This article traces the experience of four of the five Department of Health and Human Services (DHHS) recipients who received funding to provide domestic training to child welfare agencies in four areas of the United States. The article begins with the developing research which documents the connection between child abuse and domestic violence and explores the fertile ground for tensions between battered women and their advocates and CPS workers. The article also presents findings from the experiences of the DHHS funded programs, their accomplishments, and the obstacles they faced in integrating domestic violence into child welfare practice with the ultimate goal of protecting the mother-child unit. Finally, it concludes with policy and practice recommendations for researchers and practitioners who are working at the intersection of these abuses.

Domestic violence describes a pattern of battering or abusive acts in the context of an intimate relationship. It takes many forms, including destructive acts, and hurtful words or actions. Physical, emotional, and sexual abuse are all forms of domestic violence. In 95% of the cases men abuse women, although in some cases, women are the primary aggressors (U.S. Department of Justice,1994). Indeed, a recent study suggests that in the small percentage of cases involving women who are violent, the abuse is less significant than when perpetrated by men (Ross, 1996). It is now well documented that domestic violence probably occurs at the same rate in heterosexual and homosexual relationships (Lie, Schiltt, Bush, Montague, & Reyes, 1991; Lockhart, White, Causby & Isaac, 1994; Loulan,1987) and is evident in all cultural, ethnic and racial populations, including immigrants (Brownell & Congress, 1998) and people with disabilities (Carlson, 1997; Murphy & Razza, 1998). Some cross-cultural studies reveal cultures where domestic violence is not evident, raising the possibility that the incidence and prevalence of intimate abuse is contingent on a culture’s tolerance for all forms of violence (Levinson, 1983).

Probably the greatest single risk factor for domestic violence is gender (Nurius and Rifino, 1996). In the United States it has been estimated that between 2 million and 4 million women experience severe violence by an intimate partner each year (Straus & Gelles, 1990). Nearly one in three adult women experience at least one physical assault by a partner in adulthood (American Psychological Association, 1996). In a national sample, 28% of married couples report at least one episode of physical violence over the course of their relationship (Straus & Gelles, 1986). In 1992, the U.S. Surgeon General ranked domestic violence as the most common cause of physical injury to women between the ages of 15 to 44 (Novello, Rosenberg, Saltzman, & Shosky, 1994). Violence against women by their intimate partners is a leading cause of injury and death to women.

Batterers often abuse their children as well as their adult partners. According to the American Humane Association (1994), a review of several studies revealed that 45-70% of survivors in shelters reported that their abusers also committed some form of child abuse. The American Humane Association (1994) concluded from these studies that child abuse is at least 15 times more likely to occur in households where domestic violence is present than in those without adult violence. The United States Advisory Board on Child Abuse and Neglect (1995) found that domestic violence is the single major precursor to child deaths in the United States. Messinger and Eldridge (1993) estimated that in 70% of the cases in which an abused child dies, there has been a pattern of abuse against the mother. Child abuse and death are therefore strongly linked to the presence of domestic violence.

Domestic violence and child abuse are linked in that one all too often begets the other. Children may be hurt accidentally, if they are in the "line of fire," and experience violence meant to harm the adult partner. In addition, children may be hurt when intervening to protect a battered parent. One study of shelter mothers and their children found that 44 percent of the children surveyed attempted to protect their mother and 37 percent of those who attempted to protect were "hit" in the process (Hazen, Miller, Laudsverk, 1995). A batterer may also use child abuse to terrorize or spy on his adult partner (Zorza, 1995). The batterer’s intent to hurt his adult partner can lead to child abuse.

Battered mothers may also abuse their children. In 1984, Walker found that mothers are eight times more likely to hurt a child when battered than when safe. More recent statistics suggest that women who have been beaten by their spouses are twice as likely as other women to abuse a child (American Humane Society, 1994). As males become more aggressive toward their spouses, they also become more violent toward their children. The same relationship holds true for women. Of those mothers interviewed for the 1985 Family Violence Survey, 23% of violent wives also engaged in at least one act of physical child abuse (Ross, 1996).

There are many reasons why battered women abuse their children. First, battered women may abuse their children to prevent the child from creating an excuse for the batterer to batter. Second, the victim may be attempting to keep the children "in line," to prevent the batterer from abusing them. Third, the victim may abuse her children as an attempt to release her frustration from being abused (Mills, 1998 b).

When parents separate or the battered mother leaves the abusive relationship, there is an increased danger of an escalation of the violence. In one study, the victimization rate of women separated from their husbands was about three times higher than that of divorced women and 25 times higher than that of married women (Bachman and Saltzman, 1996). According to Hart (1988), battered women who attempt to leave their abuser are most at risk of death. Separated battered women report abuse 14 times as often as those still living with their partner (Harlow, 1991). When parents separate, children are also at risk for abuse, especially during contact visits. Hester and Radford (1992) and Hester and Pearson (1993) reported varying types of stress during contact with a separated parent, including the witnessing of physical or verbal abuse at the meeting point; abduction and use of the child to secure the partner's return to the marriage; and "grilling" the child for information on the mother, thereby exacerbating the child's feelings of divided loyalty.

Because of the harmful effects of domestic violence on children, scholars, policymakers, and CPS agencies, have begun to incorporate it into their child abuse definitions, assessments and interventions (Mills, 1998b). One influence on how domestic violence is being treated in the child abuse community has been the growing evidence of its intergenerational transmission. Children exposed to domestic violence and children who are physically abused are more likely to use violence in their later relationships. Boys who witness family violence are more likely, as adults, to batter female partners (Hotaling and Sugarman, 1986). Straus and Gelles (1986) found that in a comparison of violent men with a control group of non-violent men, the sons of violent parents have a rate of wife beating 900 times greater than that of sons of non-violent parents. Graham-Bermann and Levendosky (1998), compared 21 preschool children of battered women with 25 same age children from nonviolent homes. They found that those children who were exposed to parental violence exhibited such problems as significantly more negative affect, aggression with peers, and ambivalent relationships with caregivers. In a 1986 study of attitudes conducted by Canadian researchers Jaffe, Wilson, and Wolfe, children exposed to violence condoned it to resolve relationship conflicts more readily than did control groups. Their findings also revealed that older children who had experienced extreme violence and other negative life events showed some tendency to hold themselves responsible.

In at least one state, California, witnessing violence between parents or the intimate partners responsible for them, has now been interpreted as a form of child abuse (In re: Heather A., 1997). This policy change is based on empirical evidence that some children who witness domestic violence, but who do not experience the violence directly, show symptoms similar to children who have been physically, sexually, or emotionally abused, including the perpetuation of violence (Echlin & Marshall, 1995; Davidson, 1995). Children exposed to domestic violence are at risk for internalized and externalized behavior problems. Internalized behaviors include withdrawal, anxiety, and somatic complaints. Externalized behaviors include aggressive actions, delinquency, substance abuse and non-compliance with parental and school requests. Fantuzzo and Lindquist (1989) reviewed 29 published papers on children observing violence with a total sample of 1069 children. They found that externalized behaviors such as aggression appeared linked with exposure to violence in almost all studies, but not consistently across all ages and genders. According to Wolfe, Jaffe, Wilson, and Zak (1985), 26 percent of the children remained well adjusted despite living with abuse. There is mixed data concerning whether girls who grow up in homes with domestic violence experience violence in their adult relationships (Fantuzzo & Lindquist, 1989).

Because the rate of co-occurrence of domestic violence and child abuse is high, understanding their interrelationship is crucial. Such understanding can improve Child Protective Services workers' ability to recognize the risks and to develop viable mother and child safety plans (Mills, 1998b; Mills & Friend, 1998). CPS workers with cases involving both abuses should consider how the dynamics of intimate abuse influence battered women’s behavior. This might help reveal whether a particular battered woman, if given appropriate resources and support, could protect herself and her child(ren) from the abuse. As witnessing becomes interpreted as a form of child abuse, there is increased danger that without proper training, CPS workers will be inclined to remove children from their victimized mothers. However, before partnerships can be forged between the domestic violence and child welfare communities in general, and between CPS workers and battered mothers in particular, it is helpful to understand the tensions and stereotypes which accompany these two practices, and the effect training might have on these issues.

Scholars have previously documented the tensions between battered women’s advocates and Child Protective Services workers (Friend, 1998; Mills, 1998a; Peled, 1996; Schechter & Edleson, 1994). These tensions have been attributed to a number of factors including high caseloads, different philosophies, different terminologies, different mandates, and competition for funding (Schechter & Edleson, 1994). Indeed, battered women’s advocates and CPS workers may find the philosophies and practices of the other threatening to the safety of each of their respective clients.

CPS workers may hold unexamined assumptions about battered women. The worry is that these beliefs influence their assessments and intervention decisions in ways that are destructive to the mother-child unit. This is consistent with data generated by Saywitz and Lyon (1997) which suggests that spouse abuse is used to support petitions to remove children from battered mothers. They found in Los Angeles County, the largest dependency court in the country, that one-third of the cases which involve petitions for the removal of children, also involve spouse abuse. The allegation of domestic violence, according to Lyon, "strengthens the claim that the mother was unable (and possibly unwilling) to protect the child from abuse" (1997, p.3).

Magen and Conroy (1998), analyzing pre-test and post-test data found that training on domestic violence produced statistically significant changes in CPS workers’ attitudes toward intimate abuse as measured by the Inventory of Beliefs about Woman Abuse (IBWA) (Saunders, Lynch & Greyson, 1987). Implicit in their findings is the suggestion that negative attitudes were pervasive before the training was provided. Following the training intervention, Magen and Conroy (1998) found that CPS workers had less of a belief that women gain from being abused, less of a belief that woman abuse is justified, more of a belief that the batterer is responsible for the abuse, and more of a belief that help should be given to battered women. Similarly, Mills and Yoshihama (1997) found that CPS workers’ tolerance of domestic violence changed significantly from pre to post test data following training on domestic violence. Before the training, CPS workers were more tolerant of men’s use of violence against women than they were after the training was provided.

These findings combined with the evidence of the tensions between CPS workers and battered women and their advocates, suggest the importance of training which can help bridge gaps and debunk myths which contribute to systems interventions which punish battered women for their partners’ abuse.

Many battered women are deterred from reporting their abuse and their children's maltreatment because of fear of losing custody (United States Advisory Board on Child Abuse and Neglect, 1995). The extent to which battered women actually "lose" custody of their children due to their status as battered women is unclear. However, the perception that reporting domestic violence could trigger a custody dispute is likely to stifle communication between the CPS worker and the battered woman (Felder & Victor, 1996). Nationally, these issues have just begun to be addressed by Child Protective Services agencies. Many of the standard methods used by child welfare agencies to investigate, assess risk and plan services have been found to be inappropriate for families experiencing domestic violence (Aron & Olson, 1997a).

The Department of Health and Human Services expressed interest in these issues in 1995, when the Federal Register announced that the National Center on Child Abuse and Neglect and the Children's Bureau Discretionary Funds Program were requesting applications to conduct child abuse training and research in innovative exemplary practice in domestic violence. The Request for Proposal directed applicants to develop a training package to address "working with families contending with domestic and/or community violence" (p. 24715).

The Department of Health and Human Services received 200 proposals (M. Mannes, October, 1995, personal communication) and five groups were awarded approximately $110,000 over a two year period. These five groups included: Columbia University School of Social Work (New York), Simmons College School of Social Work (Boston), Temple University's Center for Social Policy and Community Development (Philadelphia), Tennessee State University (Nashville), and UCLA's Center for Child and Policy Studies (Los Angeles). These five grantees became loosely connected at the suggestion of the DHHS program director. Four of the five groups, including each of the groups involved in this research, met informally in March 1997.

To understand more about the experiences of these five programs whose broad mandate was to develop models to integrate domestic violence into CPS practice, data were collected from four of the five grantees. This section begins with an overview of each of the four grantees’ projects. Next, we present the scope or reach of the projects, the development of curricula, the evidence of interagency collaboration, and the common challenges faced by each grantee, including what the institutional climate was for systematically integrating domestic violence into CPS agency practice in a way that does not automatically prejudge and dismiss the battered woman. We conclude with implications for future practice and policy.

Columbia University School of Social Work developed a two-day training curriculum for child protective services workers in New York City. They also developed a World Wide Web site for dissemination of information about the project. The first day of training focused on self-awareness of victimization and attitudes about domestic violence. Day two focused on batterers, assessment, and intervention.

Simmons College worked with the Department of Social Services' (Massachusetts’ CPS Agency) Domestic Violence Unit, a unique and important innovation in child welfare practice. This is a separate unit, based in the central agency, staffed at that time by eleven domestic violence advocates and two supervisors. Using one area office as a pilot site, the Simmons grantees conducted a needs assessment, provided extensive on-site training for supervisors, and evaluated the outcomes. The training included seminars throughout the two years. After the seminars were delivered and evaluated, a manual for domestic violence training was created for the Department. The Simmons grantees also supported the development of, facilitated, and evaluated an interagency consultation team, which continues to meet. Curriculum development for social work students included delivering a course at Simmons on domestic violence and child welfare and collaborating on two workshops for Massachusetts social work faculty.

Temple University developed the "Training Project on Collaborative Responses to Community and Domestic Violence." They planned two sets of training sessions for violence prevention and intervention professionals and brought together a diverse group of violence prevention and intervention workers to foster understanding of the problems of family violence, to identify the roles and mandates of the different systems and to establish linkages between them, to recognize the interrelatedness of family violence, and to propose policy systems change.

UCLA's training program took a two-fold approach. First, UCLA trained a large cohort of workers and supervisors on the intersections of domestic violence and child abuse, and developed an assessment instrument that would make their interventions more effective when domestic violence was suspected. This training was a one day program. In addition, UCLA trained a select group of CPS workers and supervisors over a six-day training period as "fellows" or experts in domestic violence. These experts were scattered throughout Los Angeles County and were viewed as the domestic violence resource person by the other workers and supervisors in those offices. Policy change was key to UCLA's interest in the training program; they wanted to ensure that their Assessment Instrument and intervention method became integral to child abuse practice in Los Angeles (Mills, 1998b; Mills & Friend, 1997).

The specific successes of the four programs have been many and varied. The first domain we examined was the scope of the training. Some projects assessed scope by counting the number of trainees; others focused on culture or systems change.

The Columbia Program trained over 400 CPS workers in New York City in a sophisticated method for intervening in cases involving both abuses. In addition, they had 3000 "hits" on their web site and they mailed out several hundred copies of their training manual (http://www.columbia.edu/~rhm5). CPS workers who participated in the Columbia training completed a consumer satisfaction measure at the end of the two day training program. Over 50 percent of those trained rated the overall training program as good or excellent. Various components of the training were also rated highly such as the usefulness of the training manual and small groups exercises. The design of the training curriculum and preliminary evaluation data were presented at a national conference.

Simmons College worked with a previously established Domestic Violence Unit at the Massachusetts CPS agency (DSS). Simmons was not concerned with training large numbers, 15 supervisors being the target group. Rather, they chose intensive involvement with one urban office as a way to assess needs, build curriculum, and pilot training materials for the Domestic Violence Unit to use system-wide. Part of the "curriculum" was development and support of an interagency domestic violence consultation team, including representatives from criminal justice, women’s advocacy, and batterer treatment programs as well as children’s services and DSS.

Temple's project involved training more than child welfare workers. The diverse training group included domestic violence victims' advocates, public school educators, University faculty, police officers, drug and alcohol counselors, city government representatives, community-based organization representatives, hospital workers and others. They provided two training sessions to 200 professionals total, the purpose of which was to bring together a diverse group of violence prevention and intervention workers, especially child welfare workers, to foster better understandings of the problems of family violence.

Of the over 900 CPS workers and supervisors trained by UCLA in the Southern California region, approximately 120 of those workers and supervisors participated in the six-day fellows training session over two six-month periods. Workers and supervisors came from every region in Los Angeles County and nearly all CPS workers in Orange County were trained. Both the training concepts and the Domestic Violence Assessment Instrument developed for CPS workers were presented at three national and two international conferences. The Assessment Instrument was widely disseminated beyond trainees and was an integral part of a California Social Work Education Center (CalSWEC) curriculum developed as an off-shoot of the federal grant (Mills & Friend, 1997). CalSWEC trains MSW students to become California's child welfare workers.

A key feature of nearly all the programs was the development of a training curriculum involving multiple players from the child welfare and domestic violence communities which led to new and important alliances. Columbia gathered a diverse group of representatives together from the child welfare and domestic violence communities and worked closely with one zone, or district, at CPS which was testing a special domestic violence assessment instrument. They developed a training manual that was widely disseminated. According to pre test post test data, Columbia achieved a statistically significant change in worker attitudes toward domestic violence (Magen & Conroy, 1998). More specifically, they were particularly successful in changing CPS male worker attitudes (Magen & Conroy, 1998).

Simmons worked with the Domestic Violence Unit to bring together an interagency team, which proved invaluable in their cross-training approach. This team met and continues to meet monthly to consult on DSS cases and to surface organizational issues among the agencies and services represented. Team members participated in an evaluation conducted by research faculty and students from Simmons. Curriculum development for DSS is summarized in a training manual, produced for the Domestic Violence Unit, with modules to be used flexibly according to the training needs of particular groups. Curriculum in child protection and domestic violence for social work education was advanced through a course to be offered annually at the school of social work. This course was first given in Spring, 1997, with participation of the Domestic Violence Unit and supervisors from DSS. The Simmons grantees collaborated with the Massachusetts NASW Committee on Domestic Violence and Sexual Assault to produce two conferences for social work faculty on domestic violence in the social work curriculum.

Temple found strength in their Advisory Board, members of which represented many different systems involved in violence prevention and intervention. Because their trainees came from diverse professional organizations and agencies, networking was facilitated between people and organizations that would not otherwise have the time or inclination to communicate. The training model consisted of 24 hours of training in six three-hour sessions. An additional six-hour session on an elective topic was necessary. Those who completed the training session were awarded a certificate for Innovative Services to Violent Families. The topics covered included: new perspectives on family violence; focus on the victims of violence: case studies and applications; focus on the perpetrator of violence: case studies and applications; focus on systems: legal issues at the macro level; focus on human services: applications at the micro level; collaborations: working together to end family violence.

UCLA's interagency collaboration was also integral to its success: law enforcement, prosecutors, judges, alcohol and drug treatment agencies, adult protective services, domestic violence advocates and CPS supervisors were involved in the planning of the trainings and in the development of the curriculum. UCLA’s curriculum comes complete with overheads and handouts using the 1-day training as the basis for changing attitudes and teaching assessment and intervention techniques.

Each of the four training programs faced unique challenges when attempting to integrate domestic violence and child abuse practice. However, some of the obstacles they faced were similar and warrant discussion. These obstacles can be summarized under two main headings: Tensions between priorities on children or women; and characteristics of CPS culture.

CPS philosophy elevates the interests of the child to the highest priority. Some of the grantee groups felt that this priority came at the expense of the battered woman, exemplified by the difficulty CPS workers may have viewing a woman’s "staying" in the abusive relationship as anything other than complicity. Such views seemed to prevent some trainees from realizing that it was the batterer, not the battered woman, who was responsible for the abuse. If training is not successful at changing this view, workers may continue to judge the actions of battered women negatively and to hold them at least indirectly responsible for the abuse inflicted on the child.

Other groups formulated the philosophical tensions between the interests of battered women and the interests of their children as an inevitable tension in priorities given CPS’ mandate. In some situations it was difficult for CPS workers to see the mother’s and child’s interests as linked, and to take what felt like the risk of trusting the woman’s judgment. Anxiety about children’s safety and helplessness to intervene with violent men sometimes left workers blaming battered women for their abuse and taking actions which could be seen as punitive.

All four of the funded projects experienced the CPS culture as difficult to penetrate. Specifically, all the programs, even the Simmons program which had access to a specialized Domestic Violence Unit and the Temple trainers who had a 20 year history of working with the Department of Health Services, found an absence of formal mechanisms to facilitate collaboration between a university or Center and the public child welfare agency. This was despite the fact that each group was operating in a CPS environment that had already accepted domestic violence into their child abuse assessments and interventions. Given that a condition of the grant was the CPS agency’s participation, it was striking that this difficulty was both significant and persistent.

Each group emphasized the importance of support from key agency players. Agency managers’ support for the training effort was critical to buy-in from other CPS agency personnel. Without such support from the top, participation was limited. In addition, the crisis-driven and reactive nature of CPS work made attendance at training sessions problematic. Even those groups that felt top administrators had "bought in," found that attendance at trainings was sometimes low and that other agency pressures impeded full participation. Finally, CPS workers are understandably leery of training that seems to involve additional demands in an already overwhelming job, particularly if they perceive that trainers are not themselves familiar with the realities of CPS work.

A less expensive but equally viable approach to integrating domestic violence practice into child protection is to train a cohort of CPS workers who become specialists. Los Angeles’ method of training "fellows" or "experts" in domestic violence serves a similar function as those who are hired as Domestic Violence Specialists, without additional agency expenditure. However, agency buy-in in adopting this approach is key. If the unique training and practice experience of these expert workers goes unrecognized by supervisors and administrators, they can quickly become marginalized and dismissed. UCLA attempted to overcome this problem by anointing supervisors to deputize workers to become fellows.

Fully integrated training programs are critical for success. In other words, the issues which plague the integration of domestic violence into child abuse practice are so vexing that training should directly involve CPS agency personnel as trainers. External or outside experts have difficulty penetrating, in any significant way, the culture without true investment from the agency itself.