LIBRARY Michigan State Untversity WM-” F... PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. ll DATE DUE DATE DUE DATE DUE Iémwm MSU Is An Affirmative Action/Equal Opportunity Institmion CWMH- ' An Advocacy Intervention for Women with Abusive Partners and Its Effects on their Children Joanna M. Basta A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1991 677-7éf3 ABSTRACT AN ADVOCACY INTERVENTION FOR WOMEN WITH ABUSIVE PARTNERS AND ITS EFFECTS ON THEIR CHILDREN by Joanna M. Basta Little published research has looked at children after they have left a shelter for battered women. No research to date has empirically examined the effects of an intervention on children who witness woman abuse. This study utilized a an experimental design to assess the behavioral adjustment of children whose mothers participated in an advocacy program after leaving a shelter for battered women. Fifty- two mothers and their children were randomly assigned to an experimental (advocacy intervention) or control group (no advocacy services). Interviews with the mothers were conducted at pre- and post-intervention periods (10 week interval). The mothers' victimization, involvement with the assailant, effectiveness in obtaining resources, psychological well-being and stress, and use of disciplinary tactics were assessed. Mothers reported on the extent of their children's exposure to violence, behavioral adjustment, contact with assailant, and feelings about the assailant. One major hypothesis of this study was supported-- mothers in the experimental group were found to have significantly higher levels of perceived effectiveness in obtaining resources than mothers in the control group. The experimental group also showed trends toward having higher levels of life satisfaction and social support. Children of the experimental mothers were not significantly different from children in the control group on behavioral adjustment, experiences of abuse, mothers' involvement, or children's contact with assailant. Descriptive information about the children was also presented. In the ten weeks prior to their pre-intervention assessment, a majority of the children observed their fathers (or father-figure) emotionally and physically abuse their mothers. According to the mothers, 3 significant amount of children were also victims of their father's emotional and physical abuse. Children most likely reacted to their father's violence with fear, anger, and confusion. They also exhibited a wide range of externalizing and internalizing problems. Children's maladjustment was found to correlate significantly with their victimization experiences, mothers' pre-intervention depression levels and post-intervention stress levels. Children's contact with their fathers, experience of violence and internalizing problems decreased over the 10 week intervention period. They also expressed feelings of ambivalence about being separated from their fathers. Copyright by Joanna M. Basta, 1991 In dedication to mothers who strive to make their homes loving and safe for their children and for Marc, who suggested I keep my own name ii ACKNOWLEDGEMENTS The entire process of my doctoral study was an endeavor that consumed 7 years of my life. Many people during those years provided me support and encouragement. Foremost among them is my partner, Marc Sheehan, who endured with me, and provided me invaluable material, emotional and spiritual support. I can't thank him enough. Our relationship, due in part to his support and pride in my studies, survived the vagaries and hardships of "doing a dissertation." I want to thank my advisor and chairperson of both my masters and doctoral theses, William Davidson, for all his expertise and encouragement over the years. He provided me excellent educational and research opportunities which were invaluable. The other members on my dissertation committee were terrific: Deb Bybee, Ellen Strommen and Paul Freddolino. They were nothing but encouraging and supportive throughout the entire process. They all had the knack of being warm and supportive, while also maintaining their professionalism in providing me feedback about my work. I specifically want to thank Deb Bybee for her statistical expertise. She also shared with me her emotions about completing her dissertation that made me feel I was not alone during a iii period in which I had doubts about my abilities. My dissertation support group was a significant source of support for me since I first started attending it in 1987. The members provided me practical advice and a forum to share frustrations and victories. It was also a great way to get know some wonderful women. The original members were Cris Sullivan, Leah Gensheimer, Sara Wood-Kraft, and Andrea Solarz. As members completed their dissertations (a tribute to the group), they were made up of others. Although all members, new and old helped me, I want to especially thank the original members for asking me to join their group. I also want to thank Jennifer Stanley for re- initiating the group and providing me tremendous support in the "last stretch" of the dissertation. This research would not have been possible without the support and encouragement of Cris Sullivan, good friend and Director of the Community Advocacy Project. She provided me the opportunity to work on her project while simultaneously doing my dissertation research on it. Cris or "Sully", has been a mentor and friend. Her dedication and hard work towards empowering women and eradicating violence against them has been both challenging and inspirational to me. Thanks so much Sully--we are carrying on. I also want to thank those women on the Community Advocacy Project--Maureen Rumptz, Gwen Geml, and Cheribeth Tan with whom I worked closely. They always listened to my litany of frustrations and provided uplifting encouragement iv and support. I especially want to thank Maureen Rumptz for her excellent work in co-training the interviewers for my project. Special thanks is extended to all the students and staff persons who conducted the interviews with the mothers. It couldn't have been done without their hard work. Finally, I express enormous gratitude and admiration towards the women and children in this project. They are real people behind the data, who are struggling, often against enormous odds, for better and safer lives. Because of their decision to be interviewed, and to share intimate and painful aspects of their lives with virtual strangers, we have gained information that will hopefully make their lives better and safer. Their courage and their contribution is immeasurable. TABLE OF CONTENTS List of Tables and Figures CHAPTER I Statement of the Problem Terminology Incidence of Problem The Battered Women's Experience Research on the Children of Battered Women Effects of Marital Discord, Separation and Divorce More Recent Studies on Marital Discord and Divorce Summary Effects on Child Witnesses Long term effects of experiencing family violence Summary and Implications for the Proposed Research Potential Moderating Variables Gender Age The child as victim of abuse Measurement and Design Issues Bias in maternal reports of children's behavior Use of control/comparison groups Use of follow-ups Mechanisms By Which Violence/Discord Affects the Children Attachment and separation Modeling and Sex-role identification Discipline practices Community resources and maternal stress The Need for More Research The Current Research The Community Advocacy Project Research Hypotheses and Questions CHAPTER TWO: METHOD Setting Research Participants Recruitment The Sample Summary of obtaining research participants for CAP Eligibility criteria for the current research vi 64 64 65 65 66 66 67 CHAPTER II: METHOD (Cont'd) Dropout rate Characteristics of the sample Tracking the children over time Generalizability of the sample Research Design Procedure Recruitment of Student Advocates The Community Advocacy Project Training and supervision of advocates The intervention Recruitment of Interviewers Training of the Interviewers Interviewing Strategy and Data Collection Subject Retention Consent and Confidentiality Procedures Measures Inter-interviewer Reliability Demographic Data Measures of Community Resources Treatment Strength and Integrity Extent of the Violence Mothers Experienced Outcome Measures Psychological well—being of the mothers Stress, anxiety, and depression Depression Social support Effectiveness in obtaining resources Measures for children's behavioral adjustment Measures for children's experience of violence Intercorrelations Among the Outcome Variables Descriptive Information about the Children Disciplinary tactics of the mother Children's feelings about separation from assailant CHAPTER III: RESULTS The Sample Homogeneity of conditions Mothers' Effectiveness in Obtaining Resources and Its Effect on Children's Adjustment Resources women reported needing after leaving the shelter Effectiveness in obtaining resources Effectiveness and its relationship to children's adjustment General description of children's adjustment Mothers' and Children's Experience of Abuse History of mother's abuse and separations History of children's victimization and witnessing abuse Group differences on children's experience of abuse vii 68 68 70 73 75 75 75 76 76 78 80 80 81 83 83 84 86 86 86 89 90 91 91 91 97 97 100 100 108 112 116 116 116 120 120 120 122 122 122 123 125 127 127 130 132 CHAPTER III: RESULTS (Cont'd) Relationship between abuse and children's adjustment 132 Mother's and Children's Involvement with the Assailant 133 Mother's involvement with the assailant 133 Involvement with assailant and child adjustment 136 Children's contact with assailant 136 Group differences on children's contact with the assailant 137 Children's contact with assailant and children's adjustment 137 Assailant involvement and further abuse 138 Mothers' Psychological Well-Being and Experiences of Stress 138 Relationship between mothers' and childrens' adjustment 142 Exploratory Areas of Investigation 143 Sex Differences on Children's Behavioral Adjustment 143 Children's Immediate Reactions to Abuse 143 Children's Feelings About Assailant 147 Mothers' Disciplinary Tactics 149 Summary of Results 149 CHAPTER FOUR: DISCUSSION 153 Major Findings 153 Mothers' Effectiveness in Obtaining Resources 153 Intervention Effects on Children's Adjustment 155 Mothers' and Children's Experience of Abuse 157 Involvement with the Assailant 161 Assailant involvement and children's behavioral adjustment 163 Mothers' Psychological Well-Being and Children's Adjustment 164 Mothers' depression and stress and children's adjustment 165 Children's Feelings about the Assailant's Absence 166 Mothers' Disciplinary Tactics 167 Limitations of Study 169 Implications for Further Research 172 Larger Sample 173 More Extensive Measurement 174 More Follow-ups 17S Changes in the Intervention 175 Conclusion 17 Appendix A: Administrative Agreement with CADA 178 Appendix B: Participation Agreements 179 viii Appendix C: Appendix : Appendix E: Appendix F: Appendix G: Appendix H: References University Committee on Research Involving Human Subjects Approval of Research; U.S. Department of Health and Human Services Confidentiality Certificate 182 Pre- and Post-Interview on Children 185 Child Behavior Checklist 208 Parenting Stress Index 212 Hassles and Uplifts Scale 218 Conflict Tactics Scale 220 221 ix Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table 10: 11: 12: 13: 14: 15: 16: 17: 18: LIST OF TABLES Demographics of Research Participants Mothers' Relationship to Assailant at Pre- Intervention Experimental Design Measurement Model Mothers' Physical Abuse Scale Reliability Mothers' Psychological Abuse Scale Reliability Psychometric Properties of Modified Quality of Life Scale Uplifts Subscale Reliability Hassles Subscale Reliability Parenting Stress Index Reliability Revised Depression Scale Reliability Mother's Perceived Social Support Scale Reliability Effectiveness in Obtaining Resources Scale Reliability Children's Internalizing Problems Scale Reliability Children's Externalizing Problems Scale Reliability Children's Behavioral Adjustment Scale Reliability Witness Abuse Subscale Reliability 71 74 77 87 95 98 99 102 103 106 107 109 110 111 113 Children's Victimization Subscale Reliability 114 Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table 19: 20: 21: 22: 23: 24: 25: 26: 27: 28: 29: 30: 31: 33: 34: 35: Children's Experience of Abuse Scale Reliability Intercorrelation Among Outcome Variables Mothers' Disciplinary Tactics Scale Reliability Resources Mothers Reported After Leaving the Shelter Average Effectiveness Scores of Mothers Modified Version of the Child Behavior Checklist, Externalizing and Internalizing Subscales Percentages of Specific Child Behavior Problems Reported by Mothers to be True or Very True at Pre- and Post-Intervention Periods Percentage of Mothers Who Experienced Various Forms of Violence and Who Sustained Various Injuries in the Six Months Prior to Study Percentages of Abuse Experienced by Children at the Pre— and Post-Intervention Periods Children's Experience of Abuse Percentage of Mothers Involved and Living With the Assailant at Time One and Time Two Children's Contact with Assailant at Pre- and Post-Intervention Periods Assailant Involvement and Experience of Further Abuse Mothers' Psychological Well Being and Stress Pearson Correlations of Mothers' and Children's Adjustment at the Pre- and Post- Intervention Periods Children's Reactions to Witnessing Abuse Ranked from High to Low Frequency Children's Feelings About the Assailant at Pre- and Post-Interviews Mothers' Reported Use of Disciplinary Tactics at Pre- and Post-Interviews xi 115 118 119 124 126 128 129 131 134 135 139 140 141 144 146 148 151 152 CHAPTER ONE W While wife battering has become a widely recognized social problem and research on the topic has increased greatly, little research into the effects on the children who witness the battering of their mothers exists. Also, there are very few empirical investigations of interventions for battered women and their children. The current study examined how an advocacy intervention designed for women exiting a battered women's shelter affected their children. First, before the review of the literature on children of battered women is presented, a very brief overview of the battered women's experience will be given. Terminology As in Sullivan (1988), this study used the term wife/woman abuse interchangeably instead of conjugal violence, domestic violence or spouse abuse. The latter terms are gender neutral and hide the fact that women are primarily the victims of the battering by the men (Okun, 1986; Straus, Gelles & Steinmetz, 1980). Women are not only battered by their husbands, but also by ex-husbands, boyfriends, and ex-boyfriends. Therefore, the term wife abuse/battering will also include these non-married women. I f r Woman battering is centuries old and has been condoned and upheld by traditional law and custom. Husbands 2 justified beating their wives by viewing them as property. In fact, only a hundred years have passed since men lost the legal right to use physical violence to control their wives (Carlson, 1987). Not until the last twenty years, with the onset of the Women's Movement, has wife abuse become a recognized social problem. In 1972, the first 24-hour house of refuge for battered women and their children was set up in a small derelict house in Chiswick, a borough of London, England. From this small beginning in London, an entire social movement of national and international proportions was created (Dobash & Dobash, 1988). Political and community organizers greatly contributed to documenting the scope of the problem. Their efforts gained extensive media and public attention which resulted in the creation of shelters and efforts to change outdated laws and criminal justice practices (Morash, 1986). In the United States over 600 shelters exist today. Although very little research on wife abuse existed before the 1970's, since that time the scholarly literature on the topic has increased dramatically. Several empirical studies were done that have impacted service providers, academicians, and social policy-makers (Sullivan, 1988). One of the most widely recognized and cited studies was done by Straus, Gelles, and Steinmetz (1980). They interviewed a nationally representative sample of 2,143 couples with questions from their Conflict Tactics Scale (1979). 3 Respondents were randomly selected heterosexual couples that were asked to self—report the number of incidents in one year for violence ranging in severity from throwing things to using a knife or gun. When possible both the man and woman were questioned. One out of every six wives reported that she was struck by a husband during the course of her marriage. Thirty-eight incidents of severe wife beating per one thousand women were reported in 1975. In 1985, Straus et a1. replicated their survey with the aid of Louis Harris and Associates. Over 6,002 households were interviewed over the telephone, in contrast to the first survey which was an in-person interview. Reports of severe wife beating declined from the 38 incidents per one thousand women in 1975 to thirty per one thousand in 1985. Although this was a significant decrease (27%), the remaining numbers of women who are severely abused is still high (over a million and a half). High rates of battering have been reported by other researchers with more limited samples (Moore, 1979; Schulman, 1979). The National Crime Survey (1984) has estimated that there are 1.06 million wife assaults per year, with 13 assaulted wives for every assaulted husband. Unfortunately, in many of the families where wife abuse occurs, there are children who witness it. Carlson (1984) projects that at least 3.3 million children yearly are at risk of exposure to wife abuse. Her estimates are based on 4 Straus et al.'s (1980) estimations of the number of households affected. The result is that children have become the unintended victims of wife abuse (Rosenbaum & O'Leary, 1981). The Battered Woman's Experience In order to understand the possible effects on children who witness the abuse of their mothers, knowledge about what the mothers have experienced is necessary. Although it appears that abuse increases in severity and frequency over time (Dobash & Dobash, 1979; Okun, 1986; Roy, 1977; Walker, 1979), most women stay with or return to an abusive partner rather than leave the relationship permanently (Dobash & Dobash, 1979; Gayford, 1978; Gelles, 1979; Hofeller, 1982; Roy, 1977; Schulman, 1979; Strube & Barbour, 1983). However, the majority of battered women would rather leave their batterer than stay in such a relationship (e.g. Snyder & Fruchtman, 1981; Snyder & Scheer, 1981). Similarly, in a survey of more than 6000 women in 50 Texas shelters, Gondolf & Fisher (1988) have found that battered women respond to more severe abuse with increasing helpseeking. The helpseeking is an effort to gain safety from the dangerous men who engage in general antisocial behaviors as well as wife and child abuse. It also appears that the fewer resources and the more children the woman has, the more she is motivated to seek help. Gondolf & Fisher (1988) argue that battered women remain in abusive situations not because 5 they have been passive but because they have tried to escape unsuccessfully. So why does the battered woman remain in the relationship while making efforts to escape it? Gondolf & Fisher hypothesize that it may be a "system failure" rather than failure on the part of the battered woman. Despite the fact that the battered woman may have contacted a variety of helping sources in response to the violence, these helping services may not have been able to stop the abuse or assure her safety. Second, these authors also suggest that the antisocial behaviors of the batterers are not adequately addressed by currently existing interventions in our society. Swift (1987) outlines some of these areas where community support systems fail to address the needs of battered women. These systems include the law, courts, clergy, and the medical care system. Lerman (1986) (cited in Swift, 1987) reported on the unresponsiveness of the criminal justice system to arrest, prosecute and convict wife-heaters. She traces the dramatic drop in the number of wife abuse cases at each stage in the criminal justice process. For example, out of 15,000 domestic violence calls to police in a major mid-western city over a nine-month period in 1979, police filed reports in less than five percent of the cases, and made arrests in only three percent. In another study cited by Lerman (1986) 6 prosecutors were found to dismiss charges in almost half (45%) of the assault cases in which arrests were made. Alsdurf (1985) (cited in Swift, 1987) found in a survey of protestant clergy in the United States and Canada that one- third of the respondents felt that the abuse would have to be severe in order to justify a Christian wife leaving her husband. Stark (1984) (cited in Swift, 1987) outlined how the medical response to wife abuse is inadequate and inappropriate (medicating the women with psychotropic drugs) and how physicians generally underestimate the abuse by not identifying cases in their practice. Finally, even the woman's own family can serve to maintain the violence: for example, some family members encourage the woman to "put up with it," and/or place the blame on the woman (Straus, 1980). Needless to say, there are many obstacles that may prevent battered women from leaving their assailants. Many of these obstacles lie in the failure of community services to adequately address women's needs. Considering the extreme difficulties this poses for women, what does this mean for their children? How would a program specifically designed to address some of the most pressing needs of battered women who have recently exited a shelter also affect their children? Before this latter question is addressed, the next section reviews literature that examines how the children who witness battering are affected. In 7 addition, those factors found in the literature that are postulated to mediate the adjustment of children who witness battering are also discussed with regard to their relevance to the current research. e ‘l e In the recent past, the research literature on battered women has placed its primary focus on the women and not on their children. Children of battered women have been called the "forgotten victims" of domestic violence (Pfouts, Schopler, & Henley, 1982). However, research on the children has been steadily growing in the past 10 to 15 years. What we knew previously about the children was based on anecdotal reports from clinicians and shelter workers (Carlson, 1984). Research into the effects of marital discord and separation and divorce on children has also been used to extrapolate the possible impact on children who witness wife battering. A brief overview of the literature on marital discord, separation and divorce, and its effects on the children will first be given followed by a review of the studies focusing directly on children who have witnessed battering of their mothers. WW: A large body of both clinical and research literature has been generated which shows that children suffer at least short-term developmental disruptions, emotional distress and behavioral disorders from their parents' divorce (Emery, 8 1982; Hetherington, Cox & Cox, 1985). Two major studies on the impact of divorce on the children are the most widely cited works in this area. Hetherington, Cox & Cox (1978) and Wallerstein & Kelley (1980) conducted two separate longitudinal studies looking at the effects of divorce on children. Hetherington et a1. (1978) followed forty-eight recently divorced, white, middle-class couples and compared their households, their relationships with their children, and the functioning of their families with those found in a control group of non—divorced families. Child behavior was indexed by such measures as behavioral observations, teacher ratings, peer nominations, child self-report, and parent report. It was found that most children exhibit short-term developmental disruptions, emotional distress, and behavior disorders (Hetherington, Cox & Cox, 1985). More specifically, children in the divorced families, in comparison to the control group, showed more anti-social, impulsive acting out disorders, more aggression and noncompliance, more dependency, anxiety and depression, more difficulties in social relationships, and more problem behavior in school. Externalizing, impulsive antisocial behavior was more consistently found than internalizing disorders such as withdrawal, depression, and anxiety. In addition, these effects were more enduring for boys than for girls. Wallerstein & Kelly (1980) conducted a clinical study 9 of sixty families who were filing for divorce and volunteered to receive help for themselves and their children through a divorce counseling service at a community mental health center. Each child and his/her parents received four to six individual clinical sessions and were interviewed at these sessions for the research. The sample included 131 children and adolescents--48% males and 52% females--who had no psychological difficulties or development problems. Slightly more than half of the children were between ages two and eight when first seen; the remainder were between nine and eighteen. The families were largely middle to upper middle class. Findings from the initial assessment at the time of their parents' breakup showed considerable acute distress in over half of the children. Thirty-six percent (36%) of the sample (N = 46) faced the divorcing period with good psychological functioning, defined as high self-esteem, no depression or anger, good social relationships, good school performance and little anxiety. Another 48% (N = 63) had a mixed rating of socio-psychological functioning. For example, although some of these children were doing well academically, they had ungratifying relationships with their parents and peers. The remainder of the sample (17%) entered the divorcing period with serious problems in coping. No significant age differences in overall psychological functioning were found at the time of the 10 divorcing period. However, when sex alone was considered, there was a trend toward better psychological functioning among girls (Wallerstein & Kelly, 1980). Although distress occurred at each age, the children's responses varied in their expression according to age and developmental sequence (Wallerstein, 1985). In fact, Wallerstein & Kelly (1980) observed that one of the major findings of their study was how the children's initial responses to the divorce were closely age-related. Pre- school children responded with fear and bewilderment, fantasies to explain or deny the separation, and aggression in play. Young school age children showed a pervasive sadness, yearning for the departed parent, anger at the custodial parent, and conflicts in loyalty for either parent. The older school age children seemed to have better coping abilities for dealing with the divorce. Compared with the younger children, they seemed more organized and less immobilized by their worry and grief. Adolescents were particularly worried over sex by, and re-marriage of their separated parents. Dating by their parents often caused concern and anxiety. Both Hetherington, Cox 8 Cox (1985) and Wallerstein & Kelly (1980) conducted long-term follow-ups on their samples and found that the psychological effects were long-lasting for some of the children. Hetherington, Cox & Cox (1985) found that at six years following divorce, sons in divorced 11 families showed more externalizing behavior and were sometimes reported showing more internalizing behavior and less social competence than sons in non-divorce families. However, daughters in families with a divorced, nonremarried mother were very similar in adjustment to those in non- divorced families. Wallerstein (1985) found that after ten years following their parents' divorce, almost half of the sample still retained some anger about the divorce, and a significant number were burdened by vivid memories of unhappy events that occurred during the divorce. Forty children from 26 of the 60 original families were interviewed when they ranged in age from 19 to 29 of age. One-third of the women appeared wary of commitment and fearful of betrayal. Although over half of these children were coping well, almost all considered the divorce to be a continuing major influence in their lives. Several limitations of these two longitudinal studies exist. Wallerstein & Kelly's study is primarily clinical in nature, so no efforts to obtain a representative sample through random sampling were conducted, no control or comparison group was utilized, and assessments were made through clinical interviews. Hetherington, Cox & Cox were more rigorous; however, as in Wallerstein & Kelly's study, their sample was limited to predominantly white middle to upper middle class subjects. Also, in the Hetherington 12 et a1. study the effects of divorce cannot be sorted out from the pre-existing family conflict/problems that may have led to the divorce. Porter & O'Leary (1980) studied the effects of marital discord on children by assessing whether overt hostility between parents would be related to childhood psychopathology in boys and girls of the family. Sixty—four two-parent families who had applied for treatment for their children (5 — 16 years old) represented their sample. Age was controlled by dividing the sample into two age groups-- young school age and adolescent. Mothers rated their children's deviance on the Behavior Problem Checklist (BPC) (Quay & Peterson, 1979). This measure included four subscales: conduct disorder, personality disorder, inadequacy-immaturity, and socialized delinquency. A total pathology score for each child was computed from those subscales. Porter & O'Leary also assessed marital adjustment and overt marital hostility through maternal report measures. Overt marital hostility correlated signficantly with many behavioral problems for boys, but not for girls. No age differences were found. These results should be considered with caution for several reasons: a clinical sample was used; no control group was employed; the assessments relied solely on maternal report for both child and marital problems. In a similar study, Emery & O'Leary (1982) used a 13 clinic sample of 50 children between the ages of 8 and 17 to predict whether maternal and child ratings of marital discord would correlate with behavioral problems for boys rather than girls; whether conduct problems and delinquency would correlate more strongly with marital discord; and whether girls would report less marital conflict than boys as compared to their mothers' reports of marital conflict. The Behavior Problem Checklist was used to measure behavior problems in the children. The first two hypotheses were supported-—marital discord was significantly correlated with behavioral problems for boys and not for girls, and with conduct problems and delinquency rather than internalizing type of behaviors. Both girls and boys were similarly accurate in predicting their mother's ratings of marital discord, implying that although both genders are exposed to equal amounts of parental hostility, they are differentially affected by its exposure. This study shares the same limitations as the Porter & O'Leary study described above, so the results should be viewed with caution. Emery (1982) published an extensive review of the literature on interparental conflict and the children of discord and divorce. Evidence in his review showed that interparental conflict (open hostility between married, separated, or divorced parents) not ggp§1§t123_from parents may be the principal explanation for the association between child behavioral problems and marital discord and divorce. 14 Comparisons between children from divorced families and those children separated from their parents by death showed that children from divorced families had more behavioral problems. In addition, children from divorced but conflict- free parents were less likely to have problems than children from married parents who did engage in conflict. This review also identified common effects on the children. Although Emery found a somewhat conflicting pattern of results for several variables that might explain the relationship between marital discord and child behavior problems (i.e., age, separation, attachment, modeling) in every investigation, whether of questionable or sound methodology there was a significant relationship between some form of uncontrolled or externalizing behavior and marital discord. The associations between overcontrolled behaviors and marital discord were inconsistent. There is evidence to show that the amount and type of interparental conflict children observe differentially affects the children. For example, openly hostile conflict makes the child predisposed to future problems. In agreement with the studies cited earlier, Emery found that marital discord affected boys and girls differently. Evidence across the studies shows that boys seemrto be more affected than girls, and that boys exhibit more undercontrolled behaviors. Multiple explanations for this relatively consistent finding have been suggested by 15 Emery: the effect of same sex parental modeling of behaviors for their children (e.g., the men may be aggressive and engage in undercontrolling behaviors); the impact of prescribed sex roles on children's behavior; inconsistent parental disciplinary practices; how each sex handles stress; measurement sensitivity to girls' responses; and age in which the problem behavior manifests itself. Emery argues that girls may be just as affected as boys. Since base rate data on childhood disorders in the United States suggest that girls are predisposed toward overcontrol and boys to undercontrol (Ross, 1980), it may be that parents do not recognize girls' overcontrolled behavior or they tolerate it more than the undercontrol of the boys' behavior. For the age effects, Emery found that the results across studies were inconsistent. Some studies have found no effects (e.g. Porter & O'Leary, 1980) and others have found detrimental effects on different groups (e.g. Hetherington, 1972; Wallerstein & Kelly, 1980). WWW. Reid 8 Crisafulli (1990) examined the relationship of marital discord and child behavior problems through a meta— analysis of the results of published studies through 1988. Their results corroborated Emery's literature review, in that they found a positive relationship between marital discord and child behavior problems. Similar to previous 16 literature, this relationship was also stronger for boys than for girls. They also found that this relationship was stronger when parents were used as the sole data source. Examination of the studies that used clinic samples did not indicate a stronger relationship between marital discord and child behavior problems. This is an interesting finding since it has been noted that using clinic populations to study marital discord is limited because they have a restricted range of symptoms, and parents may be predisposed to report more problems in their children in order to justify receiving treatment (Emery, 1982). Allison & Furstenberg (1989) estimated the effects of marital dissolution on several measures of children's well- being at two time points using a representative sample of 1,197 children. Using reports from parents, teachers and children, children from divorced or separated families were in greater psychological distress and had poorer academic performance than did children from intact families at the time two period. They also found more adverse effects for younger children. In contrast to previous findings, girls were reported to have significantly more problems than boys. However, the amount of variance that marital dissolution contributed to the prediction of child well-being was small (3% or less), indicating that other factors beside marital dissolution may be contributing to the children's adjustment in families of divorce or separation. A major virtue of 17 this study was that it employed a large representative sample. However, there were several limitations, such as problems in the outcome measures, and inability to control for several key variables such as the adjustment of the child before the divorce/separation occurred. Long, Forehand, Fauber, & Brody (1987) assessed the social and cognitive competence of 40 young adolescents who were from divorced and intact homes. Adolescents from recently divorced homes perceived their cognitive and social competence to be lower than did those from intact homes. Reports from independent observers and teachers indicated that those children who had high parental conflict in their homes, regardless of marital status, had lower social and cognitive competence ratings. The researchers concluded that parental conflict, not parental divorce, is the critical variable associated with adolescents' cognitive and social competence. This finding supports Emery's (1982) and other researchers' contention that importance must be placed on parental conflict in intact and divorced families. The lower levels of self-perceived competence found among children of recently divorced families was explained by the acute distress that children may experience when their parents divorce. In addition, Long et al. found no sex differences on the outcome measures. Taking one step further, Jouriles, Murphy & O'Leary (1989) investigated whether marital aggression predicted 18 children's behavior problems more accurately than marital discord. Using 87 couples who requested marital therapy, Jouriles et al. found that children in maritally aggressive families exhibited problems at a clinical level more frequently than discordant, ngg;ggg;g§§iyg families. These results should be viewed with caution since the children's behaviors were reported by parents who came to a clinic for marital therapy. SQEEQII. A substantial amount of literature has been published examining the effects of marital discord and divorce on children. Evidence across studies of the impact of divorce on children over time and the impact of marital conflict on children from non-divorced families show that children suffer from a range of both undercontrolling behaviors (e.g. aggression, disobedience, dependence, anti-social behaviors) and overcontrolled behaviors (e.g. anxiety, withdrawal, depression). The longitudinal studies on divorce showed that the impact on children was relatively short-term, although there is some evidence to show that the impact may be enduring for some children (Wallerstein & Kelly, 1985; Allison & Furstenberg, 1989). A fairly consistent finding across studies was the relationship between undercontrolling behaviors and marital discord. Boys more often exhibited these behaviors than girls (Emery, 1982; Reid & Crisafulli, 1990). However, in more recent studies 19 that have examined sex differences, this finding has not been supported. How age affects children's adjustment to marital discord and divorce is not clear. While Allison & Furstenberg (1989) found that younger children seem to be more affected, Wallerstein (1985) suggests that children of all ages may be affected by marital discord and divorce, and their responses to it may vary according to age and developmental sequence. In the next section, results from the published research into the effects on children who witness abuse of their mothers will be reviewed. These findings will be compared with the marital discord and divorce literature. Effigggg on Child Witnesses The clinical and anecdotal reports on battered women's children vary widely, and although the observations have not been systematically gathered, the range of problems and behaviors reported are congruent (Carlson, 1984). The reports show that the children's reactions to witnessing the battering are severe and pervasive (Christopoulos et a1., 1987). Somatic complaints such as sleeping difficulties (Davidson, 1978; Levine, 1975), stomach-aches, and head- aches (Carlson, 1984; Flemming, 1978; Hilberman & Munson, 1978; Layzer, Goodson, & DeLance, 1986) were observed. Behavioral and emotional reactions have also been reported such as anxiety (Barnett, Pittman, Ragan & Salus, 1980; Layzer, Goodson, & DeLance, 1986; Levine, 1975; NSPCC School 20 of Social Work, 1977), increased aggression (Hilberman & Munson, 1977—78; Hughes, 1982) ambivalent feelings toward their parents (Barnett, 1980; Flynn, 1979) and problems in school (Hilberman & Munson, 1977-78; Hughes, 1982; Levine, 1975; NSPCC, 1977). A limited amount of empirical research exists about the children of battered women, although it has been steadily increasing in more recent years. For example, there were virtually no empirical studies published in the 1970's, but in the 1980's 21 studies were found (Fantuzzo & Lindquist, 1989). Unlike the anecdotal and clinical data, empirical findings tend to be inconsistent in terms of the types and extent of the children's difficulties (Christopoulos et al., 1987; Fantuzzo & Lindquist, 1989; Hughes, 1988; Rosenberg, 1987). When compared to the marital discord and divorce literature, the findings are somewhat congruent, in that externalizing or undercontrolling behaviors do occur among the children, particularly boys, in a good portion of the studies. On the other hand, evidence of internalizing problems was also reported in several of the studies. Inconsistent findings appear for the influence of age and sex on the children's reactions as well. These mixed findings are the result of several factors such as rigor of methodology, types of measurement, samples, and variables chosen for the study. A more in-depth review of the studies on child witnesses follows. 21 What do children observe when their parents engage in violent conflict? Using questionnaires and structured interviews with a sample of battered women and their children Rosenberg (1984) (cited in Rosenberg, 1987) assessed the types and frequency of verbal and physical aggression the children witnessed. Although one or both of the parents might have made attempts to shield the child from violence, nearly all saw and/or heard the incidents. Even when parents might have thought the children were asleep through most of the arguments and beatings, Rosenberg found that children reported they heard violence and remained in their bedroom because they were afraid to leave. The types of events children reported that they witnessed included: repeated verbal threats of injury; verbal assaults on their mother's character; objects hurled across the room; suicide attempts; beatings; threats with and actual use of a gun or knife; and homicide. Children may also be confused by the alternate eruptions of violence and loving phases between their parents depicted by Walker (1979) in her "cycle theory of violence." In a series of several well controlled studies, Wolfe, Jaffe, Wilson & Zak have consistently found behavioral adjustment problems among battered women's children. In this research group's first study to be described, Wolfe et al. (1985) compared 102 children from seven battered women shelters with a comparison group of children from 22 non-violent families solicited through newspaper ads. Both groups were independently rated on the level of violence in their families with the Conflict Tactics Scale (Straus, 1979). Children's behavior problems were rated by maternal reports through the Child Behavior Checklist (CBCL) (Achenbach & Edelbrock, 1983). Mother's level of stress and adjustment was assessed through three separate measure of sociodemographic, health and life experiences. Significantly more children from violent families fell into the high range on the Child Behavior Profile (CBP), with the association particularly strong for boys. More specifically, as Wolfe et al. report, "the estimated prevalence of the range of behavior problems defined as high (Sum T > 65) on this instrument is 42.2% for the total sample of children from the violent families compared with 19.8% for the nonviolent sample," (p.661). Compared to the normative sample estimates on this measure, only 7.5% of the sample would score this high. In addition, when the behavior problems were defined within the clinical range, 27 children (26.5%) from the violent families were included in contrast to ten children (10.4%) in the comparison sample. Sex differences were also assessed within the violent family sample. Boys' behavior problems were significantly more associated with violence in the home than girls' behavior problems. Also, there was a trend for boys from violent families to fall within the clinical range 23 on the CBP. Finally, hierarchical regression analyses were done to see what maternal stress and adjustment variables contributed to the children's behavior problems and social competence (e.g., maternal health, stressful life events and family crises). Factors associated with maternal stress and family violence variables accounted for 19% of the variance in child behavior problems. Wolfe et al. suggested that this finding implies that the impact on the child of witnessing wife battering may be partially a function of his/her mother's degree of impairment following such events as well as the disruption and uncertainty of the violence the child faces. In a similar study, Wolfe et al. (1986) compared two groups of battered women and their children with a non- violent community comparison group. The two groups of children with battered mothers consisted of current (N = 19) and past shelter residents (N = 23). Comparisons between these two groups showed that former residents' children's most recent exposure to violence was 6 months past, in contrast to 6 weeks for the children of the current residents. Comparisons of children's adjustment patterns revealed that the current resident group had the lowest reported social competence on the Child Behavior Profile (CBP), but they were not reported as showing significantly more behavior problems overall as compared to the two other 24 groups. Wolfe et al. advise caution in interpreting these findings because of the small samples and the elevated scores of the control group of children from nonviolent families (1 standard deviation above norms) which reduced the statistical significance of the results. However, trends were evident for higher internalizing and externalizing scores among the violent family sample. In the same study, comparisons of the mothers in each group indicated that the mothers currently staying at the shelter reported significantly more health problems, while the women who were former residents had a significantly higher disadvantage rating (i.e. changes in residence, changes in income, need for social services, etc.). It's interesting to note that the "former residents," while having a higher family disadvantage rating, did not report any significant behavioral or emotional problems among their children as compared to the non-violent community sample. This finding suggests that despite the past and present hardships of their living situation, children can recover from the impact of violence and separation, given that "the violence is eliminated and proper supports and opportunities for recovery are provided" (Wolfe et al., 1986, p. 102). In another study conducted by the same research group, Jaffe, Wolfe, Wilson & Zak (1986a) specifically examined the differences between girls' and boys' behavioral symptoms of a group of child witnesses of battering. The 58 child 25 witnesses were compared with 68 children from non-violent families who were solicited through newspaper ads. The children were from six to eleven years old, and were comparable on selected demographic variables. The Achenbach Child Behavior Profile (CBP) was used to measure behavioral adjustment. The Conflict Tactics Scale (Straus, 1979) was used to measure the amount of violence the children had been exposed to. For boys, a highly significant relationship for the child witness group was found between behavioral adjustment and level of family violence. However, female child witnesses were signficantly more likely to display internalizing and social competence problems when compared to girls in the non-violent sample. Boys significantly differed from the comparison group on all three factors-— externalizing, internalizing and social competence problems. These findings tend to support previous research that suggests boys are more vulnerable to parental discord than girls (Emery, 1982). In the final study of this research group to be summarized, Jaffe et al. (1986b) examined the similarities between 32 children who witnessed violence and 18 children who were victims of violence. A non-violent community comparison group was utilized. Only boys comprised this sample. The Achenbach CBP was used to measure behavioral adjustment, and the Conflict Tactics Scale was the measure for the extent of violence. The two groups of boys, those 26 who were child witnesses and those who were child abuse victims, differed significantly from the non-violent comparison group on the internalizing and externalizing scores. The abused boys demonstrated significantly more externalizing problems than those boys who only witnessed the violence in their homes. In addition, the degree of adjustment problems for both of the violent family groups was significantly different from the comparison group. That is, 90% (N = 16) of the abused children and 75% (N = 24) of the child witnesses had total behavior problem scores that were one standard deviation above the norm group mean for the CBP as compared to only 13% (N = 2) of the comparison group. These findings suggest that witnessing violence against their mothers may be as harmful to a child as physical abuse. However, the researchers had a strong suspicion that the children who witnessed the violence may have also been victims of it--even though they did not have direct evidence to validate their suspicions. Christopoulos, Cohn, Shaw, Joyce, Sullivan-Hanson, Kraft & Emery (1987) studied the adjustment of 40 child witnesses at the time of their shelter residence with a community comparison sample of 40 children from non-violent backgrounds. Multiple measures of family and child adjustment were used (Child Behavior Checklist (CBCL), Beck Depression Inventory for children, Perceived Competence, and WISC-R). Significant differences between the two groups 27 were found as measured by the CBCL. Specifically, the mothers reported more problems with their daughters on both the internalizing and externalizing scales. Overall, mothers reported more problems with both sons and daughters on the internalizing scale relative to mothers from the community sample. That is, when scores on the internalizing and externalizing scales were dichotomized into a clinical range group, and into a normal range group, significant differences between the groups were only found for the internalizing scale. Similar to the Wolfe et al. (1985) findings, Christopoulos et al. found that all the groups of children, except the girls from the community sample, obtained scores greater than one standard deviation above the mean of the normative sample for both internalizing and externalizing factors of the Child Behavior Checklist. However, the findings for sex differences were not congruent for these two research studies. Whereas Christopoulos et al. found signficantly more problems for girls, Wolfe et al. (1985) and Jaffe et al. (1986a, 1986b) found more problems for boys especially on the externalizing scale. Furthermore, Christopoulos et al. generally found higher maternal reports of internalizing behaviors, contrasting Wolfe et al. groups' findings and the marital discord literature (Emery, 1982). Several strengths and limitations of the Christopoulos et a1. study exist. Maternal reports of their children's 28 behavior were not the only measures used for child adjustment. However, although they obtained the children's reports of depression, social competence and intelligence, the CBCL was the only measure to yield significant differences between groups. Interesting to note was the finding that boys from the comparison sample scored one standard deviation above the mean for the norms. The participants' socio-economic status in the comparison group were similar to the women from the shelter, and thus the comparison women were also experiencing a number of family adversity factors that have been considered predictors of childhood maladjustment. The authors suggest that these family adversity factors, not just violence in the home, might also be contributing to childhood maladjustment. Related to family adversity factors, Hershorn & Rosenbaum (1985) compared the maternal reports of children's behaviors from a group of battered women with two comparison groups of women, one from non-violent but discordant marriages, and the other from non-violent, satisfied marriages. The mother's child—rearing practices and children's degree of exposure to violence were assessed in order to examine the ways in which they affect the occurrence of child behavior problems. Significant differences were found between the groups, with both the violent and non-violent discordant groups obtaining higher scores on the Quay & Peterson Behavior Problem Checklist for 29 conduct problems as compared to the satisfactorally married group. However, the violent and non—violent discordant groups did not differ significantly from each other. This suggests that "detrimental effects of parental marital violence may be attributable to marital discord inherent in such relationships" (p. 264, Hershorn & Rosenbaum, 1985). It was also found that increased exposure to discord/violence and punitive maternal parenting were both associated with children's problems. Punitive parenting style was only significantly related to conduct problems and not to personality problems. Finally, it was found that women from violent marriages did not differ in parenting style from either of the two comparison groups, refuting the hypothesis that abused wives tend to use harsher punishment and abuse with their children as a result of being physically abused themselves. The limitations of this study necessitate that these findings be viewed with caution since relatively small samples were used--on1y male children were examined, and the mothers from the violent and discordant marriage groups were clinically referred. In two studies of child witnesses conducted by Hughes & Barad (1983) and Hughes (1988) differences between age and sex in terms of the children's adjustment were examined. Hughes & Barad (1983) assessed a sample of 65 children, ages 3 through 13, who were residing in a shelter with their mothers for over three days. No control or comparison group 30 was used. The sample was divided into three subgroups based on age and school level: preschool, younger school age (grades 1 - 3), and older school age (grades 4 - 7). The overall functioning of the children was measured by maternal, shelter staff and teacher ratings on the Quay & Peterson (1975) Behavior Problem Checklist (BPC) and age appropriate self—report instruments measuring self-esteem and anxiety. Only the school-aged children were assessed for anxiety. Hughes & Barad concluded that their sample was similar in functioning to that of unselected samples of children, although there were several exceptions. The mothers tended to view their children's behavior more negatively than the shelter staff and teachers, based on a pattern of significant differences between the maternal reports and observers' reports. Sex differences in problems behaviors were only found for school age children. That is, boys received higher total scores on the BPC from the shelter staff and higher scores on aggressive behaviors from their mothers. In regard to age differences, preschoolers were significantly below average on a self-report self-esteem measure. There were several limitations to this study, (no comparison group; did not asssess for the severity of violence the children witnessed; sample was from a single shelter). This study raises the issue of why the mothers 31 rated their children's behavior more negatively than the other observers. Hughes & Barad suggest two explanations: the mothers' greater familiarity and sensitivity to their child's problems, and the tendency for mothers in distress to rate their child more harshly. In a further embellishment of the Hughes & Barad (1983) study, Hughes (1988) compared three groups of children on multiple measures of psychological and behavioral functioning. The three groups consisted of child witnesses who were also victims of parental abuse (abused-witness, N = 55), children who had only witnessed their mother's battering (witness, N = 40) and children who came from non- violent two-parent families (comparison, N = 83). The youngsters were divided into age subgroups: preschoolers (3 - 5 years), younger school age (grades 1 - 3), and older school age (grades 4 - 6). The mothers in the comparison group were matched on most demographic variables, although they had significantly higher incomes than the shelter group. It was not indicated whether income was used as a covariate in the analysis. Maternal reports of child behavior problems, and self reports of anxiety and self- esteem were utilized as measures. The "abused-witness" group had significantly more problems than the ”witness" or the comparison groups based on maternal reports from the :Eyberg Child Behavior Inventory (Eyberg & Ross, 1978). However, both the ”abused—witness" and "witness" groups had 32 higher scores than the comparison group on the anxiety measures. For self-esteem, the comparison group had significantly higher scores than the two other groups, but this was only for young school age children. For older children's self esteem, the comparison group still scored higher, but only as compared to the "abused-witness" group. Overall, Hughes' predictions that the "abused-witness” and "witness" groups would show more distress than the comparison group was partially supported. While the "abused-witness" group differed significantly from both groups on behavior problems, the "witness" group did not differ significantly from the comparison group on this dimension. The implication is that physically abused children who also witnessed their mother's abuse were less well-adjusted psychologically than those children who only witnessed violence but were not victims of it. Hughes' age predictions were also partially supported, with preschoolers having more behavioral problems but not having more self- reported anxiety than the other two age groups. Davis 8 Carlson (1987) found somewhat similar results in regard to age effects and type of violence experienced for a sample of shelter children. All of the children in their sample had witnessed the battering of their mothers, and half had been victims of parental abuse. Relying solely on the Achenbach Child Behavior Checklist for the assessment of the children's adjustment, they found that children who 33 both witnessed and experienced abuse directly had signficantly more problems on the Social Competence scale and internalizing dimensions than children who only witnessed the abuse of their mothers. Regarding age differences, preschool boys were found to have more behavioral problems (as in Hughes, 1988) with at least half of this age group falling within the clinical range of the aggression and depression scales. This finding is bolstered by previous studies that have found preschoolers to be more vulnerable to discord and abuse in the home (Hughes, 1988; Hughes & Barad, 1983; Wallerstein & Kelly, 1980). In addition to age effects, it was found that gender was an influence, with school age girls shown to score signficantly higher on aggressive items than school age boys. On the other hand, girls in the other age groups were found to have higher scores on more passive, depressed behaviors. Jouriles, Barling, & O'Leary (1987) specifically examined the effects of children being the victims of abuse as well as witnessing the violence inflicted upon their mothers. Their sample was comprised of 45 children who were referred for treatment by their mothers. Over 90% of the children had both witnessed the battering of their mothers and were themselves victims of battering by one of their parents. Interspousal aggression as measured by the Conflict Tactics Scale was highly associated with parental aggression towards children. Correlational data from parent 34 reports on the Revised Behavior Problem Checklist (Quay & Peterson, 1983) indicated that parent-child aggression was more strongly linked to child behavior problems than parent reports of interspousal aggression. Jouriles et al. suggested that in families where women are battered, the witnessing of the violence may not be as important in influencing child behavior as the occurrence of parent-child aggression. This contradicts Jaffe et al.'s (1986b) conclusions from their study that witnessing abuse can be just as harmful as being a victim of it. However, Jouriles et al.'s findings must be considered with extreme caution due to a small sample size, no comparison group, and lack of independent judges rating family aggression and child problems. Since the mothers were the ones who referred their children to treatment, and then were also rating their children's exposure to violence and existence of their behavior problems, the correlations between the two variables may be artificially inflated. Wally—View The studies summarized thus far have all examined the effects of witnessing and/or being a victim of violence on preschool and school age children between the ages of 3 and 16. Although it is controversial, there is some evidence that one of the long term effects of exposure to family ‘violence may be the intergenerational transmission of such violence (Gelles & Straus, 1988; Owens & Straus, 1975; Post, 35 Willett, Franks, House, & Bach, 1981; Straus, 1983). With adult samples, two studies examined the long-term effects of living in a violent home. Rosenbaum & O'Leary (1981) compared three groups of women with children--those who were victims of battering (N = 52), those who were not abused but had discordant marriages (N = 20), and a group of women in non-violent, non-discordant marriages randomly selected over the telephone (N = 20). Women were asked to report on the extent of violence in the families of origin for both themselves and their spouses (e.g. whether they say their fathers abused their mothers; whether they had been physically abused). Also, mothers were asked to rate their own child's behavior on the Quay & Peterson Behavior Problem Checklist. Only male children closest to the age of ten comprised the sample. Maternal reports of child behavior yielded no significant differences between the three groups, although there was a trend for the youngsters of abused women to have more problems. However, it was found that abusive husbands were more likely to come from families of marital violence as compared to the other two groups (discordant, and nondiscordant, nonviolent samples). In total, 70% of the abusive husbands came from violent family backgrounds, where either wife battering, child abuse, or jboth existed. Interesting to note, abused women were no inore likely to have witnessed abuse than were either of the lTOD'BbUSEd groups. Caution should be used in interpreting 36 these results since the data were obtained solely through retrospective accounts. In a second study on long-term effects of witnessing abuse, Forsstrom-Cohen & Rosenbaum (1985) conducted an exploratory study that assessed the behavioral and emotional effects of viewing wife battering on a college age sample. Two comparison groups were used: those who viewed discord but not violence (N = 43); those who viewed satisfied, happy parents with no evidence of violence or discord (N = 77). The viewed violence group consisted of 44 undergraduates at a university. All were white, middle to upper-middle class. Anxiety, depression and aggression were assessed by paper and pencil tests given at one time. The viewed violence group and the viewed discord group were significantly more anxious than the group who came from non-violent, non- discordant parents. Males and females were differentially affected by witnessing abuse. Women who had viewed interspousal aggression were signficantly more depressed and more aggressive than the women in the other two groups. These women also reported more depression than the males who had viewed interspousal aggression. These findings suggest that the effects of witnessing abuse in the home extend into adulthood and that young women are more adversely affected than men. The finding that women are just as aggressive as men is somewhat surprising based on past findings, although Davis & Carlson (1987) found school-age female witnesses to 37 score higher on aggression than school-age male witnesses in their study. Support for this finding also exists in the divorce literature, where Hetherington (1972) found that girls who experienced their parents' divorce exhibited behavioral problems and negative reactions to men when they were in older adolescence. WWW To date, little empirical research exists on children who witness the battering of their mothers. A total of 12 published studies were found on this topic. Numerous anecdotal and clinical reports of the effects of witnessing wife abuse on children's adjustment describe such behavior problems as acute distress, depression, withdrawal, aggression, problems at school, and somatic complaints. Since little research has been done in this area, the literature on marital discord and divorce and its effects on children has been extrapolated to include children who live in violent homes. Children with parents who are divorcing or who have discordant marriages tend to be rated higher by one of their parents (usually their mother) on problem behavior measures. Boys in particular seem to have more problems with externalizing behaviors. Similar results have been found in some of the studies on child witnesses of battering. However, the overall findings on children's exposure to woman battering are not conclusive, since several methodological limitations were evident in the 38 research reviewed. These findings point to the necessity for further research in this area. Many questions remain unanswered. For example, although it appears that the children have difficulties adjusting to witnessing violence, it is not clear what specific factors contribute to their maladjustment, and whether their problems are time-limited and tend to fade after the crisis of being in the shelter. Most of the research used samples of children who were residents in a shelter; none of the studies employed follow- ups of the women and their children. Other factors such as gender, age, extent and type of violence witnessed, maternal stress, child abuse, disciplinary practices and family adversity have been implicated in the literature as tending to moderate or compound the effects of witnessing battering. However, it is not known which of these factors are the most salient for contributing to the maladjustment of the children. Therefore, the question remains as to whether witnessing the battering of one's mother in itself creates behavioral and adjustment problems in the child. In addition, virtually no empirical research has examined the child's feelings about the assailant. This relationship may also be a crucial factor in explaining the child's adjustment. Finally, children's experience should be considered within the context of the battered woman's struggles to 39 escape the violence. As described earlier, although battered women try to leave the relationship by engaging in help-seeking behaviors, they are thwarted by inadequacies in and/or lack of response from, community supports and services. The end result is that the woman and her children may continue to be harrassed and abused by the violent assailant. In only one study, conducted by the Wolfe et al. (1985) research group, were such factors as stressful life events and maternal stress examined as variables that could moderate the children's behavioral adjustment. Limited income, poor living conditions, multiple moves, separations and maternal stress along with the level of violence were examined. This study found that maternal stress and violence variables contributed significantly to the children's adjustment. This limited, but compelling result helps to emphasize battered women's need for community supports and services, since not only does violence and lack of community resources to escape the violence potentially harm them, but it may also potentially harm their children. It was the current study's purpose to investigate these issues by identifying the influence, if any, of some of these factors on the child's adjustment after the mother and children have left the shelter and have received an intervention to help them remain independent of their assailant. Before a more thorough description is given of what the current study investigated, a more in-depth 40 description and discussion of the potential influence of the various factors purported to influence adjustment of battered women's children will be given. WW Gendg1‘_ In seven of the studies that looked at gender differences, boys had higher negative adjustment scores than girls. The problematic behaviors were usually the externalizing, undercontrol or conduct type. This finding is congruent with the marital discord and divorce literature (see Emery, 1982) and literature studying the effects of stress on children. For instance, Rutter (1970) found that boys are more vulnerable to physical stressors and hazards than girls and he suggests that this predisposition may also apply to psychosocial stressors (Rutter, 1979). However, base rate data on childhood disorders in the United States show that boys are predisposed toward undercontrol whereas girls are predisposed to overcontrol (Ross, 1980), suggesting that girls are not immune to psychosocial disorders. In the remainder of the studies (N=3) reviewed that assessed gender differences, girls were found to be just as affected as the boys, if not more so. In a well controlled study by Christopoulos et al. (1987) girls were found to have more problems with adjustment, and with both types of broad band behaviors, externalizing and internalizing. Davis and Carlson (1987) also found that school-age girls who witnessed their mothers' abuse scored 41 higher on aggression than did the school age boys. Finally, Forsstrom-Cohen and Rosenbaum (1985), using a college sample, found that the women who witnessed violence between their parents when they were growing up, were more aggressive than two comparison groups of women--those who only experience discord between their parents, and those who came from happily married parents. Several explanations drawn from the Emery (1982) review articles on marital discord and divorce might help to explain the gender differences in the behavioral ratings: 1) The boys tend to model their father's physical and aggressive behaviors toward their mother. This might help to explain the undercontrolled behaviors in the boys and the overcontrolled behaviors in the girls. 2) The children are responding according to their prescribed sex roles. 3) Inconsistent disciplinary practices tend to be used between parents who have marital discord, and this has been related to problems of conduct and aggression in children. Also, parents are more involved in disciplining the same sex child, so if marital turmoil causes fathers to alter their discipline more radically than the mothers, a greater effect would be seen on the sons. In fact, research indicates that sons are disciplined about equally by both parents, whereas daughters are disciplined more by their mothers. 4) As indicated above, how each sex handles stress may account for the differences. It may be that girls are less susceptible than boys to psychological stressors because of their greater resistance to physical stressors. 5) Finally, it may be that the girls are just as affected as the boys, but it is not being detected with the present measures, and it may be that overcontrolling behaviors are more tolerated by parents, and not reported on the measures. 42 Strong evidence does not exist to make definitive conclusions in this area, but as Emery (1982) stated, we should not be asking the question of whether or not problems exist for girls, but rather, to what extent do problems exist for them. Age, Only three studies specifically examined age differences in their studies. In all three studies, negative adjustment problems were found across all age groups in the samples. Age groups typically were divided into preschoolers (3-5 years), young school age (6-8 years), and older school age children (9—13 years). Several researchers have argued that preschoolers tend to be more vulnerable to stressful situations (Davis & Carlson, 1987; Hughes, 1988; Hughes & Barad, 1983) than older age groups. However, the results are inconsistent and too meager to conclude vulnerability for specific age subgroups. A similar pattern of inconsistent findings was also found for the effects of marital discord and divorce on children (Emery, 1982). It may not be that just one age group is particularly vulnerable to the violence and the discord between parents, but that all age groups are, and they respond in different ways according to their particular developmental capacities. Response differences to marital discord and inter-spousal anger according to age were observed by several researchers (Cummings, Zahn-Waxler & Radke-Yarrow, 1981, 1984; Wallerstein & Kelly, 1980). For 43 example, Cummings et al. (1984) reported that children's patterns of responding to anger between their parents changed over a five year period. They observed a sample of twenty-four children, at two time periods, for nine months when they were toddlers, and again for three months when they were six and seven years old. Toddlers were much more overt in their emotions in reponse to anger (e.g. crying, smiling, laughing, excitement, and angry yelling). By school-age, however, the same children were less likely to express emotions in anger situations, and exhibited planned strategies for dealing with the conflict. Wallerstein & Kelly (1980) observed similar patterns, and found that although children of different age groups responded differently, they were no less affected by the marital discord they experienced between their divorcing parents. 1hg_gn11g_§g_yigtim_gf_§pn§g‘_ Researchers have suggested that children who are direct or indirect victims of battering may overlap by as much as 40% to 70% (Bowker, Arbitell & McFerron, 1988; Layzer, Goodson, & DeLange, 1986; Straus, Gelles, & Steinmetz, 1980; Walker, 1984). Physical, emotional and sexual abuse have all been reported. Differing observations exist about who predominantly abuses the child in such relationships. Layzer, Goodson, and DeLange (1986) found that the mother's batterer was usually the abuser of her children, although in about one-quarter of the cases, the mother joined in the battering of the child. 44 Straus et al. (1980) found that if the mother has been hit, she is twice as likely to hit her child as compared to a mother who has not been hit by her husband. There is a substantial amount of evidence that physical abuse of children has a detrimental impact on their psychosocial adjustment. Lamphear (1985) reviewed the literature in this area, and reported that although several methodological flaws in the designs of most of the studies were evident, recurrent patterns of negative adjustment emerged, such as a greater number and frequency of behavior problems, poor peer relationships, social skill deficits and poor school performance. Three studies that were reviewed specifically examined the effects of being a victim versus witnessing abuse on the child (Hughes, 1988; Jaffe et al. 1986; Jouriles, Barling & O'Leary, 1987). The children who were abused had similar adjustment problems (and sometimes more severe) to those who witnessed the battering of their mothers. Wolfe & Mosk's (1983) study comparing a group of abused children with two comparison groups of children-~those from distressed families with no abuse present and those who reported no problems--supports these findings. Both the abused children and the children from the distressed families were clients from a child welfare agency. Assessments from the Achenbach Child Behavior Profile revealed that the abused children and those from the non-abused distressed families displayed a 45 significantly greater number of behavior problems and fewer social competencies than the non-distressed, non-abused controls. The abused children's scores resembled the wide range of behavior problems displayed by children from maritally distressed families. From these findings, Wolfe and Mosk (1983) suggested that the negative adjustment of the children may be more a function of family events and interaction patterns than isolated abusive episodes. This suggestion is reinforced by the findings that single isolated chronic stressors carry no appreciable psychiatric risk on children, but when any two stressors occur together, the risk goes up, and the existence of one stress can actually increase the likelihood of the occurrence of others (Rutter, 1979). Related to this, several researchers have argued that multiple stressors exist in homes where there is discord and violence such as multiple separations and moves, poor living conditions, and lower economic status. These stressors along with the witnessing of the violence and the probability of being a victim of it over long periods of time can all contribute to the children's negative adjustment patterns (Wolfe, Jaffe, Wilson, & Zak, 1985). So few studies controlled for the presence of child abuse that definitive conclusions about whether it is preyalent for children who have a battered mother and what contribution it makes to children's problems cannot be made. 46 Yet the child abuse literature and the few studies done on the topic of battered women's children who are also victims of abuse seem to support the view that child abuse interacts with the effects of witnessing violence, and is a factor that should be examined. WW ' v r Six out of twelve of the studies reviewed used Achenbach & Edelbrock's Child Behavior Checklist (CBCL) (1983) to assess child behavior problems and social competency. Four of the studies relied on the Quay & Peterson Behavior Problem Checklist (BPC) (1983). Both measures have demonstrated good reliability and validity (Achenbach, 1983; Quay & Peterson, 1983). However, these measures rely on parental reports, usually the mother's, of children's behaviors. Relying solely on maternal reports of the child's behavior, which the majority of these studies have, is limited and may introduce bias into the data. Reid, Kavanagh, & Baldwin (1987) compared the behavior of children in twenty-one child abusive and twenty-one non- abusive matched comparison families using home observations by independent observers and parental report measures. The parental report of child conduct problems differed significantly between the two groups, but independent observations showed few significant differences in either the child or parental behaviors. These data tend to show 47 that parents in child-abusive families overestimate the level of conduct disordered behavior engaged in by their children compared to independent observations in the field and in lab settings (see Mash, Johnston, 5 Konitz, 1983). Also, Hughes & Barad (1983) found that battered mothers in a shelter tended to rate their childrens' behavior more negatively than shelter staff and teachers of the school age children. The potential for negative or biased reporting by the battered mothers might exist for several reasons. Some mothers may have a low tolerance for essentially normal child behaviors because of constitutional factors, environmental stress or depression (Reid et al., 1987; Wolfe, Jaffe, Wilson & Zak, 1985). For example, Friedlander & Weiss (1986) found that maternal depression was found to be signficantly associated with mother's rating of her child on the Achenbach CBCL. However, the mother's ratings were still able to differentiate groups of children with and without psychiatric problems. Estroff, Herrera, Gaines, Shaffer, Gould & Green (1984) found similar results using the CBCL. In addition, mothers may over-report in order to justify treatment for their children. (This may be the case for the clinic samples used some of the studies.) On the other hand, the women may under-report depending upon their level of reluctance to share such information (Hughes, 48 1988). Finally, the mothers' reports may seem elevated only because they have a greater familiarity with and sensitivity to their children's behavior (Hughes & Barad, 1983). Suggestions for multiple informants have been made in order to supplement and cross-validate the parental reports (e.g. teachers, peers) (Christopoulos et al., 1987; Hughes, 1988; Reid et al., 1987; Wolfe et al., 1985). Obtaining child self-reports on various psychological measures has been one improvement made in two of the studies reviewed (Christopoulos et al., 1987; Hughes, 1988). However, this can be problematic too, since younger children's self- reports tend to be less reliable. The ability of young children to reflect on their feelings and subjective experiences has been questioned in relation to their level of cognitive development, attention span, memory capacity, and language skills (Hughes, 1988). In addition to the age of the child in terms of validity and reliability of reports, there also may be discrepancies between the child's versus maternal reports of the child's behavior and adjustment. In one study, diagnoses based on children's reports indicated depression, whereas mothers' reports on their children differed. Instead of indicating depression, the mothers reported high rates of oppositional and attention problems (Kashani, Orvaschel, Burk, & Reid, 1985). Whether these differences between the maternal and child reports were due to the children's inability to describe 49 their subjective states, or the mothers' inability to fully describe their children's adjustment, or due to yet other factors, is up for question. Wrongs... The majority of the studies used control or comparison groups in their designs (N=8). Four studies used multiple measures or multiple informants of the child's adjustment, either through child self-reports or through other observers such as teachers and shelter staff. Most of the research sample populations were obtained from shelters for battered women (N=8). The remaining four studies used clinic samples, and a college population. s o f w-u . Finally, no follow-ups of the women and children were carried out in any of the research reviewed here. However, in one study that was based on an evaluation of demonstration projects in selected shelters around the country, children were followed up after leaving the shelter (Layzer, Goodson & Delange, 1986). Those who were abused (N=44) were less likely to be abused because they were not living with the mother's batterer anymore. Over half of the children had no further contact with the batterer. More extensive follow-up information would be particularly valuable in order to gain more descriptive information about the children's situation and to know whether the children's adjustment changes over time. 50 ,* ,-A, $11.‘ 1 1, g 0 ’Q - 0; 00‘7" - g 1 o 1 Similar mechanisms discussed by Emery (1982) that help to explain children's adjustment to marital discord and divorce may be extrapolated to the adjustment of battered women's children. What follows is a brief discussion of some of the theories that contribute to the understanding of what influences children's adjustment. Attaghmgnt_§ng_§gp§;§;19gL, This hypothesis is based on the work of Bowlby's attachment theory that the loss of a figure to whom one is attached produces fear and anxiety (Bowlby, 1969). It particularly focuses on maternal deprivation because the mother tends to be the primary attachment figure. However, in the case of divorce, or with battered women who leave their batterer, the loss is more likely to be paternal deprivation. The effects of paternal deprivation on children of divorce are unclear, and further, no work has been done on children's reponses to leaving the batterer who may be a father. Also, as Emery argues, the attachment hypothesis predicts that the most detrimental effects occur in the child's early years, yet no conclusive age effects have been found in the marital discord or battering literature. It might not be separation per se, but the exposure to the conflict that is actually the factor in creating the behavior problems. WWW Mar ital turmoil may interfere with the child's imitation of the same-sex 51 parent; it may lead to rejection of both parents as models, or the child may imitate the negative and aggressive behaviors that occur between discordant couples. As demonstrated by Bandura (1973), children are likely to imitate aggressive behaviors. This theory may particularly hold for the children of battered women who witness their father or their mother's boyfriend attack their mothers. Modeling of same-sex behaviors may also help to explain the apparent gender differences in behavioral problems found in the literature. Some evidence exists that show children are more likely to imitate a same-sex model (Bandura, 1969; Margolin & Patterson, 1975). Hughes (1981) argued that a very powerful violent father exerts much influence upon his children's learning. The son in particular observes and is reinforced by him, learning how males behave and what characteristics are associated with being female. The boys quickly learn that violence is a way of obtaining what they need, and that physical aggression is part of being male. Girls, on the other hand, see their mothers as being in a subordinate role, attempting to get their material and emotional needs met and not being very successful. Child advocates have observed that school-age boys in particular have shown some discomfort and ambivalence toward their fathers because of their sex-role identification. Within the shelter, the boys feel their fathers are not esteemed and are uncomfortable about missing them or wanting 52 to see them. This ambivalence and discomfort the boys feel may interfere with their adjustment at the shelter and may result in behavioral difficulties (Hughes, 1983). Separation from the father-figure when the battered mother leaves him may also cause more problems for boys because the male model is missing. This tends to be supported by the finding that children adjusted better when they were in custody of the same-sex parent (Santrock & Warshak, 1979) and that contact with adults other than the custodial parent is beneficial to the child's development (Jacobson, 1978). Discipling_p;§gtigggL Emery (1982, p 321) suggests: "it is possible that marital turmoil leads to a change in the use of important discipline techniques, to the detriment of the child, or it may lead to increased inconsistency in discipline either between parents or in the practices of a single parent." Arguments between parents about discipline may occur in front of a child, and may result in inconsistent discipline. Problems of conduct and aggression have been related to inconsistent disciplinary practices. Although parents have the tendency to discipline the same sex child more, it has also been observed that male children may be disproportionately disciplined (Margolin & Patterson, 1975). This could lead to more conflicts about how to _ discipline the male child rather than the female, which could help to explain the gender differences found in behavioral problems. 53 Very little research exists on whether disciplinary practices change before the onset of separation and divorce or whether it changes afterwards. However, Hetherington, Cox 5 Cox (1976), through a rigorous research design, documented changes in postdivorce discipline. Divorced parents were found to be more inconsistent with their children as compared to parents from intact marriages. Again, boys seemed to be more affected, exhibiting less compliant behaviors. In terms of battered women, Hughes (1981, 1982) observed that women who came to the shelter tended to utilize inappropriate control with their children (either too little or too much). Hershorn & Rosenbaum (1985) found that increased exposure to discord and battering and punitive maternal parenting were both associated with children's problems. However, battered women did not differ in parenting style from either women in discordant marriages or women in non-discordant marriages; that is, the battered women were no less punitive with their children than either of the groups. W The major 1ty of battered women would rather leave their batterer than stay in such a relationship (e.g., Snyder & Fruchtman, 1981; Snyder & Scheer, 1981). However, despite their desire to leave, the majority return to living with the batterer. In fact, many of the women leave and return to their assailant repeatedly, with some studies reporting as many as six to 54 ten times (Dobash & Dobash, 1979; Labell, 1979). Unfortunately, community services that could help women leave violent relationships are often unresponsive or unavailable and thus may be a significant barrier to their leaving (Gondolf, 1988, Sullivan, 1988). For example, Gondolf (1988) found that economic dependence as defined by personal income, transportation and childcare, significantly predicted whether a woman would separate from her batterer once she left the shelter. This desire to leave and the deficiencies in community support available to the women may create undue stress on her and her children (Wolfe, Jaffe, Wilson, & Zak, 1985). For example, Walker (1979) found that battered women's maternal effectiveness may be impaired because of the stress she faced from the battering situation. Hughes (1983) observed that some abused women in shelters could be extremely depressed and/or anxious because of the trauma of leaving their husbands. These researchers have observed that the mother's functioning in turn could affect the children's psychological functioning. This was supported by Wolfe et al.'s, (1985) finding that maternal stress variables accounted for 19% of the variance in predicting behavior problems of children who witness battering of their mothers. Support from other literature on children, maternal functioning and the environment exist. Rutter (1979) 55 reported that the wider social environment can have an effect on parenting. For instance, working-class women in inner-city areas had higher levels of maternal depression and marital discord (Rutter & Quinton, 1977). Children of depressed mothers are at greater risk for maladjustment as are offspring of schizophrenic parents (Beardslee, Bemporad, Keller, & Klerman, 1983; Cohler, Gallant, Grunebaum & Kaufman, 1983). Conger, McCarty, Yang, Lahey, & Kropp (1984) found that chronic stress and maternal psychological functioning were good predictors of positive and negative parent-child interactions. These stress variables may result in reducing the amount, frequency and type of attention the child receives from his/her mother. For example, inconsistent, understimulating and inappropriate parental attention tends to exist more in violent families than in non-violent ones and this type of attention may result in child behavioral problems (Patterson, 1982; Walker, 1983; Wolfe, 1985). These factors that are postulated to create stress and malfunctioning in a battered mother can also serve to create stress directly on the child. Changes in neighborhood, schools, housing, and access to material resources with perhaps a reduced or different social network can be stressful to children. However, the existence of a warm, close relationship with one parent can mitigate some of the negative effects of these stressors, including marital 56 discord (Emery, 1982; Longfellow, 1979; Rutter, 1979). Wasatch As discussed above, few empirical studies on children of battered women exist. Also, there are several deficiencies in the research design and measurement of these studies. Therefore, further description of the characteristics of the children who witness battering of their mothers is needed in order to advance the knowledge about this social problem so that resources may be mobilized in order to aid these women and their children. More specifically, further research is needed in describing the adjustment of the children after their stay at a shelter and after separation from the mother's, and perhaps child's, assailant. Documentation of the changes in the child's living situation, such as the variables described above, and how this may affect the children are needed. Also, in light of the modeling hypotheses and gender differences that have been hypothesized to moderate the effects of the violence on the children, more description of the children's feelings about the assailant's absence would be valuable for explaining the children's adjustment. Furthermore, examination of whether the children are victims of abuse in such living situations and whether they continue to be abused after leaving the assailant would be valuable information for both treatment and theory. 57 The Current Resgarch The current research was designed to examine the effects on battered women's children of an advocacy intervention designed to help women live their lives independent of their batterer. This research is valuable for several reasons. To date, there has been no research examining the effects of an intervention designed for battered women on their children. This research is unique in that it will attempt to manipulate some of the variables theorized to contribute to the stress of battered mothers that may affect the children's adjustment--namely, access to community resources that may help to maintain the women's independence. Secondly, the research was designed to address some of the factors described above that have been reported to influence children's adjustment. Namely, it will provide information about children's adjustment after leaving a shelter, the type and extent of violence the children experience over time, (including child abuse), children's contact with and feelings about the assailant's absence, children's immediate reactions to witnessing assaults, mothers' psychological well-being and experiences of stress, and mothers' disciplinary tactics. W The current research is part of a larger project that is being evaluated on its success in helping women access 58 the community resources they need to help them escape violence (Sullivan, 1988). The participants were women who had just left a shelter for women with abusive partners. Cris Sullivan designed the program specifically to fulfill the unmet needs of these women through paraprofessional advocates who identify, mobilize and generate community resources to address their needs. The advocates not only work to gain access to the needed resources for these women, but they also teach the participants those skills necessary for becoming their own self-advocates. WW The following are the specific hypotheses and questions that were addressed in this research. First, predictions were made that assessed whether the intervention affected the women and children in the project on various outcome measures. 1. An important goal of the Community Advocacy Project is to provide project participants with trained community advocates who will have the effect of helping the women access the resources they desire. Paraprofessional advocates have been found to be highly successful advocates for various populations including women with abusive partners (Davidson et a1, 1987; Donato & Bowker, 1984). Therefore, the first hypothesis of this study was: the W W 59 W 2. The literature suggests one of the significant barriers to women leaving their assailants is lack of community resources and the unresponsiveness of the community service providers to battered women's needs (Gondolf, 1988; Hofeller, 1982; Strube & Barbour, 1983). Since this project was designed to make the community more responsive to women's needs by providing the women with community advocates to help them get resources it was hypothesized that: Wages; --v: .on w .l -- o_- _ i to .~ - 1 _ -‘--1q,; 0 .;~1 v., ; .; . -ar~- . .- .., - h: ., . 91222; 3. In addition to the physical effects of battering, women may also experience serious psychological effects, such as stress, depression and anxiety (Hughes, 1983; Walker, 1979; Wolfe et al. 1985). In addition to the stress of the violence, environmental adversities such as lack of economic resources and multiple separations and moves may also create undue stress on the women (Wolfe, et al., 1985). Women with abusive partners may also lack supportive social networks to help them (Straus, 1980). In providing the women with the resources they need to help them escape abuse, and also the social support the community advocates could give to the women on a weekly basis a third hypothesis was that: WW1: o l w -b n e red b ov r l ife o o n t d depgession) than those women in the control condition. 4. As stated above, battered women may experience stress, depression and anxiety as a result of battering and other concomitant factors such as lack of resources to escape the abuse. Several researchers have postulated that battered women's maternal effectiveness may be impaired as a result of the abuse and stress they have experienced (Hughes, 1982; Walker, 1979; Wolfe et al (1985)). Wolfe et al. (1985) found that maternal stress variables predicted child behavior problems of children who witnessed battering. The Comunity Advocacy Project provides a means for women to access their desired community resources, which may enhance their psychological well-being. This may in turn enhance their maternal effectiveness. In addition, certain resources that directly pertain to the women's children (i.e., childcare, other issues for children) may be accessed that could positively affect their children. Thus, a fourth prediction was made that: Ibs_2hildren_in_ths_eansrimental .,. ., ._ , ..; - , ~_ - ., n ,- - .- .-"9-- -_-. ~19 +1" ,1- i ll 1’! 1 1' .0 7' The following research questions were exploratory and may help in understanding what relates to, predicts or explains the behavioral adjustment of the children in the 61 two groups over time: 5. Will there be any significant differences between conditions on the amount of abuse children experience over time? 6. For women in both groups, will the type and amount of resources obtained relate to their children's adjustment at post-intervention? 7. For women in both groups, will remaining free from their batterer improve their children's adjustment at post- intervention? 8. For women in both groups, will good psychological well-being positively relate to children's adjustment at post-intervention? Current research shows that boys from maritally discordant and violent families tend to have more behavioral problems than girls of these families. The behavioral problems tend to be of the externalizing or undercontrol type. Modeling theory also supports this gender difference finding. Therefore, another prediction was: 9. Although girls will show behavioral adjustment problems, boys will have more externalizing problems. Several factors have been hypothesized to contribute to the behavioral adjustment of children who witness violence. Since there is very little research in this area, research questions were generated that explored the relationship of some of these factors to the behavioral adjustment of the 62 children. For example, there is evidence to show that child abuse tends to exist in a large percentage of families where woman battering occurs. This abuse may contribute to the problematic adjustment of some of the children who have also witnessed their mothers being battered. Therefore, the research questions were: 10. What will be the extent of child abuse among the participants in this study? Will it differ between the two groups as a result of the intervention? Will child abuse be related to the children's adjustment? The disciplinary practices of parents from discordant and violent relationships may be punitive and/or inconsistent. This has been theorized to contribute to child adjustment problems. Thus another research question was: 11. What type of disciplinary practices do the research participants of this study use and does it relate to the child's adjustment? The child's feelings about the batterer is another area for exploration. If the child particularly values her/his relationship with the mother's asssailant s/he might imitate the the batterer's aggressive behaviors. Bandura's (1973) modeling theory supports this. Shelter workers often report aggressive behaviors from the male children. Also, Hughes (1982, 1983) observed that male children tend to have ambivalent feelings about their fathers, feeling anger 63 towards them on the one hand because of the abuse, but also missing them after their separation. However, the batterer might not be the father of the child under study and/or might not have much contact with the child. What relation the batterer is to the child, the type and amount of contact the child has had with the batterer in the past and present, and the feelings the child has about the batterer's absence will be variables that may help to illuminate the nature of the child's relationship to this person. Therefore, several research questions exploring this area were: 12. How much contact does the child have with the batterer, both at pre-intervention and post-intervention? Is the level of contact related to the child's behavioral adjustment? Does the amount of contact differ between the two research conditions? 13. Based on the mother's perceptions, how does the child currently feel about the batterer being gone? Are the child's feelings about the batterer related to the child's behavioral adjustment? Finally, information about how children immediately responded to witnessing the battering of their mother would also be helpful for descriptive purposes. Therefore, a final research question was: 14. What were children's immediate responses to the battering, and do these responses relate to adjustment? CHAPTER TWO METHOD 8.955.139. The research intervention, the Community Advocacy Project, was conducted in the Greater Lansing Area of Ingham County, Michigan and involved the cooperation of two existing institutions. The first was Michigan State University, under whose auspices this project was implemented. The structure for this program was already in place at MSU, as a section of Psychology 371 and 372: Community Projects. Paraprofessional advocates were recruited from the undergraduate population. They earned college credits for their participation. The Council Against Domestic Assault (CADA) was also involved in this research, as the referral shelter for research participants. CADA is the local battered women's shelter, and houses approximately 30 women and children each month. It is the policy of the shelter to allow women to remain up to 30 days, resulting in their housing approximately 350 women (and their 500 children) a year. The critical individuals at CADA agreed to participate in this program by allowing research staff to recruit shelter residents. An administrative agreement between the Council's director and the principal investigator outlining each party's roles and responsibilities has been negotiated in writing. See Appendix A for the administrative agreement 64 65 with CADA. Re ar h Part c t Went All shelter residents were told about the availability of the Community Advocacy Project by a project staffperson at CADA's twice-weekly house meetings. Interested women were interviewed and given a more thorough description of the program. Each woman was told that if she were randomly assigned, she would receive the services of a trained advocate for 4-6 hours a week for a period of 10 weeks. The woman was told that her advocate, should she receive one, would be a female student volunteer from Michigan State University trained in accessing community resources. It was also explained that not every woman would receive the services of an advocate, but that all interested women would be interviewed before the program as well as 4 times afterwards (lo-weeks, 2-months, 6-months, and 12- months). All of the interviewees were informed that they would be paid for each of the interviews. These payments increased in dollar amount in order to encourage the women to stay in contact with the project until all of the interviews would be completed. The first interview was worth $10 ($15 for child interview), the second- 340 ($45 for child interview), the third- $60, and the fourth- $80. The interested women were randomly assigned to either the experimental or control condition. After the completion of 66 the first interview, the woman was informed of whether she was randomly assigned to the experimental or the control group. This occurred at the shelter or as soon as the woman moved out. The (CAP) staffperson opened a sealed envelope containing a card that stated whether she received the intervention services or not. The women in the control group (who received no additional services) were not contacted again until their subsequent interview (approximately 10 weeks later). They received business cards indicating the date of their next interview and how much money they were to be paid. They were asked to call the project office if they were not contacted by the date on the card. Women in the experimental condition became Community Advocacy Project participants and had an advocate assigned to them within a week after leaving the shelter. See Appendix B for the participant agreement forms. M r t t During the eight months of subject recruitment for this study, 192 women were residents of CADA. Of the 192, contact was made with 177 women. The other 15 women left CADA without filling out an exit form or telling anyone where they were going. Of the 177 women who were contacted at the shelter, 88 were not included in the project for the following reasons: forty-two women left the area, 15 could not be located for their interview, 18 women were in the 67 project before, one woman spoke no English, 3 women had to go to jail or prison after they left the shelter, and 9 women were at the shelter for reasons other than domestic violence. Of the remaining 89 women eligible for the program, 79 expressed an interest in participating, and were interviewed for the program. Ten women were not interested in participating in the project. Eligibility egiterie fer the egzgegt reeeegeh, Mothers considered for the current research were chosen from the pool of eligible applicants (79) described above that were recruited for participation in the Community Advocacy Project. The specific eligibility criteria for the current research was: 1) mothers had to have a child who was between the ages of 2 and 16; 2) if the child was not living with them, then they had to have had contact with their children at least 4 to 5 times per month in the 10 weeks prior to the interview and expected to have 4 to 5 contacts per month in the next 10 weeks; 3) if a mother had more than one child who was eligible for the study, then only one child was randomly selected. Twenty-five of the 79 eligible women for CAP were not eligible for the current research for the following reasons: five mothers had children under the age of two and seven had children over the age of 16, four mothers had no contact with their children, one mother had no custody, seven women had no children, and in one case the mother's children did 68 not know the assailant and had no exposure to abuse. Therefore, out of the total of 79 women eligible women for CAP, 54 mothers were considered eligible for the child component of the project. Two of these women dropped out of the program before receiving three weeks of service. In addition, post-interviews could not be completed for three women who either dropped out of the program after 3 weeks of services, or could not be found for their second interview. Therefore, most analyses were based on data from 49 mothers and their children (24 experimental and 25 controls). t a e. In order to qualify as a research participant, women had to receive at least three weeks of services. As stated above, two women ended their involvement before this time period and were not included in the analysis. One of these women was murdered by her boyfriend, and the other woman could not be contacted despite extensive efforts made by the project staff. Furthermore, three women who did receive three weeks of services were dropped from most of the analyses because they did not complete their post-interview. Of these women, one decided to terminate her involvement with the project, and the other two left the greater Lansing area and could not be reached. Therefore, when these three women were included in the drop-out rate, the total percentage was 9%--fairly low for such a transient population. WW. Forty-eight percent of 69 the mothers participating in this study were African American, 40% were white, 8% were Hispanic, and 4% were designated in the "other" category. Ages ranged from 18 to 38, with the mean falling at 27. The mean number of children living with the mothers was 2, and 54% had children under the age of five. Ninety-six percent of the mothers had some type of custody of their children (i.e., full, joint). (See Table l) The majority of mothers were unemployed (84%) and most were on governmental assistance (86%). Slightly under half of the women had not completed high school (42%), while 22% had completed high school, and 29% had taken some college courses. Eight percent had attended a trade school or had graduated from college. The total mean income for these mothers was $10,966. Forty-two percent of the mothers were married to their assailant at the time of the pre-interview. Thirty-seven percent were living together. (See Table 2) Seventy-five percent of the women had separated from their assailant at least once during the course of their relationship. Over half (56%) had separated three or more times. (See Table l) The majority (78%) of the women were not living with their assailants at the time of the pre-interview (this was immediately after they left shelter). The children of the women involved in the current study had a mean age of 5.7. Sixty percent of the children were 70 between the ages of 2 and 5. Over half of the children were male (60%). At the time of the pre-interview, 94% of the target children were under full custody of their mothers. All of the women, except one, were the natural mothers of the child participants. Fifty-nine percent of the assailants were their natural fathers. Twenty-two percent of the assailants who were not the natural fathers or legal stepfathers, were considered to be a father-figure to the target child. At the time of the pre-interview, 75% of the children were not living with the assailant. (See Table l) d en v i e. Since this research population is considered to be transient, and therefore difficult to contact for follow—up interviews, special efforts were made to track the living situations of the children over time. If a number of children had no contact with their mother over the 10-week intervention period, this could have affected the outcomes of this study, since measures were based solely on maternal report. At the time of the pre-interview, three women did not have their children living with them. Of these three women, one had visitation rights to see her child who was living with the assailant/father, one woman had no custody but could see her child who was living with her grandparents, and the other woman had full custody, but the child was temporarily living with the maternal grandmother while her mother was in the shelter. 71 Table 1 Demographics of Research Participants (measured at Pro-Intervention) N=52 Experimental Control Total a r1 )1 3. RACE Black 13 12 25 (48%) White 10 ll 21 (40%) Hispanic 3 l 4 ( 8%) Other 1 1 2 ( 4%) MEAN AGE 27 26. 27 MOTHERS WITH DEPENDENT CHILDREN Under 5 years old 21 19 40 (77%) 5 to 12 years old l4 17 31 (60%) 13 to 18 years old 1 3 4 ( 8%) MEAN NUMBER OF DEPENDENTS 2. 2. 2. CUSTODY Full 24 22 46 (88%) Some custody 2 2 4 ( 8%) None 1 l 2 ( 4%) EMPLOYED AT TIME OF INTERVIEW 3 5 8 (16%) CURRENTLY A STUDENT 3 2 5 (10%) EDUCATIONAL LEVEL < than high school 13 9 22 (42%) High school 5 6 11 (22%) Some college 5 10 15 (29%) College graduate 2 -- 2 ( 4%) Trade school 2 2 ( 4%) 72 Table 1 (cont'd) Experimental Control Total n a Ii 31 RECEIVING GOVERNMENT AID 24 21 24 (86%) TOTAL MEAN INCOME 10,754 11,188 10,966 NUMBER OF PREVIOUS SEPARATIONS None 7 6 13 (19%) One or two 7 3 10 (19%) Three to five 6 9 15 (29%) Six to ten 4 6 10 (19%) > ten 3 l 4 ( 8%) LIVING WITH ASSAILANT AT PRE-INTERVENTION 6 5 ll (22%) s i n MEAN AGE OF CHILDREN 5. 5.8 5.7 MODE=3 SEX Female 10 11 21 (40%) Male 17 14 31 (60%) MOTHER'S RELATIONSHIP TO CHILD Biological mother 26 25 51 (98%) Stepmother l -- 1 ( 2%) ASSAILANT'S RELATIONSHIP TO CHILD Biological father 17 13 30 (58%) Stepfather 2 2 4 ( 8%) Father figure 5 4 7 (21%) Not father figure 3 4 7 (13%) LIVING WITH ASSAILANT 7 6 13 (25%) 73 At the post-interview, only the woman who had full custody at the pre—intervention period had reunited with her child as planned and lived with her over the course of the ten weeks. Also, at the post-interview, two additional mothers were not living with the target child, making the total number of women at the post-interview who were not living with their children, but had significant contact with them over the intervention period to be 4. One of these mothers had full custody of her child at the pre-interview, but this status changed over the course of ten weeks to joint custody with the assailant/father. e a z b l o s e. Of the total women who use the shelter in the course of a year (based on 1988 figures with a total of 268 women) approximately 60% are white, 35% black, 4% Hispanic, and 1% are other women of color. About 70% of the women who use the shelter are mothers and range in age from 17 to mid-60s. The majority of the mothers (90%) brought three or fewer children with them to the shelter. In the pilot study of this project, 88% were mothers (N=36), and 53.7% had children under the age of 5 years old (Sullivan, 1988). Data about children's ages is not systematically collected at the shelter, but most tend to be pre—school to young school age. The current study sample's characteristics were similar to those women who seek shelter at CADA (see characteristics of sample, Table l). The women who seek shelter at CADA are very 74 Table 2 Mothers' Relationship to Assailant at Pre—interervention WW5. Married, living with Married, separated Divorced Girl/boyfriend, (living with) Girl/boyfriend (not living with) Ex-girl/boyfriend Dating, but not girlfriend/boyfriend N=52 WWW Q. 13 12 n. 5. 22 (42%) 5 ( 9%) 1 ( 2%) 19 (37%) 2 ( 4%) 2 ( 4%) l ( 2%) 75 similar demographically to women at battered women's shelters in general. As is typical of many shelters across the country, slightly over half of the residents are white, the majority are mothers; ages range from late teens to mid- 60s, and most of the residents lack resources, especially employment, higher education, money, and housing (Berk, Newton, & Berk, 1986; Finn, 1985; Hilbert & Hilbert, 1984). W The examination of the particular research hypotheses and questions of this study were conducted through a single factor design with pre— and post-measures. The experimental factor is an experimental condition (program or control), and there were two time intervals (pre- and post- intervention). See Table 3 for the design. Mtge. t Stud dv es Three methods were used to recruit the students for this project: letters describing the course were mailed to all female social science majors and women pursuing a women's studies thematic (a concentration of women's studies courses); through flyers posted across the university campus; through brief presentations by the director in front of students in appropriate courses who might be interested in the project. Two mandatory introductory meetings were held for interested undergraduates. The program was explained in 76 detail in the first meeting, stressing the time commitment and intensity of the course sequence. Grading procedures and class format were specifically outlined. The second meeting was required in order to assure the project directors that the students were committed to doing the project. e 't v c o Trained advocates assisted participants in identifying and accessing needed community resources and social support. Throughout their involvement, advocates taught their clients active coping strategies designed to maximize their possession of resources and to assist participants in becoming their own self-advocates by the end of the program. The services of 27 paraprofessionals were utilized in this intervention project over the course of the research. The processes involved in recruiting, training, and supervising students to be community advocates were modified from those methods already in place at MSU for the Adolescent Diversion Project (Davidson, et al., 1990). n v n o d Advocates received 40 hours of training over a 10 week period. They learned about the relationships among psychological variables, community resources, social support, and community responsiveness as they relate to battered women's future adjustment and ability to escape further abuse. They received intensive instruction detailing their role in this 77 Table 3 Experimental Design Experimental Control Pre-intervention Post-intervention (10 weeks) 27 24 25 25 78 process as social change agents: they understood, by the end of their training, the importance of assisting battered women in obtaining community and social resources as well as helping their clients develop more active coping strategies. An extremely important part of training consisted of linking advocates with existing community networks and teaching them effective strategies in obtaining sparse resources. Training included outside projects, roleplays, and guest speakers. Following training, advocates met once a week in small groups of six students and two supervisors to review the progress of their interventions. Supervisors continually stressed the conceptual model upon which the program was based, and ensured that advocates were satisfactorally meeting their short- and long-term goals. The importance of transferring skills to clients was emphasized repeatedly throughout training and supervision. The_ieee;yen§ien, Each advocate completed one intervention lasting ten weeks. Student advocates were required to spend four to six hours per week with, and on behalf of their clients, working on mutually identified goals. The intervention was comprised of four distinct phases: assessment, initiation of intervention, monitoring and termination. The assessment phase consisted of the advocate getting to know her client and assessing what areas they would like 79 to work on. This was the information-gathering stage in which short- and long-term goals were created. The next step or the second phase, was implementation of the intervention, where the advocates helped the women identify their unmet needs so they could generate and mobilize the appropriate resources to fill those needs. This included brainstorming possible resources, making phone calls, going for interviews with the person who controls the resource, assessing what strategies would be best to influence the person to give the needed resource to the client-~whatever it might take to fulfill the unmet need. The monitoring phase required that the advocate assess how effective the initial advocacy efforts were in accomplishing set goals. This phase was critical for the success of the implemented intervention. If the resource was not obtained, or if it was not satisfactory to the client, a secondary advocacy effort was implemented. This secondary advocacy effort consisted of following the same exact steps that were followed in the initial advocacy effort, i.e. the assessment procedures that include generating resources, identifying targets who hold the resources, and determining what strategy might best work to obtain the desirable resources, and finally implementation and monitoring of this strategy for its effectiveness. Finally, termination of the program began around week 8 of the intervention, as the advocate prepared for the end of 80 the program. During this time, the advocate was actively transferring the advocacy skills she possessed to the client. This was done by having the advocate gradually remove herself from playing an active role in assessing and obtaining resources for the client. Instead, the advocate increased instruction, modeling and role-playing of these skills with the client so that the client could access the resources on her own after the program has ended. This was done to ensure the probability that each woman was able to maintain and increase positive changes in her life. W The data for this research was collected through the use of face-to-face interviews with project participants. A total of 13 students were trained to do project interviews. These students were also informed of the time and energy requirements of this course, and that a commitment to two terms was necessary. Students who chose to participate signed a contract agreeing to remain throughout the two terms. W During the first term, the interviewers met once a week for two and a half hours with the interview trainers. The first few weeks were spent discussing written material in the practice of interviewing, and the remainder of the term was devoted to role-plays, class discussion, and outside assignments. Each student had extensive practice conducting 81 interviews before she was allowed to administer an actual program questionnaire. Inter-rater agreement on all interview items (assessed by calculating percent agreement across all items) was extremely high by the end of training (98-100%). After training, groups continued to meet once a week for two hours to review techniques and discuss the progress of their interviews. All of the interviews were audio-taped in order to re-play them periodically and critique them during class. Also, all of the tapes were listened to and re-coded by the interview trainer and another student so that inter-rater agreement was assessed continually. High inter—rater reliability was maintained throughout the course of the project (97-100%). Interviewers received grades for this course sequence based on attendance, expertise in interviewing techniques, and work completed. WWW Data were collected through face-to-face interviews with the mothers at their homes. Each of the 49 participants were given two interviews (pre and post). The participants were paid for each interview. The subjects received $15 for the first interview and $45 for the second. The 13 interviewers were reponsible for administering all of the face-to-face interviews for this research project. In order to minimize interviewer bias, the interviewer 82 who gave the first interview was never the same interviewer who subsequently interviewed the same woman. Interviews were conducted at the respondent's home unless otherwise requested. The first duty of the interviewer was to explain the reason for the interview and answer any questions the woman had. The second step in the interviewing process involved requesting that the session be tape-recorded. Interviewers explained to the participant that this measure was taken only to verify responses and that tapes would be used for research purposes only. They were also assured that their names would not appear on the tapes and the tapes would be destroyed after the research ended. If a mother had more than one child, the interviewer explained to her that she was only going to ask her about one of her children that was 2 years old or more. The interviewer explained that one child was randomly chosen to be asked about and was not being singled out for any reason. She further explained that obtaining information on all of her children in the family would be desirable but that time constraints made it prohibitive to do this. See Appendix B for the interviewer's introduction to the research interview on the children. Upon completing the interview, the interviewer recorded the length of the interview, and thanked the respondent and paid her. A receipt for the money was signed by the 83 respondent and returned to the project supervisors. They were paid $15 and $45 respectively. ion Procedures for retaining the women's participation over time were followed (see Rumptz, Sullivan, Davidson & Basta, 1990). At the woman's first interview, she was asked to name as many people as she could think of who could help the interviewer locate her. Participants signed a standard release of information letter for each person they mentioned, addressing it to that individual giving them permission to divulge the participant's whereabouts. The woman was also given a business card with the project's phone number, the date of her next interview, and the dollar amount she was to be paid. An answering machine was available 24 hours per day, every day, to field calls. In the pilot study of this project, similar methods were used, and 96% of the sample was retained (Sullivan, 1988). In the current research, 91% of the women were retained. WNW Data collection procedures of this research did not begin until the Michigan State University Committee for Research with Human Subjects (UCRIHS) approved this study. See Appendix C for the letter of approval from UCRIHS. The women voluntarily participated in this project. Before the first interview, each of the interested women were read the participant agreement form that outlined all 84 that was expected from them in the research and thereby obtained their informed consent. Most importantly, they were assured that all information gathered about them through the interviews was to be kept confidential. They were told that this confidentiality and anonymity was assured through the U.S. Department of Health, Education and Human Welfare certificate of confidentiality (see Appendix C for the copy of certificate). They were also informed that they could refuse to answer any questions during the course of the interviews, and that they could withdraw from the research at anytime, and not suffer any consequences. See Appendix B for the participant agreement form. fleaaurea Table 4 illustrates the measurement model for this research project. It defines what construct was measured, and identifies what variables were associated with this construct that was measured. Data sources for each construct are included. As indicated earlier, the research participants were administered an extensive verbal interview at two points in time--following the exit from the shelter (pre-intervention) and after the ten week intervention (post-intervention). In order to test the hypotheses and answer the research questions of this study, the interview contained pre- existing measures as well as those created specifically for the project. Specifically, the instruments measured: 85 l) outcome variables: women's effectiveness in obtaining resources; woman's psychological well-being and stress; mothers' involvement with assailant; extent of the violence in children's lives over time; children's behavioral adjustment; children's contact with assailant. 2) descriptive variables that are exploratory in nature: extent of violence in mothers' lives; children's reactions to the violence; extent of maltreatment of the children; feelings that children have about the assailant's absence; mothers' disciplinary tactics. The pre- and post-interviews were virtually identical except that the time reference points changed (i.e. asking respondents to think back 10 weeks ago instead of thinking back to 10 weeks at the pre-interview). The specific interview items about the children in the pre- and post- interviews can be found in Appendix D. Women in the experimental condition received the Intervention Process Interview that contained questions relating to their participation in, and satisfaction with, the advocacy program. Advocates were given face-to-face interviews at 10 weeks almost identical to the Intervention Process Interview except that they answered questions about the experiences of the women with whom they worked. This information was used to assess the validity of the information gathered. Women in the control condition were asked the same questions, but the advocate was not 86 mentioned. I — v'ew Inter—interviewer reliability was assessed by having each interview coded independently by two interviewers (one at the actual interview and one who coded responses based on the audio-tape). Most of the items on the interviews were forced-choice, Likert—type, or dichotomous type categories, which resulted in high agreement. Inter-interviewer agreement for the post-interview was .98. W The first 17 items of the interview consisted of questions eliciting information about the participant's demographic background such as race, age, number of children and their ages, type of custody of the children, employment background, educational background, income level and assistance, living situation, and access to transportation and use of shelters. We Multiple sources were used to measure this construct. Items about the women's education, employment, income, housing and access to transportation in the demographic variables section of the interview were asked of the subjects at both data collection points. The Intervention Process Interview, given at week 10, also measured what strategies the women used to obtain resources, how much time it took, whether the strategies were effective, and how Minuet Psychological well-being of mother Community resources Children's behavioral adjustment Child witness abuse Child maltreatment 87 Table 4 W Wes Sensing: depression stress social support psychological well- being legal housing, employment, assistance, education, childcare, finances, health, material goods externalizing and internalizing symptoms frequency of child observing mother being battered: verbal threats, physical harm frequency of verbal threats and physical harm, including sexual abuse 522é2££. SCL-90, self— report; CES-D, report self- Hassles and Uplifts, self- report; Parenting Stress Index; self- report Bogat et al.'s Social Support Scale, self- report Quality of Life Scale (Andrews & Withey), self- report Intervention Process scale, self-report Achenbach Child Behavior Check- list, maternal report Study interview maternal report Study interview; Revised Conflict Tactics Scale, maternal report; archival data from shelter 92mg; Mother's abuse Mother's psychological abuse Child's feelings about assailant 88 Table 4 (cont'd) W QQRELLHEL. frequency & severity of threats of being punched, pushed, kicked cut, burned, raped, restrained, shot; extent of injuries frequency of being criticized, ridiculed threatened a controlled amount of contact, and feelings about separation from assailant UEQEELQ Conflict Tactics Scale shelter files, Study interview Index of Psych- ological Abuse, self-report Study interview, maternal report 89 satisfied the women were with the results. e t e S t and i The advocates' detailed weekly progress reports, the intervention process interview (part of the post-interview), and the advocates' termination interview provided a manipulation check of the project. These measures assessed the amount of time advocates spent working with the women on various areas and what efforts the advocates made to accomplish the participants' goals. According to the women in the experimental group, an average of 5 hours was spent with the advocate each week, with a minimum of 1 hour to a maximum of 25 hours. Advocates indicated a similar amount of time (M_= 5.2). Participants generally saw their advocates 2 times per week, and spent an additional 2.8 hours per week on the phone. Advocates reported in their post—interviews how much time they spent trying to obtain resources for the participants. The least amount of time was spent on obtaining transportation (H.= 2.3; range 1 to 10 hours), while the most time-consuming area was working on legal issues (M_= 10.5, range = 1 to 60 hours). Advocates also reported that they discussed how to obtain resources an average of twice a week with the women and they obtained written materials a little less than 2 times per week (H = 1.6; range .40 to 3.5). In addition, advocates taught a new skill to the women an average of less 90 than once a week (M = .63; range 1 to 1.8). These data are similar to Sullivan's pilot study data (1988). Thus, this information provides evidence that the experimental manipulation occurred with sufficient strength and integrity. f Vi The Straus Conflict Tactics Violence subscale (1979) was used to measure the amount of physical abuse the assailants inflicted upon the mothers involved in this project. Straus found this to have an internal consistency of .83. This subscale was modified so that the mothers were asked how often they experienced various types of abuse (i.e., pushed, shoved, hit with an object, threatened with gun or knife). Two of the items were not included because of low variance: burned and drove recklessly. The Cronbach's alpha for this modified scale was .89. (See Table 5). The Index of Psychological Abuse was used to measure the extent and frequency of psychological abuse the assailant inflicted upon the mothers. This measure was created by Cris Sullivan (director of the Community Advocacy Project) and Julia Parisian at Michigan State University (Parisian, 1990). Thirty-three items ask the respondent how often the assailant engaged in ridiculing, controlling, threatening and insulting them. In the current study, the Cronbach's alpha was .94. Table 6 presents the item 91 reliability of this measure. 93W Psychological well-being of the mothers. Several instruments were used to measure the mothers' psychological well-being. Measures of depression/anxiety, social support and stress were used. Also, a modification of Andrews and Withey's (1976) Quality of Life measure was used to determine psychological well—being. Extensive psychometric research done on this scale has found it to be reliable and valid (overall alpha = .73). Twenty-five items which pertain specifically to the experiences of battered women had been selected to predict overall quality of life as well as the quality of life in specific domains-~housing, finances, family, self, independence, and personal safety. Cronbach's alpha for the global quality of life scale in this study was .87. The item reliability of this scale from the study can be found in Table 7. s i t d ' . These constructs were measured by three pre-existing measures. geeeee_was measured by The Hassles and Uplifts Scales (Kanner et al., 1981) which has 254 items, each representing a person or a situation. Each item is rated on how much of a hassle and/or uplift it has been for the subject during the last month. Kanner et al. have demonstrated that measuring hassles and uplifts of daily life is valuable for describing life stress in divergent populations and as possible 92 mediators of the effect of adaptation to major life events. The version in this study was revised and scaled down to 24 items by choosing marker items from the subscales that had a factor loading of .50 or greater (see Appenix G). Further scale revision was made based on item reliability analysis. Nineteen items were retained for the analysis. The Hassles and the Uplifts scale shared 8 items in common. The items in Tables 8 and 9 display what items were retained for the data analysis. Cronbach's alphas for the revised Hassles scale was .78 and .87 for the Uplifts scale. fit;eee_was further measured by the Parenting Stress Index (PSI) (Abidin, 1983). This measure was designed to assess the relative degree of stress between parent and child. The most current version has 120 items. The directions ask the client to think of the child she/he is most concerned about and to mark the answer that best describes his/her feelings along a continuum of 1 (Strong agree) to 5 (Strongly disagree). Items 102-120 are optional and involve a check for each life event that has occurred in the past 12 months. These items were not used in the present study. See Appenix F for the modified version of this measure. The PSI test items are divided into two source domains of stressors: 1) Child Characteristics Domain, 2) Parent Characteristics Domain. The child domain determines stress associated with specific child characteristics in areas including adaptability, 93 TABLE 5 Mothers' Physical Abuse Scale Reliability (Modified Conflict Tactics Scale) N=52 ITEM Corrected Item Total Correlation 1 TORE CLOTHING, BROKE GLASSES .61 2 PUSHED 0R SHOVED .66 3 GRABBED .69 4 SLAPPED .53 5 HIT WITH FIST .65 6 KICKED .68 7 THREW AN OBJECT .50 8 HIT WITH AN OBJECT .47 9 TRIED TO HIT WITH AN OBJECT .62 10 CHOKED .48 11 TIED UP OR RESTRAINED .46 12 RAPED .41 13 THREATENED WITH GUN OR KNIFE .76 14 USED GUN OR KNIFE .68 ALPHA .89 94 TABLE 6 Mothers' Psychological Abuse Scale Reliability N=52 ITEM Corrected Item Total Correlation madame-cowh- REFUSED TO TALK TO YOU ACCUSED OF HAVING SEXUAL RELATINSHIPS TOLD ABOUT SEX HE WAS HAVING TO HURT YOU REFUSED TO DO THINGS WITH YOU FORBID YOU TO GO OUT WITHOUT HIM TRIED TO CONTROL YOUR MONEY TRIED TO CONTROL YOUR ACTIVITIES WITHELD APPROVAL, AFFECTIONS TO PUNISH YOU LIED TO YOU OR MISLED YOU MADE CONTRADICTORY DEMANDS 0R REQUESTS OF YOU CALLED YOU NAMES TRIED TO HUMILIATE YOU IGNORED OR MADE LIGHT OF YOUR ANGER IGNORED OR MADE LIGHT OF YOUR OTHER FEELINGS RIDICULED OR CRITICIZED YOU IN PUBLIC RIDICULED OR INSULTED YOUR MOST VALUED BELIEFS RIDICULED 0R INSULTED RELIGION, RACE, CLASS RIDICULED OR INSULTED WOMEN AS A GROUP CRITICIZED YOUR STRENGTHS CRITICIZED YOUR INTELLIGENCE CRITICIZED YOUR PHYSICAL APPEARANCE CRITICIZED YOUR FAMILY OR FRIENDS TO YOU HARASSED YOUR FAMILY OR FRIENDS IN SOME WAY DISCOURAGED YOUR CONTACT WITH FAMILY OR FRIENDS THREATENED TO HURT YOUR FAMILY OR FRIENDS BROKE OR DESTROYED SOMETHING IMPORTANT TO YOU ABUSED OR THREATENED TO ABUSE PETS TO HURT YOU PUNISHED OR DEPRIVED KIDS WHEN HE WAS ANGRY THREATENED TO TAKE THE CHILDREN IF YOU LEFT THEM LEFT YOU SOMEWHERE WITH NO WAY TO GET HOME THREATENED TO END THE RELATIONSHIP TRIED TO FORCE YOU TO LEAVE YOUR HOME THREATENED SUICIDE WHEN HE WAS ANGRY AT YOU ALPHA = .94 .38 .60 .52 .39 .59 .53 .79 .75 .64 .73 .58 .74 .51 .57 .74 .72 .67 .57 .67 .58 .52 .68 .47 .69 .49 .61 .24 .53 .35 .42 .48 .65 .12 95 Table 7 Psychometric Properties of Modified Quality of Life Scale N=52 ITEM Corrected Item Total Correlation FEELINGS ABOUT.... LIFE AS A WHOLE .39 WHERE YOU ARE LIVING NOW .27 YOURSELF .54 EMPLOYMENT SITUATION .39 HEALTH AND PHYSICAL CONDITION .45 FINANCIAL SECURITY .37 AMOUNT OF PRIVACY .43 SECURE FROM THEFT OR DESTROYING PROPERTY .23 PERSONAL SAFETY .26 AMOUNT OF FUN AND ENJOYMENT ONE HAS .57 CHANCES OF GETTING A GOOD JOB .31 RESPONSIBILITIES FOR FAMILY .35 ACCOMPLISHMENTS IN LIFE .52 INCOME .26 FRIENDS .57 INDEPENDENCE OR FREEDOM .53 STANDARD OF LIVING .35 CLOSE ADULT RELATIVES .38 EMOTIONAL AND PSYCHOLOGICAL WELL-BEING .53 HANDLING PROBLEMS THAT COME UP .32 SOCIAL SERVICES DEALINGS .26 FAMILY LIFE .53 HOW MUCH ACCEPTED AND INCLUDED BY OTHERS .46 WAY SPENDS SPARE TIME .63 LIFE AS A WHOLE .67 COEFFICIENT ALPHA - ALL ITEMS = .87 96 acceptability, demandingness, mood, destractibility/hyperactivity, and reinforcement of the parent. The parent domain measures stress associated with the parental role and includes the subscales of: depression, attachment, restrictions of role, sense of competence, social isolation, relationship with spouse, and parental health. Several studies have provided evidence of the concurrent, construct and discriminant validity of the PSI (Abidin, 1982; Awalt, 1981; Bristor; 1982; Lafiosca, 1981). In this study, the PSI was reduced to 38 items for the following reasons: the PSI has a reported high internal consistency (Cronbach's alpha = .95); some of the items were not appropriate for the target population; and time constraints of the interview. The revised scale contains two to three items within each of the 13 subscales with factor loadings of .50 or greater. In addition, seven items were dropped from the scale because of low corrected item total correlations, resulting in a total of 31 items that were used in the data analysis. The Cronbach's alphas for the child domain and parent domains were .87 and .88, respectively. Initially, it was proposed to keep the Parent and Child Domains separate for the purposes of this study. However, the two domains were highly correlated (.73), and they correlated in a parallel manner with many of the study's other outcome measures. Therefore it was deemed 97 best to use the total scale score in the analyses for data reduction purposes. Cronbach's alpha for the total scale was .93. (See Table 10) er;eeeien_ was measured by two different scales. Forty items were taken from the Symptom Checklist-90 (SCL- 90) (Derogatis, 1977) which asked the respondent to describe how much discomfort a specific problem had caused them during a certain time period. The SCL-90 has high internal consistency (.77—.90) and stability, and its validity has been demonstrated in numerous studies (Derogatis, 1983). The Center for Epidemiological Studies Depression Scale (CES-D) contains 20 items that asks the respondent how often certain feelings or behaviors were encountered in the past week. Adequate psychometric properties have been demonstrated for this scale (Radloff, 1977). Because the average scale scores for the SCL-90 and the CES-D correlated highly with each other (.70 - .90), and also for data reduction purposes, the scales were combined into one depression scale. More specifically, the scale score for the sixteen items from the CES-D was standardized and combined with the standardized scale score for the SCL-90. See Table 11 for the item reliability analysis of the combined items for the depression scale. SQQLEL_EHEEQLL.W38 assessed by a measure developed by Bogat et al. (1983). It assesses five domains of social support: 1) emotional support; 2) practical assistance; 98 TABLE 8 Uplifts Subscale Reliability N=52 ITEM Corrected Item Total Correlation 2. Other relatives .32 5. Your supervisor employer .38 6. Meeting deadlines goals at work .56 7. Your work load .67 8. Enough money for necessities .60 9. Enough money for emergencies .57 10. Enough money for extras .50 11. Your medical care .52 12. Your health .62 13. Your physical ability .44 14. News events .48 16. Political or social issues .44 17. Cooking .45 18. Housework .49 23. Being organized .43 24. Social commitments .54 COEFFICIENT ALPHA = .87 99 TABLE 9 Hassles Subscale Reliability N=52 ITEM Corrected Item Total Correlation 8. Enough money for necessities .40 9. Enough money for emergencies .38 10. Enough money for extras .43 12. Your health .37 15. Your environment .54 17. Cooking .51 18. Housework .53 20. Yardwork .30 22. Home entertainment .46 23. Being organized .47 24. Social commitments .40 COEFFICIENT ALPHA = .78 100 3) advice and information; 4) companionship; and, 5) overall social support. Questions are asked within each of the above subscales regarding the amount of support received and satisfaction with the support. Cronbach's alpha was .81. (See Table 12.) WE:— All of the mothers were asked at the post-interview how effective their efforts had been in achieving their goals in each specific target area they worked on (e.g., housing, legal, education). Response categories were in the form of a Likert-type scale (range = 1 to 4): 1 = not effective at all, 2 = not very effective, 3 = somewhat effective, 4 = very effective. For the purposes of determining the item reliability of this scale, those areas a woman did not work on (coded as not applicable or 8 by the interviewer) were coded to a neutral score of 2.5. Cronbach's alpha was .73. (See Table 13.) For the data analyses, an overall effectiveness score was calculated by averaging the- responses across all of the areas a woman worked on. NEW The children's adjustment was assessed by several pre-existing measures and items specifically developed for this study. The Child Behavior Checklist (CBCL) (Achenbach & Edelbrock, 1983) is designed to assess the social competencies and behavior problems of children ages 4 through 16 as reported by their parents and others who might know them. A version 101 for 2-3 year old children is also available and shares 57 items from the version for the older children. The CBCL was developed with a clinically referred population in mind, therefore the content, scales and profile types all reflect this clinical emphasis (Mooney, 1984). It contains 118 childhood problems and parent(s) are asked to rate how often each occurs. (The version for 2-3 year olds has 99 items). This instrument is standardized so that a child's scores can be compared to clinical and "normal" scores. The items measuring the social competency of the children were not used in this study due to time constraints. Achenbach & Edelbrock (1983) suggested that either one of the portions of the test (behavior problem portion or social competency) can be used alone without jeopardizing the usefuleness of the scales. Several "narrow band" behavior problem scales were derived for each age and sex group, along with two second-order factors that appear to tap the dimensions of Externalizing and Internalizing problems. These "narrow band" scales are placed along the continuum of Internalizing-Externalizing along with a Mixed designation of those scales to comprise the Revised Child Behavior Profile that provides the basis for the normative comparisons of a child's problem behavior scores. Test- retest reliability of the CBCL scores filled out by mothers of nonreferred children at one-week intervals were generally in the eighties or low nineties (.81 to .96). Content, 102 Table 10 Parenting Stress Index Reliability (Modified Version) N=52 ITEM Corrected Item Total Correlation l. VERY HARD FOR MY CHILD TO GET USED TO NEW THINGS .41 3. USUALLY AVOIDS NEW TOY BEFORE PLAYING WITH IT .44 5. CHILD IS NOT ABLE TO D0 AS MUCH AS EXPECTED .60 6. CHILD HAS DIFFICULTY CONCENTRATING .54 7. SOME THINGS CHILD DOES REALLY BOTHERS ME .63 8. LIST OF NUMBER OF THINGS THAT BOTHER MOTHER .50 10. CHILD CRIES OR FUSSES MORE OFTEN THAN MOST .55 ll. CHILD IS VERY MOODY AND EASILY UPSET .38 12. CHILD SEEMS HARDER TO CARE FOR THAN MOST .40 13. CHILD IS SO ACTIVE IT EXHAUSTS ME .65 14. CHILD 13 MORE ACTIVE THAN EXPECTED .55 15. CHILD SQUIRMS AND KICKS A LOT WHEN DRESSED .37 16. CHILD LIKES ME AND WANTS ME TO BE CLOSE .60 18. CHILD SMILES AT ME MUCH LESS THAN EXPECTED .55 19. DEPRESSED & SAD AFTER LEAVING HOSPITAL WITH BABY .64 20. WHEN CHILD MISBEHAVES,FUSSES, I FEEL RESPONSIBLE .62 21. AFTER HOME WITH BABY, MORE SAD AND DEPRESSED .57 22. EXPECTED TO HAVE CLOSER AND FEELINGS FOR CHILD .56 23. TAKES TIME TO DEVELOP FEELINGS FOR CHILD .46 25. OFTEN FEEL CHILD'S NEEDS CONTROL MY LIFE .57 26. CHILD PREVENTS ME FROM DOING THINGS I LIKE TO DO .48 27. GIVE UP MORE OF LIFE TO MEET CHILD'S NEEDS .52 28. WHAT KIND OF PARENT MOM FEELS SHE IS .63 30. NOT INTERESTED IN PEOPLE AS I USED TO BE .64 31. OTHERS MY OWN AGE DON'T LIKE MY COMPANY .41 32. WHEN I GO TO PARTY I DON'T EXPECT TO ENJOY IT .55 33. SINCE HAD CHILD, SPOUSE GIVES LITTLE SUPPORT .44 34. HAVING CHILD CAUSED TROUBLED IN RELATIONSHIP .36 35. SINCE HAD CHILD, SPEND LITTLE TIME WITH SPOUSE .46 36. PHYSICALLY I FEEL GOOD MOST OF THE TIME .45 38. DURING PAST 6 M08, SICKER THAN MOST .45 ALPHA = .93 103 TABLE 11 REVISED DEPRESSION SCALE RELIABILITY (Standardized Scores) n=52 ITEM Corrected Item Total Correlation Sixteen items from CES-D : l BOTHERED BY THINGS THAT USUALLY DON'T BOTHER .32 2 DIDN'T FEEL LIKE EATING .46 3 COULDN'T SHAKE OFF THE BLUES, EVEN WITH HELP .60 5 HAD TROUBLE KEEPING ATTENTION .65 6 FELT DEPRESSED .64 7 FELT EVERYTHING WAS AN EFFORT .49 9 FELT MY LIFE WAS FAILURE .36 10 FELT FEARFUL .42 ll SLEEP WAS RESTLESS .71 13 TALKED LESS THAN USUAL .60 14 FELT LONELY .47 15 PEOPLE WERE UNFRIENDLY .44 17 HAD CRYING SPELLS .48 18 FELT SAD .57 19 FELT PEOPLE DISLIKE ME .33 20 COULD NOT "GET GOING" .46 Items from SCL-90 That Were Combined with CES-D: 1 LOW IN ENERGY OR SLOWED DOWN .50 2 FEELING OF BEING CAUGHT OR TRAPPED .50 3 FEELING ALONE .66 4 FEELINGS OF WORTHLESSNESS .59 5 AFRAID OF OPEN SPACES OR STREETS .60 6 AFRAID TO GO OUT OF HOUSE ALONE .48 7 AFRAID TO TRAVEL ON BUSES, SUBWAYS OR TRAINS .50 8 AVOID CERTAIN THINGS, PLACES OR ACTIVITIES .46 9 UNEASY IN CROWDS, SUCH AS SHOPPING OR MOVIE .61 10 NERVOUS WHEN LEFT ALONE .54 11 TROUBLE FALLING ASLEEP .72 12 AWAKENING IN THE EARLY MORNING .34 13 SLEEP THAT IS RESTLESS OR DISTURBED .58 14 SUDDENLY SCARED .51 15 FEARFUL .70 16 SPELLS OF TERROR OR PANIC .75 17 SO RESTLESS YOU COULDN'T SIT STILL .61 18 FEELING SOMETHING BAD IS GOING TO HAPPEN .70 19 THOUGHTS AND IMAGES OF A FRIGHTENING NATURE .72 20 FEELINGS MOST PEOPLE CAN'T BE TRUSTED .55 21 FEELING THAT YOU ARE WATCHED OR TALKED ABOUT .64 104 Table 11 (cont'd) Revised Depression Scale Reliability ITEM Corrected Item Total Correlation 22 23 24 25 26 27 28 29 30 FEELING THAT PEOPLE WILL TAKE ADVANTAGE OF YOU FEELING SHY OR UNEASY WITH OPPOSITE SEX FEELING OTHERS DON'T UNDERSTAND YOU UNEASY WHEN PEOPLE ARE WATCHING VERY SELF-CONSCIOUS WITH OTHERS HEARS VOICES OTHERS DO NOT HEAR LONELY EVEN WITH PEOPLE THOUGHTS ABOUT SEX THAT BOTHERS YOU A LOT IDEAS THAT SOMETHING IS WRONG WITH YOUR MIND .51 .53 .46 .67 .67 .24 .47 .58 .47 105 construct and criterion-related validity have been adequately demonstrated for this measure. In this study, a total of 55 items from both the CBCL for 4-16 year olds and 2-3 year olds were used because of the time constraints of the interview. (See Appendix E for the items.) The measure was conducted verbally instead of in writing. Generally, items that had .50 or higher factor loadings on both the externalizing and internalizing dimensions for both girls and boys across all age groups were chosen. Five items from the version for 2-3 year olds that were not in the 4-16 age version were also added. Seven items that did not load .50 or higher on the scales were included because they were considered appropriate for the target population. In addition, one new item, created for the study, "acts too old for age, overprotective of others," was added because it was considered to be a behavior often found within the target population. After data were collected and preliminary analyses were conducted, six items were dropped from the analyses because of low response frequency. Further, 15 additional items were dropped from the analysis because of low corrected item total correlations. This resulted in a final number of 36 items that were used in the data analyses to measure children's adjustment. Cronbach's alphas for the internalizing and externalilzing dimensions were .88 and .90, respectively. See Tables 14 and 15 for the item 106 TABLE 12 Mother's Perceived Social Support Scale Reliability N=52 ITEM Corrected Item Total Correlation FEELINGS ABOUT.... 1 AMOUNT OF COMPANIONSHIP .22 2 QUALITY OF COMPANIONSHIP .44 3 AMOUNT OF ADVICE AND INFORMATION OBTAINED .61 4 QUALITY OF ADVICE AND INFORMATION OBTAINED .30 5 AMOUNT OF PRACTICAL ASSISTANCE ONE RECEIVES .47 6 QUALITY OF PRACTICAL ASSISTANCE ONE RECEIVES .66 7 AMOUNT OF EMOTIONAL SUPPORT ONE RECEIVES .61 8 QUALITY OF EMOTIONAL SUPPORT ONE RECEIVES .66 9 OVERALL AMOUNT AND QUALITY OF SOCIAL SUPPORT .67 COEFFICIENT ALPHA = .81 107 TABLE 13 Effectiveness in Obtaining Resources Scale Reliability N=49 ITEM Corrected Item Total Correlation Perceived effectiveness in... 1 HOUSING .48 2 EDUCATION .39 3 TRANSPORTATION .50 4 EMPLOYMENT .39 5 LEGAL ISSUES .25 6 PHYSICAL HEALTH .26 7 SOCIAL SUPPORT .39 8 FINANCIAL ISSUES .45 9 MATERIAL GOODS AND SERVICES .35 10 CHILDCARE .32 11 OTHER ISSUES FOR CHILDREN .40 ALPHA = .73 108 reliability analysis of both these dimensions. The alpha for the total revised CBCL was .92. See Table 16. Meeepres for children's experience of violence. Three items were developed specifically for this project that directly assessed the amount and type of abuse the children had witnessed. The mothers were asked how many times their children had witnessed specific types of emotional and physical abuse from a specific time reference point (10 weeks before their stay at the shelter, and 10 weeks after leaving the shelter.) These three items were combined into a "witness subscale." (See Appendix D for these specific items.) (See Table 17 for the item reliability analysis.) Six items were created for this study that measured the assailant's maltreatment of the target child. These included psychological/verbal abuse, physical and sexual abuse and neglect. Also, the mothers were asked in two items whether the assailant intentionally or unintentionally abused the child when he was also abusing her. These items appear in Appendix D. All of these items were combined into a ”victim" scale. The sexual abuse item was not included because of low response frequency. The Cronbach's alpha for the victim scale was .78. (See Table 18.) The total of 8 items measuring the children's experience of abuse were combined into an overall experience of abuse scale. Four items were recoded to have the same response categories as those of the other abuse items so they could be combined 109 TABLE 14 Children's Internalizing Problems Scale Reliability N=52 ITEM Corrected Item Total Correlation 1 ACTS TOO YOUNG FOR AGE .41 2 FEARS WILL DO SOMETHING BAD .66 3 FEELS HAS TO BE PERFECT .60 4 FEELS NO ONE LOVES THEM .58 5 FEELS WORTHLESS .63 6 ANXIOUS, FEARFUL .59 7 FEELS TOO GUILTY .78 8 ACHES OR PAINS .43 9 REFUSES TO TALK .46 10 SECRETIVE .65 ll UNHAPPY OR SAD .78 12 WITHDRAWN .54 13 WORRYING .65 14 CHEATS AT SCHOOL .43 15 POOR SCHOOL WORK .57 COEFFICIENT ALPHA - ALL ITEMS .90 110 TABLE 15 Children's Externalizing Problems Scale Reliability N=52 ITEM Corrected Item Total Correlation l ARGUES A LOT .55 2 CRUELTY, MEANNESS TO OTHERS .40 3 DEFIANT .45 4 DEMANDS A LOT OF ATTENTION .43 5 DESTROYS OWN THINGS .49 6 DESTROYS THINGS OF OWN FAMILY OR FRIENDS .69 7 DISOBEYS AT HOME .56 8 DOESN'T GET ALONG WITH OTHER CHILDREN .42 9 IS TOO JEALOUS .42 10 GETS IN FIGHTS .42 11 ATTACKS OTHER PEOPLE .43 12 PUNISHMENT DOESN'T CHANGE BEHAVIOR .58 13 SCREAMS A LOT .57 14 STEALS AT HOME .41 15 STEALS OUTSIDE OF HOME .40 16 IS TOO STUBBORN .53 17 MOODY OR SULLEN .49 18 TEASES A LOT .49 19 HOT TEMPER, TANTRUMS .52 20 THREATENS OTHERS .46 21 IS TOO LOUD .39 COEFFICIENT ALPHA - ALL ITEMS .88 111 TABLE 16 Children's Behavioral Adjustment Scale Reliability (Revised Version of Child Behavior Checklist) N=52 ITEM Corrected Item Total Correlation l ACTS TOO YOUNG FOR AGE .38 2 FEARS WILL DO SOMETHING BAD .58 3 FEELS HAS TO BE PERFECT .47 4 FEELS NOONE LOVES THEM .59 5 FEELS WORTHLESS .67 6 ANXIOUS .60 7 FEELS TOO GUILTY .65 8 ACRES OR PAINS .43 9 REFUSES TO TALK .42 10 SECRETIVE .52 11 UNHAPPY OR SAD .56 12 WITHDRAWN .48 13 WORRYING .48 14 CHEATS AT SCHOOL .41 15 POOR SCHOOLWORK .46 16 ARGUES A LOT .55 17 CRUELTY, MEANNESS OR BULLYING .40 18 DEFIANT .49 19 DEMANDS A LOT OF ATTENTION .39 20 DESTROYS OWN THINGS .40 21 DESTROYS THINGS OF FAMILY OR FRIENDS .58 22 DISOBEYS AT HOME .51 23 DOESN'T GET ALONG WITH OTHER CHILDREN .33 24 IS TOO JEALOUS .35 25 GETS IN TOO MANY FIGHTS .32 26 ATTACKS OTHER PEOPLE .24 27 PUNISHMENT DOESN'T CHANGE BEHAVIOR .57 28 SCREAMS A LOT .46 29 STEALS AT HOME .56 30 STEALS OUTSIDE OF HOME .43 31 IS TOO STUBBORN .53 32 SULLEN OR MOODY .58 33 TEASES A LOT .51 34 HAS A HOT TEMPER, HAS TANTRUMS .41 35 THREATENS OTHERS .51 36 IS TOO LOUD .42 COEFFICIENT ALPHA - ALL ITEMS = .92 112 into one scale. Overall alpha for this scale was .86. (See Table 19.) In addition to these items, the mothers were asked how their children reacted to the violence (see Appenix 0.) These items had extremely low internal consistency, and thus were not used together as one scale. t n on e c e V r e The outcome variables were examined for the their inter-relatedness in order to guage whether they were assessing similar or different constructs. As shown in Table 20, there are several significant correlations between the outcome variables, and several high but not significant correlations. Parenting stress (PSI) and children's adjustment were expected to correlate because many of the items in the PSI concern the mothers' perceptions about their children's characteristics. Although not significant, mothers' hassles and depression correlated with children's adjustment. These correlations were not surprising and are often found in maternally reported child behavior measures. This issue is discussed more thoroughly below and in Chapter 4. Also, a significant inverse relationship was found between mothers' depression and her life satisfaction (-.64 and social support (-.47). Since all of these variables are considered measures of the mothers' psychological well-being, these correlations were desirable, thus indicating they are tapping a similar construct. 113 Table 17 Witness Abuse Subscale Reliability ITEM Corrected Item TotalCorrelation Child witnessed threats .57 Child witnessed physical abuse .76 Child witnessed psychological abuse .74 ALPHA = .87 114 TABLE 18 Children's Victimization Subscale Reliability N=52 ITEM Corrected Item Total Correlation When Mom psychologically abused so was child .62 When Mom physically abused so was child .51 Psychological abuse of child .77 Physical abuse of child .62 Neglect of child .33 COEFFICIENT ALPHA - ALL ITEMS = .78 115 Table 19 Children's Experience of Abuse Scale Reliability N=52 ITEM Corrected Item Total Correlation Child witnessed threats .67 Child witnessed psychological abuse .64 Child psychologically abused when mom also abused .62 Child witnessed physical abuse .70 Child phys. abused when mom also physically abused .51 Child psychologically abused .73 Child deprived of food and other necessities .36 Child physically abused .60 Alpha = .86 116 However, the high significant correlation (.77) found between quality of life and social support may have indicated that perhaps the social support variable was not measuring a distinct enough construct to warrant using it as a separate measure. Caution was used in interpreting the results from this variable. MSW W The Conflict Tactics Scale (Straus, 1979) was revised in order to make the wording more appropriate for younger children. In this scale, widely used in national surveys to determine the incidence and prevalence of child maltreatment, the parent is asked to rate how often they engaged in each of the ways of dealing with conflict described by the items. See Appendix H for this measure. Two items were dropped from the original scale because they were considered to be inappropriately worded for the target population. The resulting 13 items had a Cronbach's alpha of .87. (See Table 21.) ' b ara Items were developed for this project to assess what type of feelings the child had about the assailant if he was not living with them. The following questions were asked of the mothers to obtain this information: how often the child was allowed to see the assailant, how often the child asked to see the assailant, how they felt about the assailant being 117 gone, and how often they asked if their mothers would reunite with the assailant. (See Appendix D for these items.) m: Some of the measures that were used in this study are on file at the Community Advocacy Project, Psychology Department, Michigan State University. The measures included in the appendices were especially developed for this dissertation thesis. amch No Hammanonamwmnwoo >aoc=m Ocnnoam ccmm N orwwanm=.m >aucmwamsn w mommwmm b cvwwmnm m moansnwam mmammm o Zonrma.m cmcnmmmwo: u Zonsmn.m Ocmpwnw 0m rwmm m zonrma.m monHmH mcwponn c Mmmmonwmmmwwmsn Hs