Conjoint Interventions for Child Witnesses of Domestic Violence

Introduction

One of the most damaging aspects of domestic violence, and also one of its least studied, is its effect on children who witness it. Most children whose parents are involved in domestic violence witness that violence, and these children frequently experience behavioral and emotional problems as a result of that exposure. Despite these facts, the development of intervention programs aimed at helping children who have witnessed domestic violence is fairly recent.

One of the key issues surrounding intervention is whether children's programs should include the non-offending parent. The existing literature does little to resolve that question, because some studies have found that conjoint intervention is helpful while others have failed to support those findings. To address this issue, Sullivan, Egan, and Gooch (2004) studied the effect of conjoint interventions on adult and child victims of domestic violence.

In order to study the effect of conjoint interventions, Sullivan et al. (2004) followed a nine-week group intervention program designed to address the needs of female domestic violence victims and their children who had witnessed violence. The intervention had several goals: increasing parenting skills, providing coping skills to mothers and children, safety planning, and decreasing post-violence stress (Sullivan et al., 2004).

One troubling aspect of the study is that it concentrated exclusively on female victims of domestic violence. Although statistics vary widely, it is now widely accepted that females can be abusers as well as victims. In addition, many households are non-traditional and may feature both an abusing father and a victimized father. Therefore, the results of the study are valid only for adult female victims of domestic violence and their children. Gender differences in intervention outcomes may change the results if applied to male victims and their children.

Another troubling aspect is that the intervention programs did not appear to address abuse by the mother. Secondary abuse by a parent in a domestic violence situation is receiving increasing attention in the field, and an intervention program's failure to address it raises concerns. Sullivan et al. (2004) make clear that the intervention came from a feminist perspective and did not include any victim blaming. However, advocates for children may question whether a program that fails to address a child's feelings about a parent's failure to protect them can truly be effective. Practitioners concerned with age-appropriate intervention may also be troubled by an emphasis on safety planning and coping skills rather than on ensuring that children are not required to live in environments where such measures are necessary. There is ongoing conflict between child advocacy and domestic violence advocacy groups on this point, and it is evident that Sullivan et al. approached the issue from the perspective of adult victims rather than child victims.

Methods

The study followed members of a nine-week intervention program aimed at helping adult and child victims of domestic violence. Subjects were included only if both the children and their mothers completed the entire nine-week program. Those who completed the study included 79 children and 46 mothers. Sullivan et al. (2004) also analyzed the sample of subjects who completed the study compared to those who began but did not complete the intervention program, finding no significant differences in race, gender, or age. However, there were significant differences in the mothers' pretest parental stress scores and the children's trauma symptom checklists (Sullivan et al., 2004). Because children presumably had no control over whether they completed the intervention program, the meaning of those differences is difficult to discern. In any case, the results of the study should not be generalized to populations that do not share similar pretest profiles. Furthermore, the results may be skewed by the possibility that completing mothers were more motivated to help their children than mothers who did not finish the program.

Mothers and children both completed two types of measurements aimed at studying the effectiveness of the intervention. All measurements had "known validity and reliability coefficients as well as norm distributions and clinical cutoff scores" (Sullivan et al., 2004). Children were measured using the Child Behavior Checklist (CBCL), completed by parents, which assessed behavioral problem areas in three domains: activities, social, and school, as well as internalizing and externalizing dimensions. Sullivan et al. (2004) chose a clinical cutoff score of 67 to separate normal from clinically symptomatic groups.

Children were also assessed using the Trauma Symptom Checklist for Children (TSCC). "The TSCC subscales include Anxiety, Depression, Anger, posttraumatic stress disorder (PTSD), and three measures of dissociation" (Sullivan et al., 2004), with a clinical cutoff of a T-score of 67. The Parenting Stress Index (PSI) was completed by parents and includes both a child domain and a parent domain. "The child domain measures parents' perceptions of their children's distractibility or hyperactivity, adaptability, reinforcement of the parents, demandingness, mood, and acceptability" (Sullivan et al., 2004). Children above the 85th percentile on the PSI "might be characterized as difficult" (Sullivan et al., 2004). "The parent domain measures parents' perceptions of their own competence, isolation, attachment, health, role restriction, depression, spouses, and life stress" (Sullivan et al., 2004). The final measurement was the Children's Perception of Interpersonal Conflict (CPIC), a self-report instrument for children aimed at assessing self-blame for the domestic violence in the home.

One of the main problems with the measures used is that active victims of domestic violence were being asked to complete surveys about their children. Generally, research is compromised when conducted by an interested party. It is inconsistent for Sullivan et al. to characterize the mothers solely as victims of domestic violence — thereby excusing their failure to protect their children from witnessing violence — and then to use those same mothers as objective reporters of their children's behavior. It is well documented in domestic violence research that children from violent homes frequently experience emotional and behavioral symptoms, and that both offending and non-offending parents are often unable to recognize when their children need help. The absence of a normative home environment makes it difficult, if not impossible, for mothers to accurately assess whether their children's behavior meets normal standards. This concern could be addressed by having uninvolved parties — such as teachers, caregivers, or therapists — assess the child's behavior. Accordingly, any results of the study should be understood as reflecting the mothers' perceptions of changes in their children's behavior, not necessarily actual behavioral changes.

The data collection methods may also not have been sufficiently sensitive to the target population. Because a mother who permits her child to be abused — regardless of whether she is also a victim — may bear legal responsibility for that abuse, there are legitimate concerns about the honesty of parental responses. No mechanism appears to have been put in place to assess the veracity of the mothers' statements. It is also well established that children from violent homes are frequently discouraged from discussing the violence, or have heard it minimized so often that they minimize both the violence and its effects on themselves. No measures appear to have been implemented to detect or correct for this tendency.

However, if the research is understood as measuring mothers' and children's reported perceptions of the efficacy of a completed conjoint intervention program, the research design is valid. Sullivan et al. (2004) issued retests to verify consistency of results and identified areas of difference between those who completed the intervention and those who did not. They also acknowledged five limitations of the research: small sample size, inclusion of only those who completed the intervention, data collapsed over time, general rather than specific findings, and the absence of a control group (Sullivan et al., 2004).

The findings on children suggested that those who scored above normal prior to the intervention benefited more than children who scored in the normal range prior to the intervention (Sullivan et al., 2004). However, those children continued to score above the normal range after the intervention (Sullivan et al., 2004). The most dramatic improvements were observed in the anger and trauma domains (Sullivan et al., 2004). Perhaps the most significant finding was that the intervention substantially reduced children's self-blame for the domestic violence (Sullivan et al., 2004).

The parental findings were also noteworthy. "Isolation, life stress, and health of the parent were significantly improved at the posttest" (Sullivan et al., 2004). However, the mothers' health scores remained in a nonclinical range during both pre- and posttest, and their life stress scores remained in a clinical range at both time points (Sullivan et al., 2004).

Conclusion