collection points.

*p < .05. **p < .01. ***p < .001, one-tailed test for “Postintervention” and two-tailed test for all other covariates.

Results

Sample Descriptives on Study Covariates

Table 3 shows the sample descriptives on all covariates. Almost half the sample (n = 33, 47.1%) was solely court mandated, whereas 20% (n = 14) of the sample was solely mandated by CPS. About a third of the sample (n = 23, 32.9%) were involved with both the court and CPS systems. The sampled participants ranged in age from 20 to 51, with a mean age of 31. Most women identified their race/ethnicity as African American/Black (n = 37, 52.9%), followed by White (n = 22, 31.4%), multiracial (n = 7, 10%), and other (n = 4, 5.7%). Participants who indicated their race/ethnicity as other included women who self-identified as Asian, American Indian or Alaska Native, and Latina. The level of education was high among participants with nearly 70% having completed some post-high school coursework or having obtained diplomas at the undergraduate or graduate levels.

Table 3.

Participants’ Descriptive Characteristics: Covariates.

Covariate	Preintervention		Postintervention		3-Month Follow-Up
Covariate	N	% or M (SD)	N	%	N
Age	70	30.94 (7.53)
Education (HS/GED or below)	70	30.0%
College/technical school or some	70	55.7%
Graduate or some	70	14.3%
Income (personal employment)	68	44.1%
Government assistance	68	11.8%
Other	68	44.1%
Number of children live with participant	70	1.60 (1.12)
Number of children aged 5 or under	70	1.04 (0.73)
Race/ethnicity (Non-African American)	70	47.1%
African American	70	52.9%
Employment (full time)	70	41.4%
Part time	70	18.6%
Unemployed/homemaker	70	40.0%
Referral sources (CPS and Court)	70	32.9%
Child protective services	70	20.0%
Court	70	47.1%
Relationship with abusive partner
No	69	58.0%	69	68.1%	59	81.4%
Yes	69	42.0%	69	31.9%	59	18.6%

Note. HS = high school; GED = general equivalency diploma.

Over a third of participants (n = 29, 41.4%) were working full time, nearly 20% were working part time (n = 13, 18.6%), and 40% were unemployed or full-time homemakers (n = 28). About 44% (n = 30) received income through their own employment, nearly 12% (n = 8) received income through government assistance (i.e., Work First check, Unemployment, Social Security/Supplemental Security Insurance), and approximately 44% (n = 30) identified other sources of income (e.g., no income, other’s employment, multiple sources of income).

With the exception of two participants, all the women were mothers (biological, adoptive, foster, or step); the two exceptions were women who were primary caregivers for their male partners’ children at the time they entered the program The mean number of children living with the participant was 1.60 (SD = 1.12) with a range from 0 to 5 children. The mean number of children 5 years old or younger was 1.04 (SD = 0.73) with a range from 0 to 3 children. At program entry, 42% of the participants were in a relationship with the abusive partner; this percentage decreased to about 32% at program completion and 19% at the 3-month postcompletion follow-up.

Further, we note that on average, participants attended 11 MOVE sessions of the 13 offered. Additional information on the MOVE mothers’ baseline characteristics, needs, and resources (e.g., mental health, parenting practices, help-seeking efforts) is available in another publication (Macy et al., in press).

Sample Descriptives on Study Outcome Variables

Table 4 presents the sample descriptives of the eight IPV victimization and perpetration outcomes at each of the three time points in conjunction with the hypothesized direction of change. This table shows that the mean value of each IPV outcome variable decreased from program entry to program completion. We used the logistic regression analysis to examine whether statistically significant effects existed after controlling for various covariates. Table 4 also shows that a control of clustering effects was warranted because temporal clustering effects existed across all outcomes.

Table 4.

Intimate Partner Violence Victimization and Perpetration Descriptive Statistics.

CTS Outcome		Preintervention			Postintervention			3-Month Follow-Up
(Hypothesized Sign of Effect)	ICC	N	M	SD	N	M	SD	N	M	SD
Victimization
Psychological abuse (−)	0.4411	69	0.96	0.21	55	0.78	0.42	48	0.75	0.44
Physical abuse (−)	0.1801	69	0.90	0.30	55	0.44	0.50	48	0.52	0.51
Sexual abuse (−)	0.4474	68	0.38	0.49	55	0.20	0.40	48	0.21	0.41
Injury (−)	0.3741	69	0.65	0.48	55	0.29	0.46	48	0.31	0.47
Perpetration
Psychological abuse (−)	0.4535	69	0.96	0.21	55	0.75	0.44	48	0.73	0.45
Physical abuse (−)	0.3674	69	0.81	0.39	55	0.40	0.49	48	0.46	0.50
Sexual abuse (−)	0.4361	69	0.09	0.28	55	0.07	0.26	48	0.02	0.14
Injury (−)	0.4222	69	0.44	0.50	55	0.22	0.42	48	0.21	0.41

Note. CTS = Revised Conflict Tactics scales; ICC = intraclass correlation coefficient; SD = standard deviation.

Logistic Regression Analysis of Partner Violence Victimization

A Wald χ² test was conducted to test the model fit of the logistic regression models for each of the victimization outcomes. Table 1 shows the results for the logistic regression models of the IPV victimization outcomes found to be statistically significant: psychological abuse, physical abuse, and injury. The logistic regression model for sexual abuse victimization was not statistically significant.

Psychological Abuse

The logistic regression model for psychological abuse victimization was statistically significant, χ²(15) = 54.08, p < .001. This result indicated an excellent model fit to the data and that we could reject the null hypothesis that all coefficients were equal to 0. Using a one-tailed test, the postintervention variable was also statistically significant. The odds ratio for postintervention was 0.162, meaning that other things being equal, the probability of having experienced any form of psychological abuse victimization at any postintervention time point was 83.8% lower relative to program entry (p < .01). Using a two-tailed test, the only significant covariate for this model was education; specifically, the indicator variable graduated or attended college/technical school. The odds ratio for this variable was 5.739, meaning that other things being equal, participants who reported having completed some coursework or a degree at a college or technical school were 473.9% more likely than participants with a high school/GED or less education to have experienced any form of psychological abuse (p < .05).

Physical Abuse

The logistic regression model for physical abuse victimization was statistically significant, χ²(15) = 53.46, p < .001, indicating an excellent model fit to the data. Using a one-tailed test, the postintervention variable was also statistically significant. The odds ratio for postintervention is 0.035, meaning that other things being equal, the probability of having experienced any form of physical abuse victimization at any postintervention time point was 96.5% lower than at program entry (p < .001). Using a two-tailed test, the only significant covariate for this model was education, specifically, the indicator variable graduated or attended college/technical school. The odds ratio for this variable was 3.738, meaning that other things being equal, participants who reported having completed some coursework or a degree at a college or technical school were 273.8% more likely than participants with a high school/GED or less education to have experienced any form of physical abuse (p < .05).

Injury

The logistic regression model for injury resulting from IPV victimization was statistically significant, χ²(15) = 27.42, p < .001, indicating excellent model fit. Using a one-tailed test, the postintervention variable was also statistically significant. The odds ratio for postintervention was 0.141. The probability of having experienced any injury as a result of IPV victimization at any postintervention time point was 85.9% lower than at program entry (p < .01), other things being equal. Using a two-tailed test, two covariates emerged as statistically significant, education (i.e., indicator variable graduated or attended college/technical school) and income at program entry (i.e., indicator variable other). The odds ratio for graduated or attended college/technical school was 3.687. Thus, with all else held constant, participants who reported having completed some coursework or a degree at a college or technical school were 268.7% more likely than participants with a high school/GED or less education to have experienced injury as a result of their IPV victimization (p < .05). The odds ratio for other sources of income at program entry was 4.506. Other things being equal, participants who reported sources of income other than personal employment or government assistance were 350.6% more likely to have experienced IPV-related injury (p < .05) than participants who indicated income from personal employment.

Plotted Change Trajectories for Significant Victimization Outcomes

We plotted the change trajectories of the three significant IPV victimization outcomes (see Figure 1). In each plot, we present both the observed mean change trajectory and the model-predicted mean change trajectory. The observed trajectory is the outcome mean at each of the three time points of data collection (i.e., at program entry or Month 0; at program completion or Month 3; and at the 3-month follow-up, or Month 6). The model-predicted trajectory is based on the estimated logistic regression model presented in Table 1 that controls for all covariates at the sample mean level while allowing the time and postintervention variables to vary according to the designed scenarios. Given that the observed trajectory is the mean outcome at each time point, the model does not control for covariates. Hence, a departure of the observed trajectory from the model-predicted trajectory is attributable to multivariate analysis that controls for covariates. On all presented outcomes, the victimization outcome at a postintervention point (i.e., either at Month 3 or at Month 6) was significantly lower than that at program entry. The change trajectory for psychological victimization also demonstrated a continued beneficial trend from program completion (i.e., Month 3) to the 3-month follow-up (i.e., Month 6). However, on physical abuse victimization and victimization-related injury, results showed an increase from program completion (i.e., Month 3) to 3-month follow-up (i.e., Month 6). For these variables, the model-predicted mean at program follow-up was higher than that at program completion.

Figure 1.

Victimization outcomes: observed and model-predicted mean change trajectories (p < .05). Note. The model-based predictions employed the estimated effects presented in Table 1, controlled for all covariates at the sample mean level, and allowed the time and postintervention variables to vary based on the designed scenarios.

Logistic Regression Analysis of Partner Violence Perpetration

A Wald χ² test was conducted to test the model fit of the logistic regression models for each of the perpetration outcomes. Table 2 shows the statistically significant results for the logistic regression models of the IPV perpetration outcomes: psychological abuse, physical abuse, and injury. The logistic regression model for sexual abuse encountered an empty cell problem due to a small proportion of observations endorsing sexual abuse perpetration.

Psychological Abuse

The logistic regression model for psychological abuse perpetration was statistically significant, χ²(15) = 54.45, p < .001, indicating excellent model fit. Using a one-tailed test, the postintervention variable was also statistically significant. The odds ratio for postintervention was 0.108, meaning that other things being equal, the probability of having perpetrated psychological abuse at any postintervention time point was 89.2% lower than at program entry (p < .001). Using a two-tailed test, no covariate emerged as statistically significant.

Physical Abuse

The logistic regression model for physical abuse perpetration was statistically significant, χ²(15) = 52.27, p < .001, indicating excellent model fit. Using a one-tailed test, the postintervention variable was also statistically significant. The odds ratio for postintervention was 0.064, meaning that other things being equal, the probability of having perpetrated physical abuse at any postintervention time point was 93.6% lower than at program entry (p < .001). Using a two-tailed test, several covariates emerged as significant: education (i.e., indicator variable completed or attended graduate degree program), income at program entry (i.e., indicator variable other), and race. The odds ratio for completed or attended graduate degree program was 0.151, meaning that, other things being equal, participants who reported having completed a graduate degree or some graduate coursework were 84.9% less likely than participants with a high school/GED or less education to have perpetrated any form of physical abuse (p < .05). The odds ratio for other sources of income at program entry was 5.651. Other things being equal, participants who reported sources of income other than personal employment or government assistance were 465.1% more likely to have perpetrated physical abuse (p < .05) than participants who indicated personal employment as their source of income. The odds ratio for race was 4.414, meaning that other things being equal, African American participants were 341.4% more likely than non-African American participants to have reported perpetrating physical abuse (p < .01).

Injury

The logistic regression model for IPV perpetration–related injury was statistically significant, χ²(15) = 50.22, p < .001, indicating excellent model fit to the data. Using a one-tailed test, the postintervention variable was statistically significant. The odds ratio for postintervention was 0.174. The probability of having caused injury as a result of IPV perpetration at any postintervention time point was 82.6% lower than at program entry (p < .05), other things being equal. Using a two-tailed test, five covariates emerged as statistically significant: education (i.e., indicator variable graduated or attended college/technical school), income (i.e., indicator variable other), number of children 5 years or younger living with the participant, race, and referral sources (i.e., indicator variable court). The odds ratio for graduated or attended college/technical school was 7.994. This result suggests that all else held constant, participants who reported having completed some coursework or a degree at a college or technical school were 699.4% more likely than participants with a high school/GED or less education to have caused injury as a result of their IPV perpetration (p < .01). The odds ratio for other sources of income was 6.626. Other things being equal, participants who reported sources of income other than personal employment or government assistance were 562.6% more likely than participants who indicated personal employment as their source of income to have caused IPV perpetration-related injury (p < .05). The odds ratio for number of children 5 years or younger living with the participant was 2.963. Other things being equal, every unit increase in the number of children 5 years or younger living with the participant increased the chance of causing perpetration-related injury by 196.3% (p < .01). The odds ratio for race was 6.674, meaning that other things being equal, African American participants were 567.4% more likely than non-African American participants to report causing IPV perpetration–related injuries (p < .01). The odds ratio for the court referral variable was 3.320. Holding all other variables constant, participants solely court referred to MOVE program were 232% more likely to have caused perpetration-related injury (p < .05) than participants referred by both the CPS and court systems.

Plotted Change Trajectories for Significant Victimization Outcomes

We also plotted the change trajectories of the three significant IPV perpetration outcomes in the same manner as the significant victimization outcomes discussed above (see Figure 2). The graphs in Figure 2 demonstrate that for all presented outcomes, the perpetration outcome at a postintervention point (i.e., Month 3 or Month6) were significantly lower than perpetration at program entry. However, the graphs for each of the perpetration outcomes also showed an increase from Month 3 to Month 6. On these variables, the model-predicted mean at program completion is higher than that at program follow-up.

Figure 2.

Perpetration outcomes: observed and model-predicted mean change trajectories (p < .05). Note. The model-based predictions employed the estimated effects presented in Table 2, controlled for all covariates at the sample mean level, and allowed the time and postintervention variables to vary based on the designed scenarios.

Discussion and Applications to Social Work

The overall study aim was to investigate changes in participants’ experiences with IPV perpetration and victimization over the course of their participation in a community-based program that focused on IPV, parenting, and safety. The study was guided by two research questions. First, were participants less likely to report IPV victimization and perpetration at program completion and program follow-up relative to program entry? Second, to what extent did participant characteristics relate to participants’ experiences of IPV victimization and perpetration? By conducting this research, we endeavored to help address the acute insufficiency of intervention research conducted with IPV victims involved with CPS and/or the court systems. In addition, we were concerned with the feasibility of conducting intervention research with this vulnerable group of women in a community-based setting and in collaboration with community-based human service organizations. Taken together, the statistically significant reductions in the number of participants reporting IPV victimization and perpetration from program entry to program completion on physical abuse, psychological abuse, and IPV-related injury suggests that MOVE may hold promise for helping IPV victims involved with the court and/or CPS with safety to address family violence. This study also provides information about IPV victims who are involved with the court and/or CPS that can be used to develop programs and policies for this vulnerable group of women and their children.

This group of IPV victims varied in relationship status with the abusive partner, socioeconomic status, and number of children. Overall, this was a highly educated sample with nearly 70% of the women having attended at least college or technical school.

More than half of the sample described their race/ethnicity as African American or Black. Recent national research shows high levels of IPV among African Americans (Black et al., 2011), and this sample may reflect this national trend. However, this demographic finding could be interpreted as suggesting that communities of color are subject to disproportionate levels of surveillance by CPS and the justice system. Consistent with such an interpretation and in contrast to the demographic information from this sample, the most recent census describes Wake County, North Carolina, where MOVE is offered and this research was conducted, as 70% White and 21.3% Black.

In interpreting these demographic findings, we also encourage readers to be mindful that prior research has suggested that women of color tend to find typical domestic violence services less welcoming and helpful than their majority counterparts (Bent-Goodley, 2005). To the extent that African American IPV victims are not able to secure help through usual services, the violence in their families may escalate to the point at which they are likely to come to the attention of CPS and the criminal justice system.

Though most of the participants were involved with either the CPS system or the court system, over a third of the participants were involved with both systems. Prior research has typically investigated female IPV victims in relation to single-system involvement only. To the best of our knowledge, this research is among one of the first to show that some parenting, female IPV victims may be involved in the court system, CPS, and community-based services.

Further, the descriptive statistics show a notable trend of participants leaving their intimate partners over the course of the program and at follow-up. At program entry, 42% of the participants were with their abusive partners, whereas at the 3-month follow-up, this number had fallen below 20%.

For IPV victimization, three models were statistically significant, including psychological abuse, physical abuse, and injury due to IPV victimization. Based on these models, fewer research participants reported these three IPV victimization types from program entry to program completion. The findings also show that fewer participants reported these three IPV victimization types at program follow-up relative to program entry.

Women who reported attending or completing college or technical school were more likely to report psychological abuse, physical abuse, and injury due to IPV victimization relative to women who had a high school degree or less education. These findings are noteworthy because higher levels of education are typically thought to be protective against IPV victimization (Jewkes, 2002). However, given the small sample size, we encourage future research to investigate the connections between women’s IPV victimization and education.

In addition, participants who reported sources of income other than government assistance or their own employment were more likely than employed participants to report injury due to IPV victimization. A review of our data showed that women with these other sources of income were typically relying on others for economic support, especially boyfriends and husbands. We posit that this subgroup of women may have been especially vulnerable to IPV-inflicted injury because of their limited economic power and their reliance on their male partners (Jewkes, 2002). Thus, we speculate that this subgroup of CPS and court-involved IPV victims may need economic, educational, and job-focused interventions in addition to the MOVE program to maintain the positive IPV trends found in the sample overall.

For IPV perpetration, three models were statistically significant, including psychological abuse, physical abuse, and perpetrating IPV that resulted in a partner’s injury. Based on these models, significantly fewer research participants reported perpetrating these types of IPV against their partners from program entry to program completion. The findings also showed that fewer participants reported perpetrating these types of IPV types at follow-up relative to program entry.

Regarding the covariates and IPV perpetration, participants who had completed a graduate degree or attended some graduate school were less likely to report perpetrating physical abuse relative to women who had attended or completed high school. In addition, women referred to the program exclusively by the court were more likely to report perpetrating IPV that resulted in a partner’s injury relative to women referred from both the court and CPS. In addition, women who had attended or completed college (but had not attended graduate school) were more likely to report perpetrating IPV that resulted in a partner’s injury relative to women who had attended or completed high school. Further, women who reported sources of income other than government assistance or their own employment were more likely to report perpetrating physical IPV, as well as perpetrating IPV that resulted in a partner’s injury relative to women whose primary income source was their personal employment.

We consider these results in light of the victimization findings discussed above. Unfortunately, the data from this study do not allow for an investigation of how participants’ IPV experiences of perpetration and victimization were sequenced or related. However, the elevated levels of both IPV victimization and perpetration for the women in these particular educational and income subgroups suggests severely violent relationships in which women may have been trying to protect themselves from abusive partners. Such interpretations are consistent with Henning, Renauer, and Holdford’s (2006) research on court-involved IPV victims that showed few women in their study could be considered the primary aggressors in their relationships. Further, prior research that examined service-mandated female IPV victims has recommended that programs for these women should address IPV perpetration in relation to women’s victimization and self-defense (Hamberger & Potente, 1994; Stuart et al., 2006). Our findings support such recommendations.

In a related vein relative to women of other race and ethnic groups, African American participants were more likely to report perpetration of IPV and physical IPV that resulted in a partner’s injury. Moreover, women with children 5 years or younger living at home were also more likely to report perpetrating IPV that resulted in a partner’s injury. We speculate that these subgroups of participants may have been more likely to perpetrate these types of IPV as a function of their vulnerable statuses as women of color and as the mothers of young children (Stuart et al., 2006; Swan & Snow, 2006). Again, we encourage future research to investigate the connections between women’s IPV perpetration and factors such as race, as well as the ages of women’s children in light of the small study sample.

We note the slight increase in the numbers of participants reporting physical IPV perpetration, as well as physical, sexual, and injury IPV victimization, from program completion to program follow-up (see Table 4, Figures 1 and 2). Whether these increases in IPV perpetration and victimization are indicators that program effects might be short term is unclear. Likewise, whether these increasing trends continue after the 3-month follow-up remains unknown based on the available data. The relatively small increases indicate only a few participants went from not reporting these types of IPV at program completion to reporting this IPV at follow-up. Further, our current study did not include information about why such increases might have occurred. For example, research on community domestic violence safety resources has shown that for some female IPV victims, their decision to seek resources might lead to retaliatory IPV from male partners (Dugan, Nagin, & Richard, 2003). Accordingly, we recommend that future research on specific programs similar to MOVE or research on safety interventions for IPV victims in general, collect postintervention data over longer periods. We also recommend that providers of safety interventions for IPV victims consider offering follow-up services to help women maintain any safety gains they have established during their program participation.

Although the findings from this research suggest that MOVE may be a promising program for CPS and court-involved IPV victims in terms of reducing reports of IPV perpetration and victimization, we encourage readers to be mindful of the study’s limits. Foremost, we employed a widely used and accepted measure of IPV in this study, specifically the CTS2. However, this and other measures of IPV have been critiqued for self-report bias and recall bias. Moreover, the phenomenon of IPV is complex, dynamic, and multifaceted. Accordingly, it is unclear whether any existing IPV measures do an adequate job of capturing such experiences accurately and fully. Thus, we urge future research on IPV measurement that attends to both perpetration and victimization.

In addition, the IPV subscales (i.e., CTS2 subscales for Psychological Abuse, Physical Abuse, Sexual Abuse, Injury) used as this study’s outcomes were dichotomized to account for unequal periods of data collection. At program entry, participants were asked to reflect on their IPV experiences over the past year, whereas at program completion and the 3-month follow-up, participants were asked to reflect on their IPV experiences in the preceding 3 months. By dichotomizing these IPV subscale variables, our team conducted a conservative, rigorous test of the MOVE program. However, we also lost information about IPV severity and types, as well as statistical power. In a related vein, we modified the sexual abuse scale used in this study. Our research team made that decision based on feedback from MOVE program graduates. Although this decision was consistent with the collaborative approach used to conduct this research, the lack of significance for that IPV outcome might be due to our modifications.

An analysis between the women who did and did not complete MOVE showed that there were significant differences between the two groups on the number of children living with the participants, with the three women who dropped out of the program having slightly more children on average living with them than women who remained in the program. Though only a few women did not complete the program, the finding might suggest that women with more children living with them need great support to ensure program completion.

Although an RCT is deemed the gold standard in intervention research, our team—in collaboration with our community partners—decided to conduct a quasi-experimental study of MOVE to garner preliminary evidence regarding whether the program showed enough promise to warrant further evaluation using a more rigorous research design (Fraser & Galinsky, 2010). Consequently, findings from this pilot research provide important insights regarding the program’s potential effects that, in turn, can be used to inform a more rigorous study.

We also sought to determine whether a group-level intervention effect existed, given the program’s intervention approach with participants nested in groups. We had posited that there would be a group intervention effect, given some differences among MOVE service providers as well as differences among group members across each cohort. Notably, our findings showed no group-level intervention effect. This information will be valuable for conducting power calculations for future studies of MOVE.

Despite the limitations, this study has notable strengths, including (a) robust participant recruitment and retention; (b) postintervention follow-up data collection; (c) the use of dynamic time-varying statistical analyses to account for relevant and important covariates; and (d) the collaborative nature of this research with our community partners. Another important study strength was the way in which the research was conducted in a community-based setting with mainstream service providers delivering the program.

Accordingly, the pilot study findings from this novel IPV program begin to answer the question of how to help IPV victims who are involved with CPS and the court system. As one of the first evaluations of an intervention for this vulnerable group of IPV victims, this research also contributes to the overall evidence base of interventions for IPV victims and their families. Further, this study provides evidence for the feasibility of and successful strategies for recruiting and retaining this vulnerable group of IPV victims in research studies. In light of these positive findings and in collaboration with our community partners, our research team aims to conduct a rigorous, multisite, RCT with a large participant sample. In the meantime, the findings from this study suggest that the MOVE program exerts a positive influence on IPV victimization and perpetration among women mandated to community services by CPS, the court systems, or both.

Footnotes

Acknowledgments

We acknowledge the staff of InterAct and SAFEchild for their collaboration in and help with this research. In particular, we acknowledge all of the MOVE Collaboration Team members: Delana Epps-Robinson, David Farrell, Kathy Johnson, Jo Lawson, Marjorie Menestres, Phillip H. Redmond Jr., and Stacey Sullivan. We acknowledge the North Carolina Governor’s Crime Commission for their support of the MOVE Program. We acknowledge Natalie B. Johns and Melissa Goodman for their help with this research. We also acknowledge Diane Wyant for her comments on an earlier draft of this article.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Funding for this research came from The Duke Endowment.

References

Bandura

(1991). Social cognitive theory of self-regulation. Organizational Behavioral and Human Decision Processes, 50, 248–278. doi:10.1016/0749-5978(91)90022-L

Bent-Goodley

T. B.

(2005). Culture and domestic violence: Transforming knowledge development. Journal of Interpersonal Violence, 20, 195–203. doi:10.1177/0886260504269050

Black

M. C.

Basile

K. C.

Breiding

M. J.

Smith

S. G.

Walters

M. L.

Merrick

M. T.

… Stevens

M. R.

(2011). The national intimate partner and sexual violence survey (NISVS): 2010 summary report. Atlanta, GA: National Center for Injury Prevention and Control, Centers for Disease Control and Prevention. Retrieved from http://www.cdc.gov/ViolencePrevention/pdf/NISVS_Report2010-a.pdf

Briere

Jordan

(2004). Violence against women: Outcome complexity and implications for assessment and treatment. Journal of Interpersonal Violence, 19, 1252–1276. doi:10.1177/0886260504269682

Busch

A. L.

Rosenberg

M. S.

(2004). Comparing women and men arrested for domestic violence: A preliminary report. Journal of Family Violence, 19, 49–57. Retrieved from http://www.springerlink.com/content/104903/

Busch

N. B.

Valentine

(2000). Empowerment practice: A focus on battered women. Affilia, 15, 82–95. doi:10.1177/08861090022093840

Campbell

(2002). Health consequences of intimate partner violence. Lancet, 359, 1331. doi:10.1016/S0140-6736(02)08336-8

Casanueva

Martin

S. L.

Runyan

D. K.

(2009). Repeated reports for child maltreatment among intimate partner violence victims: Findings from the National Survey of Child and Adolescent Well-Being. Child Abuse & Neglect, 33, 84–93. doi:10.1016/j.chiabu.2007.04.017

Dugan

Rosenfeld

Nagin

D. S.

(2003). Exposure reduction or retaliation? The effects of domestic violence resources on intimate-partner homicide. Law & Society Review, 37, 169–198. doi:10.1111/1540-5893.3701005

10.

Edleson

(1999). The overlap between child maltreatment and woman battering. Violence Against Women, 5, 134–154. doi:10.1177/107780129952003

11.

Ermentrout

D. M.

Rizo

C. F.

Macy

R. J.

(In press). ‘This Is About Me’: Feasibility findings from the children’s component of an IPV intervention for justice-involved families. Violence Against Women.

12.

Evans

S. E.

Davies

DiLillo

(2008). Exposure to domestic violence: A meta-analysis of child and adolescent outcomes. Aggression and Violent Behavior, 13, 131–140. doi:10.1016/j.avb.2008.02.005

13.

Fraser

M. W.

Galinsky

M. J.

(2010). Steps in intervention research: Designing and developing social programs. Research on Social Work Practice, 20, 459–466. doi:10.1177/1049731509358424

14.

Guo

(2005). Analyzing grouped data with hierarchical linear modeling. Children and Youth Services Review, 27, 637–652. doi:10.1016/j.childyouth.2004.11.017

15.

Hamberger, L. K. & Potente, T. (1994). Counseling heterosexual women arrested for domestic violence: Implications for theory and practice. Violence and Victims, 9(2), 125–137.

16.

Hamby

Finkelhor

Turner

Ormrod

(2010). The overlap of witnessing partner violence with child maltreatment and other victimizations in a nationally representative survey of youth. Child Abuse & Neglect, 34, 734–741. doi:10.1016/j.chiabu.2010.03.001

17.

Hazen

A. L.

Connelly

D.D.

Kelleher

Landsverk

Barth

(2004). Intimate partner violence among female caregivers of children reported for child maltreatment. Child Abuse & Neglect, 28, 301–319. doi:10.1016/j.chiabu.2003.09.016

18.

Henning

Renauer

Holdford

(2006). Victim or offender? Heterogeneity among women arrested for intimate partner violence. Journal of Family Violence, 21, 351–368. doi:10.1007/s10896-006-9032-4

19.

Holden

G. W.

(1998). Introduction: The development of research into another consequence of family violence. In Holden

E. W.

Geffner

Jouriles

E. N.

(Eds.), Children exposed to marital violence: Theory, research, and applied issues (pp. 1–20). Washington, DC: American Psychological Association.

20.

Huber

P. J.

Ronchetti

E. M.

(2009). Robust statistics (2nd ed.). Hoboken, NJ: John Wiley.

21.

Ireland

T. O.

Smith

C. A.

(2009). Living in partner-violent families: Developmental links to antisocial behavior and relationship violence. Journal of Youth and Adolescence, 38, 323–339. doi:10.1007/s10964-008-9347-y

22.

Jewkes

(2002). Intimate partner violence: Causes and prevention. Lancet, 359, 1423–1429. doi:10.1016/S0140-6736(02)08357-5

23.

Kitzmann

K. M.

Gaylord

N. K.

Holt

A. R.

Kenny

E. D.

(2003). Child witnesses to domestic violence: A meta-analytic review. Journal of Consulting and Clinical Psychology, 71, 339–352. doi:10.1037/0022-006X.71.2.339

24.

Kohl

P. L.

Edleson

J. L.

English

D. J.

Barth

R. P.

(2005). Domestic violence and pathways into child welfare services: Findings from the National Survey of Child and Adolescent Well-Being. Children and Youth Services Review, 27, 1167–1182. doi:10.1016/j.childyouth.2005.04.003

25.

Macy

R. J.

Ermentrout

D. M.

Rizo

C. F.

(2012). An innovative program for justice-involved partner violence survivors: “No man is worth me getting locked up.” Journal of Family Violence, 27, 453–464. doi:10.1007/s10896-012-9436-2

26.

Macy

R. J.

Giattina

Sangster

T. H.

Crosby

Montijo

N. J.

(2009). Domestic violence and sexual assault services: Inside the blackbox. Aggression and Violent Behavior, 14, 359–373. doi:10.1016/j.avb.2009.06.002

27.

Macy

R. J.

Rizo

C. F.

Ermentrout

D. M.

(In press). Characteristics, needs, and help-seeking of partner violence victims mandated to community services by courts and child protective services. Journal of Orthopsychiatry.

28.

Miller

S. L.

(2001). The paradox of women arrested for domestic violence: Criminal justice professionals and service providers respond. Violence Against Women, 7, 1339–1376. doi:10.1177/10778010122183900

29.

National Network to End Domestic Violence. (2012). Domestic violence counts 2011: A 24-hour census of domestic violence shelters and services across the United States. Washington, DC. Retrieved from http://nnedv.org/resources/census/2011-report.html

30.

Pachankis

J. E.

Goldfried

M. R.

(2007). On the next generation of process research. Clinical Psychology Review, 27, 760–768. doi:10.1016/j.cpr.2007.01.009

31.

Rajan

McCloskey

K. A.

(2007). Victims of intimate partner violence: Arrest rates across recent studies. Journal of Aggression, Maltreatment & Trauma, 15, 27–52. Retrieved from http://www.informaworld.com/smpp/title∼content=t792303964

32.

Riger

Bennett

Wasco

S. M.

Schewe

P. A.

Frohmann

Camacho

J. M.

Campbell

(2002). Evaluating services for survivors of domestic violence and sexual assault. Thousand Oaks, CA: Sage.

33.

Rizo

C. F.

Macy

R. J.

Ermentrout

D. E.

Johns

N. B.

(2011). A review of family interventions for intimate partner violence with a child focus or child component. Aggression and Violent Behavior, 16, 144–166. doi:10.1016/j.avb.2011.02.004

34.

Roustit

Renahy

Guernec

Lesieur

Parizot

Chauvin

(2009). Exposure to interparental violence and psychosocial maladjustment in the adult life course: Advocacy for early prevention. Journal of Epidemiology and Community Health, 63, 563–568. doi:10.1136/jech.2008.077750

35.

Russell

Springer

K. W.

Greenfield

E. A.

(2010). Witnessing domestic abuse in childhood as an independent risk factor for depressive symptoms in young adulthood. Child Abuse & Neglect, 34, 448–453. doi:10.1016/j.chiabu.2009.10.004

36.

Schechter

Edleson

J. L.

(1999). Effective intervention in domestic violence and child maltreatment cases: Guidelines for policy and practice. Reno, NV: National Council of Juvenile and Family Court Judges. Retrieved from http://www.vaw.umn.edu/documents/executvi/executvi.html

37.

Simmons

C. A.

Lehmann

Collier-Tenison

(2008). From victim to offender: The effects of male initiated violence on women arrested for using intimate partner violence. Journal of Family Violence, 23, 463–472. doi:10.1007/s10896-008-9173-8

38.

Simmons

C. A.

Lehmann

Dia

D. A.

(2010). Parenting and women arrested for intimate partner violence. Journal of Interpersonal Violence, 25, 1429–1448. doi:10.1177/0886260509346064

39.

Snijders

T. A. B.

Bosker

R. J.

(1999). Multilevel analysis: An introduction to basic and advanced multilevel modeling. Thousand Oaks, CA: Sage.

40.

Stata. (2007). Stata user's guide, Release 10, [U]. College Station, TX: Author.

41.

Stover

C. S.

Meadows

A. L.

Kaufman

(2009). Interventions for intimate partner violence: Review and implications for evidence-based practice. Professional Psychology: Research and Practice, 40, 223–233. doi:10.1037/a0012718

42.

Straus

M. A.

Hamby

S. L.

Boney-McCoy

Sugarman

D. B.

(1996). The revised Conflict Tactics Scales (CTS2): Development and preliminary psychometric data. Journal of Family Issues, 17, 283–316. doi:10.1177/019251396017003001

43.

Stuart

G. L.

Moore

T. M.

Hellmuth

J. C.

Ramsey

S. E.

Kahler

C. W.

(2006). Reasons for intimate partner violence perpetration among arrested women. Violence Against Women, 12, 609–621. doi:10.1177/1077801206290173

44.

Sullivan

C. M.

(2011). Evaluating domestic violence support service programs: Waste of time, necessary evil, or opportunity for growth? Aggression and Violent Behavior, 16, 354–360. doi:10.1016/j.avb.2011.04.008

45.

Swan

S. C.

Snow

D. L.

(2006). The development of a theory of women’s use of violence in intimate relationships. Violence Against Women, 12, 1026–1045. doi:10.1177/1077801206293330

46.

Turner

Gilbert

Hendricks

Demaree

(Eds.). (2006). Growing beyond conflict: The path to building safer families parent group curriculum. San Diego, CA: Chadwick Center for Children & Families, Children’s Hospital and Health Center.

47.

U.S. Department of Health and Human Services. (2009). Child witness to domestic violence: Summary of state laws. Retrieved from http://www.childwelfare.gov/systemwide/laws_policies/statutes/witnessdvall.pdf

48.

Wathen

C. N.

MacMillan

H. L.

(2003). Interventions for violence against women: Scientific review. Journal of the American Medical Association, 289, 589–600. doi:10.1001/jama.289.5.589

49.

Zolotor

A. J.

Theodore

A. D.

Coyne-Beasley

Runyan

D. K.

(2007). Intimate partner violence and child maltreatment: Overlapping risk. Brief Treatment and Crisis Intervention, 7, 305–321. doi:10.1093/brief-treatment/mhm021