an}: a?! 3.. ‘ .9...” 34......» .i. n. . .55.? .. r 5 .1 . {-101 . .. I ,5 2 ..., a. i: . 7.: £329.? “maximum; . _.u..”§$ vsfififiafit A... .f. , THY: $13 ’1. ,1 M I This is to certify that the dissertation entitled F OF EXPOSURE TO DOMESTIC VIOLENCE ON CHILDREN: A ONGITUDINAL EXAMINATION OF CHILDREN '8 OUTCOMES presented by Angela M. Wolf has been accepted towards fulfillment of the requirements for U 4 2 Major professor Date -' é" ' 00/ MSUiJ A ff 1' A (flan/Equal Opportunity anlitution 0" Irma we 0-12771 LIBRARY Michigan State University PLACE lN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 2009 Flt/392075324 1 05 0 9 i) 6/01 c:/C|RCJDaioDua.pB5—p.1 5 THE EFFECT OF EXPOSURE TO DOMESTIC VIOLENCE ON CHILDREN: A LONGITUDINAL EXAMINATION OF CHILDREN’S OUTCOMES By Angela Marie Wolf A DISSERTATION Submitted to Michigan State University In partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 2001 ABSTRACT THE EFFECT OF EXPOSURE TO DOMESTIC VIOLENCE ON CHILDREN: A LONGITUDINAL EXAMINATION OF CHILDREN’S OUTCOMES By Angela Marie Wolf There is increasing evidence that exposure to domestic violence against their mothers has a negative effect on children’s functioning including their levels of depression, self-concept, problematic behaviors, and somatic complaints. However, conclusions from the current literature are limited given that most findings have been based on cross-sectional research. Further, there is a limited knowledge base regarding how exposure to domestic violence affects children and equivocal finding regarding the degree to which the effect of abuse is different for boys and girls. In the current study, the problematic behaviors of children exposed to the psychological and physical abuse of their mothers were examined longitudinally. In addition, the presence of disruptions (e.g., household moves, changes in schools) in the lives of children was examined as a potential mediator of the relationship between exposure to abuse and children’s outcomes. Finally, to account for differences regarding the effects of exposure on boys and girls, gender was examined as a moderator. Data used to examine these relationships Were collected fi'om interviews with eighty mothers and their children as part of a larger Study examining the effects of emotional and physical abuse over time on women and their school-age children. The findings from this study demonstrated that exposure to domestic violence was positively related to children’s levels of “internalizing” or withdraWal behaviors and “externalizing” or aggressive behaviors. This study also contributed to the current knowledge base regarding how and when children are affected by domestic violence. In particular, findings indicated that the relationship between exposure to domestic violence and some aspects of children’s behaviors were moderated by child’s gender and partially mediated by the number of disruptions in children’s lives. The conclusions and implications for future research were explored. ACKNOWLEDGMENTS I would like to thank some very special people in my life who supported me during this process. First and foremost, thanks to Dr. Cris Sullivan and Dr. Bill Davidson. As my dissertation chair, Sully was a wonderful source of support, guidance, and also laughter. Sully is the kind of researcher/activist that I strive to be, and I appreciate her wisdom and scholarship. Bill Davidson has been a fabulous mentor throughout my graduate career. He was generous with his time, experiences, and advice, and was a constant source of encouragement. Cris and Bill have had an important impact on my professional life, but in addition, they have also been wonderful friends. From their friendships, the lessons I have learned have been innumerable and will forever contribute to the way I approach my life. Thanks also to my committee members Dr. Deborah Bybee and Dr. Alytia Levendosky. This project would not have been possible without the support and guidance of Deb. Her outstanding teaching gave me the confidence to approach these analyses and I learned a great deal. I also appreciate my meetings with Alytia and her careful feedback. I had the support of many wonderfirl fi'iends who made my graduate school eXperience enjoyable. In particular, my friendship with Nicole Allen has been a Wonderful gift and has encouraged me to grow many ways. Nicole also provided a great deal of instrumental support throughout this process. She read many drafis of this project and her feedback was invaluable. David Loveland also made many contributions to the iv completion of this project. David was always available both for help and fim. I would also like to thank Dr. Craig Blakely for putting me on this path of work that I love so much and for supporting me all the way. Thanks to Chris and Michelle Coulter forvv opening their home to me. And thanks to Dr. Madeline Wordes for the many pep talks and the freedom to finish this degree While starting my career. Lastly, I would like to thank my family. Thanks to my dad for his high expectations, to my mother for always listening, and my brother for keeping things in Perspective. TABLE OF CONTENTS List of Tables ......................................................... ix List of Figures ......................................................... x Introduction ............................................................ 1 Literature Review .................................................. 3 What is “Exposure?” ............................................... 3 Effects of Exposure to Domestic Violence on Children .................... 4 Emotional and Behavioral Functioning of Children Exposed ................ 6 Variability in Children’s Outcomes ................................... 10 Limitations of the Literature ........................................ 13 Children’s Resiliency .............................................. 14 Risk and Protective Factors Influencing the Effects of Exposure ............ 15 Severity of abuse ........................................... 15 Psychological abuse ......................................... 16 Disruptions in child’s life ..................................... 17 Gender ................................................... 18 Children’s direct experience of abuse. ........................... 19 Conclusions ..................................................... 21 Current Study .................................................... 22 Methods .............................................................. 24 Participants ...................................................... 24 vi Procedure ....................................................... 25 Conducting Interviews ...................................... 25 The Intervention ........................................... 26 Measures ....................................................... 26 Assailant’s psychological abuse of mother ....................... 27 Assailant’s physical abuse of mother ............................ 27 Disruptions in Child’s Life ................................... 28 Assailant’s abuse of child .................................... 28 Children’s internalizing and externalizing behavior. ................ 29 Analysis Strategy ................................................. 30 Effect of Mother’s Experience of Abuse on Children’s Functioning . . . 31 The Moderating Effect of Gender .............................. 33 The Mediating Effect of Disruptions in Children’s Lives ............ 33 Results ............................................................... 35 Retention over time ............................................... 35 Psychological Abuse .............................................. 35 Physical Abuse .................................................. 36 Disruptions ...................................................... 37 Children’s experience of abuse by assailant ............................ 37 Intemalizing Behaviors ............................................ 38 Intemalizing behaviors over time .............................. 38 Moderating effect: Gender. .................................. 42 vii Mediating effect: Disruptions in child’s life. ...................... 43 Extemalizing behaviors ............................................ 49 Shape of the data over time ................................... 49 Moderating effect of gender ................................... 52 Mediating effect: disruptions in child’s life. ...................... 54 Analysis summary ................................................ 56 Discussion ............................................................ 58 Children’s behaviors over time ...................................... 59 Exposure to psychological abuse against mother ......................... 59 Disruption as a mediating variable .............................. 61 Exposure to physical abuse against mother ............................. 62 Limitations ...................................................... 65 Conclusion ...................................................... 66 Appendix A Assailant’s Psychological Abuse of Mother ........................ 98 Appendix B Assailant’s Physical Abuse of Mother ............................ 101 Appendix C Assailant’s Injury of Mother ................................... 103 Appendix D Disruptions in Child’s Life .................................... 105 Appendix E Child’s Experience of Abuse-Mother’s Report ......................... 107 Appendix F Mother’s Report of Children’s Intemalizing/Extemalizing Behaviors . . . 108 References ........................................................... 1 l 1 viii LIST OF TABLES Table 1 ............................................................... 69 Table 2 ............................................................... 70 Table 3 ............................................................... 73 Table 4 ............................................................... 73 Table 5 ............................................................... 75 Table 6 ............................................................... 76 Table 7 ............................................................... 77 Table 8 ............................................................... 78 Table 9 ............................................................... 79 Table 10 .............................................................. 81 Table 11 .............................................................. 82 Table 12 .............................................................. 84 Table 13 .............................................................. 86 Table 14 .............................................................. 87 Table 15 .............................................................. 88 Table 16 ................. l ............................................. 90 Table 17 .............................................................. 92 Table 18 .............................................................. 94 ix LIST OF FIGURES Figure l .............................................................. 96 Introduction Although the subject has only recently received attention by social scientists, service providers, and the general public (Tomkins, et al., 1994), intimate partner violence has occurred for centuries (Edleson, 1999). The statistics representing the numbers of women living in violent relationships demonstrate that such violence is widespread. It is estimated that the number of woman who are battered ranges fi'om three to six million each year (Tjader & Thoemmes, 1999; Tomkins, et al., 1994). To place these estimates in perspective, while 39,000 American soldiers were killed in the Vietnam War, 17,500 American women and children were killed by family members during the same time period (Grusznski, Brink, & Edleson, 1988). . Clearly, the current level of violence against women constitutes a vast social problem, and the effects of intimate partner violence are far-reaching. Many survivors of such abuse are mothers of children who are forced to cope with violence in their homes. Given the prevalence of domestic violence, it is not surprising that estimates of the number of children exposed to their mother’s abuse range from 3 million to over 10 million (Carlson, 1984; Straus, 1992; Tomkins, et al., 1994). Straus (1990) estimates that one third of American children have witnessed violence between their parents. These figures are supported by the work of F antuzzo, Boruch, Abdullahi, Atkins, and Marcus (1997) who, in an analysis of police arrest data from five major US. cities, found that between 9% and 27% of domestic violence calls directly involved children. Further, at least 70% of women seeking shelter have children with them and 17% of women have three or more children with them (MacLeod, 1987). Clearly, from any of these estimates, many children are exposed to violence against their mothers. Such prevalence is concerning given that children who are exposed to abuse against their mother are negatively affected in two overarching ways. First, domestic violence places the child at risk of direct harm. Children face direct risk of physical injury, either intentional or accidental, especially if they attempt to interrupt an assault (Carlson, 1984; Straus, Gelles, & Steinmetz; 1980; Tomkins, et al., 1994). Straus, Gelles, and Steinmetz (1980) reported that the extent of the overlap between woman battering and child abuse is 40% to 60%. Second, research to date indicates that children who are exposed to violence against their mothers are at risk for a number of negative developmental consequences. Yet, despite this association with negative outcomes, children exposed to violence against their mothers had been largely ignored by researchers until the 19805 (F antuzzo & Mohr, 1999). Within the last 20 years, there has been a marked increase in research on the effects of exposure to domestic violence on children. Although much of the early research in the area predominantly relied on anecdotal reports from families residing in domestic violence shelters, recently there has been a more systematic attempt to examine the efi’ect that witnessing their mothers’ abuse has on children. This research points to a 7 number of detrimental outcomes for children who witness domestic violence. However, there is evidence that children are not uniformly affected by exposure to abuse against their mothers. In fact, some children appear to be fairly resilient against the negative consequences of witnessing abuse. More work in this area is needed to further our understanding of how and to what extent witnessing domestic violence has an impact on children. Further, conclusions drawn from the current research are limited by a number of methodological issues, including the over-reliance on research with families in domestic violence shelters and the lack of longitudinal research. The proposed study builds on this body of research by addressing both conceptual and methodological gaps. Conceptually, the proposed study contributes to current research by focusing on children’s resiliency and by moving beyond the examination of the direct relationship between witnessing domestic violence and children’s outcomes to include ecological variables theorized to mediate or moderate this relationship. Methodologically, the proposed study advances available research by utilizing a community sample and a longitudinal perspective. These issues are important to consider if we are to diminish the negative consequences of witnessing domestic violence for children. Li_te__rat11re Review What is “Exposure?” Children exposed to domestic violence are often labeled as “witnesses.” Although this term is often widely defined to capture a range of children’s experiences, it cay be misleading. Children may not actually view the violence in their homes but may be exposed to it in other ways. Children may overhear an incident, see the consequences (e. g., mother’s injuries), or experience changes in their mother’s behavior after the violent incident (Jouriles, McDondald, Norwood, & Ezell, 2001). Further, there is some evidence that the occurrence of violence is associated with negative consequences for children, even when children are not direct witnesses. For instance, J ouriles, McDonald, Norwood, et al. (1998), found that increased child behavior problems were associated with the occurrence of knife or gun violence, against their mother, not necessarily with witnessing the violence directly. Give the discrepancy in definitions, children’s exposure to domestic violence has been measured in a variety of ways. Children and/or mothers have sometimes been asked to report on the violence the child has actually witnessed. However, it remains unclear as to how accurately young children can report incidences of violence in their homes, especially within specific time fi'ames. Further, some researchers have found that women underestimate their children’s exposure to domestic violence and that children are often aware of violence even if they have not witnessed it directly (e. g., Jaffe, Wolfe, & Wilson, 1990). Other studies assess children’s exposure to violence by using the mother’s report of the violence she has experienced (Bookless—Pratz & Mertin; F antuzzo, DePaola, Lambert, Martino. Anderson, & Sutton, 1991; Hughes & Luke, 1998; Levendosky & Graham-Bermann, 1998; Kolbo, 1998). Although each method has associated limitations, the measurement method should be chosen to address specific research questions. For example, to assess questions related to children’s interpretation of violent events, using children’s report of the violence they have witnessing may be the optimal method. In other cases, the violence the mother has experienced may be the most appropriate measure of children’s exposure. For the remainder of this proposal, the term “witness” will be broadly defined to include the range of children’s experiences and exposure to violence. ffects o E s to omestic Violence 11 'l Across the majority of studies conducted in this area, there is evidence that exposure to domestic violence is related to a variety of negative outcomes for children (Edleson, 1999; Holden, 1998; Kolbo, Blakely, & Engleman, 1996). When compared to a comparison group or population norms, children who are exposed to domestic violence often exhibit a higher number of problems, including school related problems and somatic complaints (such as headaches, stomach aches, and intestinal problems) (Kolbo, 1996). Also, there is substantial evidence suggesting that exposure to domestic violence can have negative consequences on children’s behavior. Across studies, children exposed to domestic violence are likely to display greater aggressive behaviors or externalizing behaviors (Davis & Carlson, 1987; Hershom & Rosenbaum, 1985; Holden & Ritchie, 1991; Hughes, 1988; Hughes, Parkinson, & Vargo, 1989; Porter & O’Leary, 1980; Westra & Martin, 1981; Jafi’e, Wolfe, Wilson, & Zak, 1986; Wolfe, Jaffe, Wislon, & Zak, 1985), and withdrawal or internalizing behaviors (Fantuzzo et al., 1991; Hughes, 1988; Hughes, Parkinson, & Vargo, 1989; Jaffe, Wolfe, Wilson, & Zak, 1986; Wolfe, Jaffe, Wislon, & Zak, 1985) compared to comparison populations. Children who are exposed to domestic violence have also been found to exhibit more depression, anxiety, trauma, and other psychological problems (Graham-Bamann & Levendosky, 1998; Hughes, 1988; Maker, Kemmelrneier, & Peterson, 1998; Stemberg et al., 1993) and are at risk for feelings of lower self-competence (Adamson & Thompson, 1998; F antuzzo, et al., 1990). Originally, the link between children’s exposure to domestic violence and children’s behavior problems was made through descriptions of children and their behavior in domestic violence shelters (J affe, Wolfe, and Wilson, 1990; Margolin, 1998). Based on these initial reports, several researchers began to empirically investigate the relationship between witnessing abuse and children’s outcomes, often focusing specifically on children’s emotional and behavioral functioning. Although there have been a small number of empirical studies examining the outcomes of children outside shelters, the available research is often criticized for its over-reliance on samples residing in shelters. Shelters are predominately utilized by women with few other resources and are associated with a period of crisis for families, and consequently, it is difficult to separate the effects of witnessing violence fiom the efi‘ects of limited resources or the specific challenges associated with the crisis period. In addition, the shelter experience itself is often hectic and stressful, and represents an artificial environment for families. While studies with women residing in shelters are informative and critical to our understanding of the effects of witnessing domestic violence on children, studies with women who are not shelter residents have made unique contributions to our knowledge in that they are more generalizable. Emotional and Behavioral Fflcfioning of Children Exmsed to Domestic Violence. The majority of research on the outcomes for children who have witnessed domestic violence has focused on children’s emotional functioning and behavior problems. In general, the research has attended to two types of behavior problems: externalizing and internalizing behaviors. Overall the literature has found that children who witness violence against their mothers are at a greater risk of developing externalizing behaviors, also known as aggressive or “acting out” behaviors (e. g., Fantuzzo et a1, 1991; McCloskey, Figueredo, & Koss, 1995; O’Keefe, 1994), and are also at risk for problems with internalizing behaviors, or fearful or inhibited behaviors (e.g., Fantuzzo et a1, 1991; Graham-Bermann & Levendosky, 1998; Holden & Richie, 1991; Maker, Kemrnelmeier, & Peterson, 1998). Although studies on children’s exposure to domestic violence have been conducted with participants in multiple settings, the majority of our knowledge has been gained fi'om research with mothers and children who are temporarily residing in domestic violence shelters (Grych, Jouriles, Swank, McDondald, & Norwood, 2000; Holden & Richie, 1991; Hughes & Luke 1998; Mathias, Mertin, & Murray, 1995; O’Keefe, 1994; Wolfe, Jaffe, Wilson, & Zak 1985). In one of the early empirical studies, Wolfe, Jaffe, Wilson, and Zak (1985) compared 102 children recruited fiom domestic violence shelters or transitional housing to who had exposure to domestic violence 96 children recruited fiom the community who had no exposure to domestic violence. Their findings indicated that, according to their mothers’ reports, children who were exposed to domestic violence experienced significantly more problems with externalizing and internalizing behaviors than comparison children. Further, more children exposed to domestic violence exhibited behavior problems in the clinical range than comparison children (26.5% versus 10.4%). Such outcomes persist even after controlling for other variables such as the relationship quality between mother and child, socioeconomic status, and children’s individual characteristics (Mathias, Mertin, & Murray, 1995; O’Keefe, 1994). For example, in a sample of 184 children fiorn 120 families living in a battered women’s shelter, O’Keefe (1994) examined the effect of witnessing domestic violence on several child outcomes while controlling for a number of other variables such as the relationship quality between mother and child, and children’s temperament, school performance, self- worth, and age. Although there was no comparison group, analyses indicated that the amount of violence witnessed by children was a significant contributor to children’s externalizing and internalizing behaviors, even after controlling for other factors. Clearly, the research indicates that many children residing in domestic violence shelters have problems with internalizing and externalizing behaviors (Grych, et al., 2000; Holden & Richie, 1991; Hughes & Luke, 1998; Mathias, Mertin, & Murray, 1995; O’Keefe, 1994; Wolfe, et al., 1985). These findings are consistent when controlling for other variables such as temperament, academic ability, and socioeconomic status (Mathias, Mertin, & Murray; 1995; O’Keefe, 1994). However, as previously stated, research with children residing in domestic violence shelters is often subjected to criticism. It is difficult to ascertain whether behavior problems are associated with the upheaval of the shelter stay or with the exposure to violence against their mothers. Further, shelters are predominantly utilized by women with few other available resources and limited social support. Although the predominance of research on children who witness domestic violence has been conducted with shelter residents, there has been some research with community participants. These studies are important because they begin to separate the effect of the disruption of shelter stay children’s outcomes. . Even with the removal of the stressful shelter conditions, studies with community- based participants also indicate that children who witness domestic violence experience negative outcomes (Fantuzzo et al., 1991; Graham-Bermann & Levendosky, 1998; Hershom & Rosenbaum, 1985; Jouriles, et al., 1989; Levondosky, et al., 2002; McCloskey, Figueredo, & Koss, 1995). However, studies with community-based participants offer insights about factors that may contribute to the adverse consequences of children exposed to domestic violence such as parental discord and disnrptions in the child’s life. For example, Jouriles, et a1. (1989) found that children exposed to domestic violence experienced higher levels of behavior problems than children in discordant but nonviolent homes. In their study of 87 children of couples who had requested marital counseling, children from “aggressive” families (n=5 6) were compared to children from “nonaggressive” families (n=31). Their findings indicated that children of maritally aggressive families exhibited more behavior problems than children of parents experiencing nonviolent conflicts. These results are important because they provide evidence that marital violence has a harmful impact beyond the efl’ects of parental discord. Likewise, Fantuzzo and his colleagues (1991) found that exposure to interparental violence adversely affected children beyond the effects of nonviolent parental conflict. Further, they found that children in shelters had more negative outcomes than children who witnessed domestic violence but remained in their homes. In this study, pairs of women and children were recruited from a school program (n=84) and from a domestic violence shelter (n=23). Children were placed in one of four groups according to their exposure to interparental violence: 1) living in shelter and exposed to verbal and physical violence (n=23); 2) living at home and exposed to verbal and physical violence (n=27); 3) living at home and exposed to verbal abuse only (n=30); and 4) no exposure to abuse and living at home (n=27). Their results indicated that young children were adversely affected by witnessing interparental physical and verbal abuse. Children who witnessed interparental violence, whether living in shelters or in their homes, had significantly more problems with externalizing behaviors (both group means were in the clinical range) than the other two groups. For internalizing behaviors, children in shelter had a mean score in the clinical range and had significantly more problems than children who witnessed abuse and were at home. The importance of Fantuzzo et al.’s work is twofold. First, it supports the previous work of Jouriles, et a1. (1989) by finding that witnessing domestic violence negatively affects children beyond the effects of living with parental discord. Second, it provides evidence that children in shelters appear to have more detrimental outcomes than children who witness domestic violence but who are not in shelter. In summary, all children exposed to domestic violence are at risk for problems with internalizing and externalizing behaviors (F antuzzo, et al., 1991; Graham-Bermann & Levendosky, 1998; Grych, et al., 2000; Holden & Richie, 1991; Hughes & Luke 1998; Jouriles, et al., 1989; Mathias, Mertin, & Murray, 1995; McCloskey, Figueredo, & Koss, 1995; O’Keefe, 1994; Wolfe, et al., 1985), with children who are in shelters, at greater risk (F antuzzo, et al., 1991). Additionally, exposure to domestic violence appears to affect children’s behavior beyond the effects of nonviolent interparental conflict (Fantuzzo, et al., 1991; Jouriles, et al., 1989). Although the preponderance of studies indicates that children who witness domestic violence experience negative consequences, it is important to note that not all children are equally affected by exposure to violence. First, evidence indicates that children have varying response patterns. Some children experience significant deficits while other children appear to be relatively unaffected. Second, some studies did not report children who witnessed domestic violence to be at a significant deficit (Chistopoulos, et al. 1987; Wolf, Zak, & Wilson; 1986). MW Current research with children exposed to domestic violence clearly indicates that many children are adversely affected by their experiences of witnessing abuse. However, several studies have noted that some children seem to be more resilient against the negative consequences associated with witnessing violence. For example, the early work of Jouriles, Murphy, and O’Leary (1989) noted that although more children fi'om martially aggressive families had adjustment problems at the clinical level than children fi'om discordant, nonaggressive homes, 50% of children fi'om aggressive families did not 10 report problems at the clinical level. Likewise, Hughes and Luke (1998) found that 40% of children living in shelter at the time of their study were reported as distressed and experiencing difficulties. This illustrates that the majority of children were not experiencing distress. In addition, the work of Hughes and Luke illustrates that there is variability in the outcomes of children who are exposed to domestic violence. Using cluster analyses, Hughes and Luke found distinct patterns in children’s responses to witnessing domestic violence. The 40% of children reported to be experiencing distress were grouped into one of three clusters: “high behavior problems,” “high general distress,” or “depressed kids.” Children in the “high behaviors problem” (n=9) experienced high amounts of behavioral difficulties; children in “high general distress” (n=9) were characterized by exhibiting behavioral and emotional problems and having low self-esteem; and children in the “depressed kids” (n=4) group had fewer behavior problems but exhibited emotional problems including moderately high anxiety and high depression. The remaining children fell into one of two clusters that indicated that children were experiencing few adjustment difficulties. The largest group of children (n=21) were described as “hanging in there” and experienced average problems, some mild anxiety, and average self-esteem. Fifteen children were described as “doing well” and experienced few behavior problems, no anxiety, and high self-esteem. Although the generalizibility of their findings is limited due to the small number of participants (n=58), these results are a significant contribution to the literature as they demonstrate that the outcomes of children who are exposed to domestic violence are not uniform. Building upon the work of Hughes and Luke (1998), Grych, et al., (2000) ll conducted a study with 228 children in shelter to examine patterns of outcomes among children of battered women as reported by both mothers and children. The results of their study supported the findings of Hughes and Luke (1998); they found that children had varying levels of emotional and behavior problems and that some children reported low distress. The results of their cluster analysis indicated that 51% of children experienced no problems or mild problems. They found that children clustered into five patterns of adjustment, with the largest cluster of children (31%) in the “no problems reported” group. An additional eighteen percent of children were in the “mild distress” group described as slightly elevated scores on internalizing measure (only 5% of children in this group were in the clinical range). This variability in children’s outcomes indicates two points. First, not all children are equally, adversely affected by exposure to domestic violence. In fact, some children appear to be rather resilient against the harmful effects of witnessing abuse against their mothers. Second, the variability in children’s outcomes suggests that there are other variables that may mediate or moderate the effects of exposure to domestic violence. Further, while some studies have found that children experience varying degrees of detriment, other studies have not found children to be significant deficits. For example, Wolfe and his colleagues (1986) expanded upon previous findings in a subsequent study of 63 children who were placed into one of three groups: 1) children residing in a domestic violence shelter, 2) children who had not been exposed to domestic violence within the prior six months and whose families were considered “generally stable,” and 3) children who had never been exposed to domestic violence. In contrast to their previous work, no between group differences on externalizing and internalizing 12 behaviors were uncovered. Further, results from a study conducted by Chistopoulos et a1. (1987) indicated that the domestic violence group experienced greater internalizing behavior problem than the comparison groups. However, there was not a significant between-group difference for children’s externalizing behaviors. Taken together, these studies indicate that there are gaps in our knowledge of how and to what extent witnessing domestic violence impacts children. One reason for this is that there are a number of common methodological limitations in the current literature. W In addition to the limitations of relying on data collect while women are in shelter mentioned earlier, the literature is plagued with a number of other problems. One problem with current research is that few studies follow children over time. Although researchers often make statements about the developmental consequences of witnessing domestic violence, the majority of studies have been cross-sectional in nature, and thus, a cause-and-effect relationship between witnessing domestic violence and children’s functioning cannot be determined. It is important to examine changes over time in order to examine what factors contribute to children’s wellness. Wolfe, et a1. (1986) found that the amount of time since the violent event was important to the degree to which children exhibited behavior problems. Results of this study indicated that children residing in shelters during the study exhibited greater social problems than children who stayed in shelter in the past. Thus, the passage of time could ameliorate the negative effects of witnessing mother’s abuse. A third issue with the research on children’s exposure to domestic violence is that it has yet to examine moderating factors that may affect the relationship between 13 children’s witnessing domestic violence and outcomes. The relationship between children’s witnessing violence and their fimctioning is likely complex, and while studies Show that children exposed to their mothers’ abuse are at risk for a number of negative outcomes, the specific model has yet to be determined (J affe, Wolfe, & Wilson, 1990). Moreover, children do not appear to be uniformly affected. While some children experience detrimental consequences, other children appear to be relatively resilient. Although exposure to violence is clearly undesirable, these findings indicate that additional work should be conducted to explore factors that may contribute to the development, and more importantly, the prevention of negative consequences. '1 ’ 'li c Although little work has been done in the area regarding resiliency of children wimessing domestic violence (Hughes, Graharn-Bennann, & Gruber, 2001; Kolbo, 1996), related fields have identified that many children escape negative outcomes or even grow stronger despite stressful environments and experiences (Garbarino, Dubras, Kosteny, & Pardo, 1992; Garmezy, 1993; Werner, 1993). In fact, it has been estimated that up to 80 percent of children exposed to extreme stressors do not sustain developmental damage (Garbarino, et al., 1992). Resiliency in children can be characterized by the presence of a positive outcome despite potentially harmful circumstances (Masten, 2001). It is the ability to successfully overcome adversity (Garmey, 1993) but should not imply that children are invincible to risk factors (Masten, 2001; Zimmerman & Arunkumar, 1994). Researchers have proposed several models of resiliency (see Garmezy et al., 1984 for summary); however, the most widely studied model (Zimmerman & Arunkumar, 14 l994)-- the protective factor model-- will be discussed here. A protective factor is a variable that moderates the effect of exposure to a risk factor (Zimmerman & Arunkumar, 1994). There are a number of factors that seem to serve as protective factors for children in stressful circumstances. For example, individual factors such as activity level, temperament, ability to elicit positive attention from adults, and cognitive skills have been associated with adjustment in children (Garmezy, 1993; Gore & Eckenrode, 1997; Werner, 1993). Ecological factors, such as environmental resources, social support, and positive parenting have also been found to relate to children’s adjustment (Kolbo, 1996; Garmezy, 1993; Rhodes, 1994; Gore & Eckenrode, 1997; Werner, 1993). L .rq Pto-v. 'v Factors __1 iuencin .re i ects ._ .. -osure t- I . L - ' V . arce While the literature on domestic violence finds that, overall, witnessing domestic violence is detrimental to children’s development, the variability in children’s adjustment remains unexplained and insufficiently explored. It is important to consider factors that may influence the degree to which exposure to domestic violence is detrimental. Thus far, the research points to several aspects of witnessing violent conflict in the home that seem to be related to children’s outcomes. For example, evidence suggests that the severity of the violence (F antuzzo, et al., 1991; Jouriles, Norwood, McDonald, Vincent, & Mahoney, 1996), exposure to psychological abuse (Levendosky & Graham-Bermann, 1998), and disruptions in the child’s life (Peled, 1998; Spaccarelli, et al., 1994) influence the relationship between exposure to violence and children’s adjustment. However, there is evidence that other factors may alter the impact of exposure to domestic violence on children’s functioning. Each of these factors are explored below. Severig of abuse. Several dimensions of witnessing abuse, such as the frequency 15 and severity of the violence, appear to be related to children’s outcomes. It has been proposed that frequent and severe violence may sensitize children to violence and also increase the child’s risk for developing behavior problems (Adamson & Thompson, 1998; Fincharn & Osborne, 1993; Grych & Frincham, 1990; Hershom & Rosenbaum, 1985; O’Keefe, 1994). Several studies have found that increased frequency of exposure to domestic violence is associated with an increase in children’s adjustment and behavior problems (F antuzzo, et al., 1991; Hershom & Rosenbaum, 1985; Jouriles, Norwood, McDonald, Vincent, & Mahoney, 1996). For example, Grych, J ouriles, Swank, McDonald, and Norwood (2000) found that the severity of children’s adj ustrnent problems was positively related to children’s perception of the amount of abuse against their mother. Pacholo gica_l abuse. Although more research is being conducted to examine children’s outcomes related to physical abuse, the effect of psychological abuse has been largely neglected. Not only do children of battered women witness their mother’s physical abuse but they are also exposed to the psychological or emotional abuse that is in the home. Although few studies have directly examined the effects of emotional abuse, those that have found that it does impact to children’s development (Fantuzzo, et al., 1991; Jouriles, et al., 1996; Levendosky & Graham-Bermann, 1998). For example, Levendosky and Graham-Bermann (1998) found that the fiequency of emotional abuse against the mother and mother’s self-esteem were the largest predictors of children’s externalizing behaviors. Moreover, their results indicated that young children exposed to their mothers’ emotional abuse had a higher rate of name-calling, insults, and put-downs towards their peers than children in a comparison group. In fact, findings fiom their 16 research indicated that psychological abuse was an even stronger predictor of children’s adjustment than physical abuse (Levendosky & Graham-Bermann, 1998). Further, evaluations of treatment programs for batterers have indicated that treatment is somewhat successful in reducing physical violence but the majority of men continue to be verbally abusive (Edleson & Syers, 1990; Gondolf, 1993). Thus, the children may continue to be exposed to emotional abuse even after violence has discontinued. Clearly, because emotional abuse is detrimental and may continue to be prevalent after physical abuse has stopped, the relationship between children’s exposure to emotional abuse and their functioning requires further attention. Dimtigns m child’s life. Another factor that may illuminate the impact of domestic violence on children is the level of disruptions in their lives. Children who live in homes with domestic violence often experience critical disruptions in the continuity of their home environment. Many women and children are forced to flee their homes in search of safety and end up in shelters or with extended family. Such disruptions cause immediate upheaval in children’s daily routines and may also include changing schools and/or neighborhoods (Peled, 1998; Spaccarelli, et al., 1994). Research has found that major disruptions in children’s lives contribute to their functioning. For example, Fantuzzo et a1. (1991) indicated that children who were temporary shelter residents exhibited higher levels of internalizing behaviors and lower social competence than children in their homes who witnessed violence and a nonviolent comparison group, even after controlling for the amount of violence witnessed. Some researchers have suggested that the transitions in children’s lives-- such as a move to a shelter- may separate children from their natural buffers to stress, and thus, exacerbates the negative 17 consequences of exposure to domestic violence for children (F antuzzo et al., 1991; Jouriles, Pfiffner, & O’Leary, 1988). For instance, O’Keefe (1994) examined frequent stressful events (e.g., moving to a new school) and found that these events exacerbate the effects of witnessing domestic violence. However, few researchers have included this as a mediating or moderating variable in empirical research. find—er. Although it is often hypothesized that exposure to domestic violence will affect boys and girls differently, research is this area has produced mixed results. Early discussions of gender differences supposed that girls would have less exposure to domestic violence because families would tend to shield girls from the household violence (Emery, 1982). Empirical investigations of this hypothesis have indicated that girls and boys are actually exposed to similar levels of domestic violence (O’Keefe, 1994), and in some studies girls are more adversely impacted by this exposure than boys (Spaccarelli, Sandler, & Roosa 1994; Stemberg et al., 1993). For example, in their study of l 10 children in Israel, Stemberg et al. (1993) reported that girls exposed to family violence experienced more behavior problems and depression than boys, according to both children’s and mother’s reports. Similarly, in a study with 303 fourth through sixth graders recruited from inner-city schools, Spaccarelli, Sandler, and Roosa (1994) found that history of violence against their mothers was associated with girls’ but not boys’ adjustment. However, other studies have reported that exposure to domestic violence has a greater effect on boys’ fimctioning than girls’ (J ouriles et al., 1986; J omiles & Norwood, 1995; Wolfe et al., 1985). For instance, Wolfe et al., (1985) found that boys exposed to domestic violence had more total behavior problems than did girls. Similarly, Jouriles 18 __‘ Viv and Norwood (1995) reported that, while family violence was positively related to externalizing behavior problems for both boys and girls, boys exhibited greater levels of these behaviors. Children’s direct experience of abuse. One of the limitations that is often cited in the “child witnessing” literature is the failure to investigate the differential effects of witnessing mother’s abuse alone versus witnessing mother’s abuse and also being abused (F antuzzo et al., 1991; Kashani, Daniel, Dandoy, & Holcomb, 1992; O’Keefe, 1996). Research has shown that there is often an overlap between interparental aggression and parent-child aggression (Grych, et al, 2000; Henning, et al., 1996; Jouriles & LeCompte, 1991; J ouriles & Norwood, 1995; McCloskey, Figueredo, & Koss, 1995). Thus, parent- child aggression is an important variable to consider in examining the effects that witnessing domestic violence has on children. In studies that have addressed the differential effects of witnessing domestic violence versus experiencing parent-child abuse, the results indicate that children are at a deficit when compared to a comparison groups on a number of outcomes. Findings across studies indicate that children who witness their mothers’ abuse and experience direct abuse are themselves at risk of a “double whammy” (Hughes, Parkinson, & Vargo, 1989). Many researchers have found a continuum of outcomes for children. Children who are both abused and exposed to their mothers’ abuse demonstrate significantly more problems than children who only witness abuse, while children who witness their mother’s abuse have significantly more problems than non witnessing comparison children (Hughes, 1988; Markward, 1997; O’Keefe, 1995; Rossrnan, 1998; Stemberg, er al., 1993). For example, Stemberg, et a1. (1993) examined 110 children and their parents 19 recruited fi'om social workers in Israel. Families were divided into four groups depending on their exposure to family violence: “child abuse” (n=33, children who had experienced abuse by one or more parents); “spouse abuse” (n=l6, children who had witnessed violence between their parents but were not themselves victims); “abused witness” (n=30, children who both witnessed violence between their parents and experience abuse); and “comparison” (n=30, children who had neither witnessed interparental violence nor been abused). According to children’s own reports, children who were exposed to family violence (whether “child abuse,” “spouse abuse,” or “abused witness”) were significantly more depressed than comparison children. Further, subsequent analyses indicated that the family violence groups did not differ significantly from each other. Children in the “abused witness” and “child abuse” groups reported significantly more internalizing and externalizing behaviors than children in the “spouse abuse” or “comparison” groups. Mothers’ reports of children’s behavior also indicated significant between-group differences. Mothers who were victims of domestic violence (spouse and abused witness groups) reported that their children had significantly more externalizing behaviors and a trend towards more internalizing behaviors than the other two groups. Interestingly, mothers of the child abuse group did not report more child behavior problems than the comparison group. In conclusion, children who were both witnesses of their mothers’ abuse and victims of abuse have significantly greater problems associated with anxiety (Hughes, ‘ 1988), depression (Stemberg, etal., 1993); and internalizing and externalizing behavior problems (Hughes, 1988; O’Keefe, 1995; Stemberg, et al., 1993; Rossman, 1998). These findings speak to the importance of controlling for child’s experience of abuse in studies 20 designed to assess the effects of children’s exposure to their mother’s abuse. Conclusions The current research on the impact of exposure to domestic violence on children indicates that exposure to violence negatively affects many children’s behaviors and levels of emotional problems. However, not all children appear to suffer these consequences. Some children appear to be resilient against the negative effects of exposure to domestic violence. Presently, there is a need to examine factors that may bufi’er the negative effects of witnessing domestic violence. There is evidence that a number of variables may alter or mediate the relationship between exposure to domestic violence and children’s functioning. In order to understand what may contribute to children’s resiliency, more longitudinal work is needed to provide a clear picture of the process by which exposure to domestic violence affects children. Now that researchers are attending to the ramifications of domestic violence on children, the next step is to investigate factors that are likely to moderate or mediate the impact of witnessing domestic violence on children (Hughes & Luke, 1998), and longitudinal research is key to this goal. The present study intended to examine the relationship between children’s exposure to domestic violence and children’s firnctioning while addressing three of the common criticisms of the literature in this area. First, although most of the families were recruited from shelters, none of the families were interviewed during their shelter residence. Second, children were followed over time. Explanatory variables as well as indicators of fimctioning were examined at three points in time over the course of a year. Third, the process by which children are affected through exposure to domestic violence was examined including a both a mediating and 21 moderating relationship. ELEM The present study was part of a larger project examining the effects of emotional and physical abuse over time on women and their school-age children. It was also a pilot study designed to explore an intervention with mothers and children experiencing family violence. The goal of the proposed research was to examine the effect that exposure to domestic violence has on children and to explore what factors may be contributing to children’s resilience over time by addressing three research questions. It is important to note at the explanatory and dependant variables were measured at three points in time in order to reflect the changing nature of these factors over time. . The first purpose of this study was to examine the degree to which domestic violence affects children’s fimctioning over time. To address this purpose the following research question was examined. To what extent did exposure to domestic violence impact children’s internalizing and externalizing behaviors over time? It was hypothesized that children’s exposure to the abuse mothers were experiencing (physical and psychological) had a negative impact on children’s internalizing, and externalizing behaviors over time. The second purpose of this study was to examine how domestic violence influences children’s lives over time. To address this purpose, this study examined whether exposure to mother’s abuse impacted children’s functioning by creating disruptions in children’s lives (e.g., household moves, moves to shelter, changing schools)? It was hypothesized that, in addition to the direct effect of exposure to violence on clli 1dren’s functioning, exposure to domestic violence negatively affected children by 22 creating disruptions in their lives. The third purpose of this study was to investigate the effect of children’s gender on the relationship between mother’s experience of abuse and children’s fimctioning over time. Given that the research in this area remains equivocal, it was important to examine this relationship since it seemed plausible that boys and girls are affected differently by exposure to domestic violence. This issue was addressed by the following research question and hypothesis. Did children’s gender predict variation in the impact of their exposure to violence over time? It was hypothesized that the relationship between children’s exposure to domestic violence and children’s functioning was moderated by their gender. Insert Figure 1 about here The current study contributes to the available literature in three important areas. First, the current study included families residing in the community, rather than interviewing family during shelter residence. Second, the current study is longitudinal and examines children’s resiliency over time. Third, in an effort to understand why children have different outcomes, the current study examined two variables that may mediate or moderate the relationship between witnessing violence and children’s positive fimctioning. 23 v Methods Participants Eighty women and eighty of their children were recruited after exiting two local domestic violence shelters (79%), a Social Service department (18%), and a community- based family services agency (4%), all located in a medium-sized Midwestem city. All women who participated in this study had to meet the following requirements: (1) had experienced abuse at the hands of their intimate partners within the previous four months; (2) had at least one child between the ages of seven and eleven; (3) and planned to remain in the area for the following four months. About half of the women identified themselves as non-Hispanic, white (49%), while 39% identified as Afiican American, 5% as Latina/Hispanic, 5% as bi-racial, 1% as Asian Americans, and 1% as Native American. The average age of the women was 31 with an age range fi'om 23 to 46. Most of the sample (77%) was under 35 years of age. Women’s monthly family income ranged fi'om $0 to $3,472 with average monthly income at $1,199. Forty-four percent of women were employed and the majority (88%) were receiving government assistance. Twenty-two percent of the sample had not completed a high school degree, 26% were high school graduates, 8% had completed a trade school degree, 39% had attended some college, and 5% had college degrees. Forty-four percent of the children in the sample were African American, and 40% were identified as non-Hispanic white. The remaining children were identified as multiracial (10%), Hispanic (5%), and Asian (1%). The children’s average age was 8.3 and the average level of school was third grade. A slight majority (55%) of the children Were girls. 24 Procedure All participants were part of a longitudinal study examining outcomes of intimate partner violence on mothers and their children, and were interviewed three times over a period of one year (at four-month intervals). Once mothers agreed to participate in the study, children within the appropriate age group were approached separately and the project was explained to them, in order to minimize the possibility of coercion by their mothers. In cases where the mother had more than one child within the age group interested in participating, one child was randomly selected to be interviewed with the full instrument (target child) and the other children were interviewed with a shortened version of the instrument. Only data from the target children’s interviews were utilized in this study. After the initial interview, families were randomly assigned to an experimental (n=40) group, who participated in the intervention, or a control group (n=40). All families were interviewed again four and eight months later. W Time 1 interviews were conducted immediately following recruitment. Initial attempts to schedule Time 2 and Time 3 interviews were made by telephone, with an emphasis on taking precautions to assure women’s safety. In cases where phone calls were insufl'rcient to schedule an interview appointment, subsequent tracking procedures were followed. Specifically, project representatives went in person to the home, contacted people that the women had provided to the project as alternative contacts to be used in the event that the women could not be located, and contacted school personnel. Interview times were arranged when it was most convenient and safe for the women and her children. The majority (90%) of the interviews took place in the women’s homes, 25 with the remainder of interviews conducted in safe places at the women’s discretion. Women were paid $15.00 for their initial intervieW, $50.00 for the post interview, and $75.00 for the follow-up interview. Children were paid (in their choice of cash or toys) $5, $10, and $20 respectively for each interview. The Intervention Families in the experimental group participated in a 16-week intervention program and worked with a trained female advocate. All interventions were individually tailored to meet the strengths and needs of each family but did share several components. Specifically, all advocates actively worked on (1) promoting families’ access to resources (2) developing or enhancing a positive, mentoring relationship between each child and a significant adult in his/her life, and (3) taking children to “The Learning Club,” a support and education group run by the project. It was hypothesized that working with an advocate would be beneficial to both ‘ mothers and children. The intervention was expected to facilitate improvements in children’s well-being, mother’s psychological well-being, and also be associated with reductions in violence. Thus, all analyses will include controls for intervention effects on children’s behaviors. Measures Measures used in this study included both preexisting measures and measures created specifically for this research. Some preexisting measures were adapted to be more appropriate for the study, often because the original length was prohibitive. All instruments were administered in interview format, except measures for emotional availability, and children behavior problems, which mothers completed in paper-and-pen 26 format. AME. To address the research questions pertaining to the effect of children’s exposure to domestic violence on their functioning, it is important to include the abuse that mother is experiencing with the assumption that children directly witnessed violence, overhear the violence, or are exposed to it aftermath. For this reason, mother’s report of the abuse she has experienced will be used as the measure of children’s exposure. Assailant’s psychological abuse of mothg. The assailants psychological abuse of mothers was assessed using a modified version of the Index of Psychological Abuse (Sullivan, Parisian, & Davidson, 1991) (e.g., How often has he refiised to talk to you?, How often has he accused you of having or wanting other sexual relationships?) Participants responded using a 4-point Likert-type scale. At pre, this 22-item scale had a reliability of .90 and the corrected item-total correlations ranged from .31 to .75. At post, this scale had a reliability of .91 and the corrected item-total correlations ranged fi'om .08 to .79. At the four-month follow-up, this scale had a reliability of .91 and the corrected item-total correlations ranged from .24 to .79. See Appendix A for this scale. ’s h ic buse of 0 er. To create the scale for mother’s experience of physical abuse, a composite score was created from a modified version of the Conflict Tactics Scale (CTS) (Straus, 1979) and mother’s experience of injuries. The CT S included seven yes/no items (e.g., Has he broken your glasses or torn your clothing?, Has he pushed, grabbed, shoved, or slapped you?) Mother’s experience of injuries was assessed through twelve yes/no items (e. g., Have you ever experienced soreness without bruises?, Have you ever experienced internal injuries?) Mother’s experience of abusive 27 tactics and injuries combined to create an index of physical abuse. At pre, this scale had a reliability of .86 and the corrected item-total correlations ranged from .10 to .77. At post, this scale had a reliability of .90 and the corrected item-total correlations ranged from -.O4 to .86. At follow-up, this scale had a reliability of .87 and the corrected item-total correlations ranged from .09 to .88. See Appendix B and Appendix C for physical abuse and injuries scales. Disruptions in led’s Life Several questions in the interview with mothers were used to assess the number of disruptions the children experienced in their lives in the prior four-month period. Three questions ascertained the number of moves the family experienced in the last four months, whether the family had resided in shelter, and the number of times the child changed schools. The items were combined to form a single scale of disruptions. At pre, this scale had a reliability of .56 and the corrected item-total correlations ranged fi'om .32 to .50. At post, this scale had a reliability of .49 and the corrected item-total correlations ranged fiorn .30 to .36. At follow-up, this scale had a reliability of .48 and the corrected item-total correlations ranged fiom .27 to .43. See Appendix D for these items. Physical Abge of Child To assess mother’s perception of the assailant’s physical abuse against the child, a modified version of the Conflict Tactics Scale (CTS) (Straus, 1979) was used. This scale included five yes/no items (e.g., Did the assailant every break child’s glasses or tear his/her clothing?, Did the assailant every punch, kick, choke or burn child?) From this scale, four items were combined (one item was never endorsed) at pre and resulted in the reliability being equal to .54 and the corrected item total correlations ranged from .32 to 28 .44. At post, this scale had a reliability equal to .77 and the corrected item total correlations ranged from .52 to .73. At post, this scale’s reliability was equal to .77 and the corrected item total correlations ranged fiom .52 to .73. At the four-month follow-up this scale’s reliability was equal to .81 and the corrected item total correlations ranged from .55 to .75. See Appendix E for these items. Children’s internalizing and externalizing behavior A scale was developed specifically for this study utilizing 19 items similar in structure to a sample of items from the Achenbach Child Behavior Checklist (CBCL) (Achenbach & Edelbrock, 1983). This scale contained two subscales, one to assess the internalizing behavior of the child and the other to assess externalizing behavior of the child. Only mothers completed this measure. Mothers responded using a Likert-type scale. At pre, the subscale for 10 items assessing children’s externalizing behaviors had a reliability of .80 and the corrected item-total correlations ranged from .35 to .62. At post, the reliability for the externalizing behaviors subscale was .85 and the corrected item-total correlations ranged fiom .34 to .69. At the four-month follow-up the reliability of the externalizing subscale was .83 and the corrected item-total correlations ranged from .39 to .66. At pre, the subscale for the 9 items assessing children’s internalizing behavior had a reliability of .75 and the corrected item-total correlations ranged from .18 to .55. At post, the reliability of the internalizing subscale was equal to .78 and the corrected item- total correlations ranged from .28 to .70. Finally, at the four-month follow-up, this scale’s reliability was .79 and the corrected item-total correlations ranged from .30 to .66. See Appendix F for the items assessing children’s externalizing and internalizing 29 behaviors. Analysis Strategy Univariate statistics (descriptive statistics, histograms, frequency distribution and examination of outliers) were conducted for all variables. To respond to the three research questions examined in this study, multilevel modeling (MLM) was used. This method estimates individual trajectories or change over time nested within individuals (level 1) and models between-subj ects-variation of individual change curves (level 2) with treatment factors, individual attributes, and circumstances that vary across time. This analysis strategy was appropriate because of its ability to address cross-level interactions and its flexibility in handling repeated measures data. In addition, MLM is flexible in dealing with situations where there may be subjects with incomplete data. To examine the difl‘erence between models, the likelihood ratio (LR) chi-square test was utilized. This test uses the deviance values of two models, one nested within the other, to examine whether additional variance in the outcome variable was explained by the inclusion of other explanatory factors in the model. The deviance values are computed by multiplying the log likelihood by —2. The difference in deviances (D0 — D1) is the test statistic, x2, and the degrees of freedom are computed by calculating the difference in the number of parameters in the two models (Snijders, 1996). Because maximum likelihood (ML) methods were used for estimation, LR x2 could be used to test the significance of both random and fixed effects . All estimation was done with IVIIXREG multilevel analysis software (Hedeker, D. R., & Gibbons, R. D. (1996). All tests of statistical significance utilized a one tailed test. This was decided based on two factors. First, and most importantly, all proposed hypotheses were 30 directional. Second, the use of a one-tailed test increased the ability to detect statistical significance in a relatively small size (N=80). Effect pf Mother’s Experience pf Abuse on Children’s Functioning For the purposes of the current study, analyses for each dependent variable were conducted in two stages. The purpose of the first stage was to establish the basic shape of the individual trajectories or regression equations over time. In the second stage, the hypothesized explanatory variables were added to the model. In stage one, the shape of the data was established through three steps. The initial step of stage one was to estimate a baseline model or a model without explanatory variables. This model, Model 1, included only the effects of the random intercept. The purpose of the baseline model was to reveal the basic partitioning of the variance between the two levels (i.e, level 1 or variance due to changes over time nested within individuals and level 2 or variance due to differences across individuals) without explanatory variables. The baseline model provides two valuable sources of information. First, the baseline model illuminates whether there is sufficient variability at both levels (i.e., level 1 and level 2) to justify multi-level modeling. Second, if there is sufficient variability, the baseline model provides a point of comparison to determine the explanatory power of subsequent variables added to the model. The second step of stage one was to estimate the best level 1 model that explained variance within individuals across measurement occasions using random intercept plus slope effects. This model, Model 2, resulted in a linear regression line for the three measurement occasions for each individual in the analyses (i.e., approximately 80 regression lines). In this step, time was added as a random variable, meaning that the 31 modeled trajectories of children’s scores on the outcome variable were allowed to reflect different rates of change over time. If time was not significant as a random variable, then it was added as a fixed variable, meaning that children’s scores change, but change at a fixed or similar rate over across individuals. The third step, presented in Model 3, was to ascertain whether there was an acceleration or deceleration in the rate of change on children’s outcome variables over time. It should be noted that for all analyses, time was coded such that Time 1 (pre) was defined as the intercept. The purpose of the second stage was to examine variability in the regression coefficients across the 80 individuals and the explanatory value of predictors such as exposure to abuse against mothers on children’s outcomes (internalizing and externalizing behaviors). To that end, a series of models including fixed and time- varying predictors was estimated. First, Model 4, containing only the control variables (children’s experience of abuse during the three periods of data collection and experimental assignment to intervention condition), was estimated. The next two models estimated, Models 5 and 6, included the control variables as well as the time-varying effects for psychological and physical abuse, respectively. The analyses outlined above were conducted for children’s outcome variables (internalizing and externalizing behaviors) as well as the proposed mediating variable, disruptions in children’s lives. The test for mediating relationships requires that the mediating variable be examined both as an explanatory variable and an outcome variable. This process and the process for examining moderating effects are described in detail in the following sections. 32 The Moderating Effect of Gender After the main effects were assessed, the moderating effect of gender on the relationship between exposure to abuse against mothers and children’s behaviors was examined. Using MLM, Models 7 and 8 included the main effects of abuse (psychological or physical), gender, and the interaction between abuse and gender. Guidelines outlined by Barron and Kenney (1986) were applied to the MLM analyses to ascertain whether gender had a significant moderating effect on the relationship between children’s exposure to domestic violence and children’s positive functioning. Specifically, a moderating effect can be represented as an interaction between an explanatory variable and a factor that specifies the conditions for its effect. In the event that the relationship between exposure and children’s behaviors did significantly depend on gender, further analyses were conducted to investigate how gender affected this relationship. To facilitate interpretation of the interaction effect, variables assessing physical and psychological abuse against mothers were grand mean centered prior to analysis (Aiken & West, 1991). The Mediating Effect of Disgrptiops in Children’s Lives Guidelines outlined by Baron and Kenny (1986) were applied to ascertain whether disruptions in children’s lives had a significant mediating effect on the relationship between children’s exposure to domestic violence and children’s behaviors. Three relationships are required to establish the presence of a mediating relationship. In the current study, a) exposure to mother’s abuse must significantly predict disruptions in the child’s life, b) exposure to mother’s abuse must significantly predict children’s behaviors, and c) when the effect of disruptions in child’s life is controlled, the previous relationship 33 between exposure to mother’s abuse and child’s behaviors is no longer significant or (for partial mediation) is reduced in magnitude. To examine the third criterion, Models 9 (for psychological abuse) and 10 (for physical abuse) were generated. The process outlined above was conducted for both outcome variables. Each model was labeled to discern both the step and outcome variable it which it was referring. For example, Model l-I described the results for the random intercept model (Model I) for internalizing behaviors as the outcome (indicated by I), Model l-E described the results for the random intercept model (Model 1) for externalizing behaviors as the outcome (indicated by E), and Model l-D described the results for the random intercept model (Model 1) for disruption’s in children’s lives as the outcome (indicated by D). 34 Results Retention over time The rate of retention of women in this study was high. At all time points over 90% of the sample were interviewed. Specifically, of the eighty women that were interviewed at time one, 92.5% of women were interviewed at time two, and 95% of women were interviewed at time three. Women’s Expm‘gpce pf Abuse P h 'c buse Women experienced a great deal of psychological abuse in the four months preceding the first interview. Eighty-eight percent of women said that the assailant criticized their family or fiiends, 86% said the assailant called them names, and 71% said the assailant had tried to control their activities in the four month preceding the interview. 9, 6‘ (Percentages were generated by collapsing the response options “rarely, sometimes,” and “often”). Women experienced less abuse across time. At the post interview, 72% of women said the assailant criticized their family or fiiends, 63% said the assailant called them names, and 53% said the assailant had tried to control their activities in the four month preceding the interview at time one. At the follow-up interview, 59% of women said the assailant criticized their family or fiiends, 51% said the assailant called them names, and 47% said the assailant had tried to control their activities in the four month preceding the interview at time one. See Table l for means and standard deviations of variables across time, and Table 2 for a list of the types of violence that women experienced during pre, post, and follow-up. 35 Insert Tables 1 and 2 about here Phjpipal Abuse In the four months prior to their initial interview, women experienced a considerable amount of violence from the assailant. The majority of women (65%) participating in the project reported having been pushed, grabbed, shoved, or slapped by their assailants. Twenty-nine percent of women reported having an object thrown at them or having been hit with an object. Over a quarter of women reported having sexual activity forced upon them, and 18% reported being threatened with a weapon or being assaulted with a weapon. Over time, women’s experience of abuse decreased. At the post interview, 32% of women reported having been pushed, grabbed, shoved, or slapped by their assailants in the previous four months, while 20% reported having an object thrown at them or having been hit with an object. At the follow-up interview, 20% of women reported having an object thrown at them or having been hit with an object and 18% reported forces sexual activity. Table 3 lists the type of violence women experienced during pre, post, and follow-up. Insert Table 3 about here Women also reported experiencing substantial injuries in the 4-month period prior to their initial interview. Forty-three percent of women reported having had soreness without bruises and 41% reported cuts, scrapes, bruises, and red marks at the pre 36 interview. Women’s experience of injuries decreased at the post interview and the 4- month follow-up with 27% and 15% of women reporting soreness without bruises respectively. Table 4 includes a list of the physical injuries that women reported at all three time points. Insert Table 4 about here W Children experienced a substantial amount of disruptions in their lives. Referring to the four-month period prior to their interview at pre, 34% of mothers reported that their children had changed schools at least once and 59% of women reported being a shelter resident. Additionally, 73% of women reported moving their residence at the pre, with 25% moving more than two times. Over time, families experienced fewer disruptions. Twenty percent of mothers reported that their children had changed schools during the post interview and 26% reported their children had changed schools during four month follow-up. At post, 18% of women reported they had been a shelter resident, and 10% reported they had been a shelter resident during the four-month follow-up. See Table 5 for disruptions in children’s lives over time. Insert Table 5 about here Children’s expg'ence of abye by assailant According to mother’s report of abuse the child experience by the assailant, 37 children experienced more abuse at the pre interview than at the post and follow-up interviews. Fifty-three percent of mothers reported that their children were pushed, grabbed, shoved or slapped by their assailant. Twenty-one percent were hit with an object and 8% were punched, kicked, choked, or burned. At the post interview and follow-up interview respectively, 6% and 7% of mothers reported their children were punched, kicked, choked, or burned. See Table 6 for children’s experience of abuse by the assailant. Insert Table 6 about here Intmalizing Bemviprs sme pf the data over m e As the first step, the baseline model, Model H (internalizing), was generated. This model included only the random intercept. The results of Model H indicated that the intercept or grand mean across all three time points was significantly different from zero. The relatively large coefficient for the random effects variance, which at .096 was nearly five times the magnitude of its standard error (.020), indicated substantial variability in level 2 or the level of internalizing across individuals (i.e., some children had higher levels of internalizing behaviors while others had lower levels of internalizing behaviors). Further, the intraclass correlation at .571 indicated that a majority of the variance in internalizing scores was attributable to differences between individual children, although a substantial portion (43%) was due to within-individual variability. In Model I-2, time was added as a random variable to examine the linear effect of 38 time. See Table 6 for Models I-1 and I-2. The addition of the random effect of time to the model containing only the random intercept (Model H) did not result in a significant improvement in the explained variance in children’s internalizing behaviors. In Model I- 2, the coefficient for time was significant ([3=-.012, l-tailed p<.05) but the model offered only a marginally significant improvement in the prediction of children’s scores on internalizing behaviors as indicated by the log likelihood test, x2(3) = 5.946, p=.055. To test whether time could be represented as a fixed term, Model I-2 was then estimated with time as a fixed variable, not a random variable. In this model, Model I- 2a, the coefficient for time was significant (B=—.012, l-tailed p < .05) and the model significantly improved the prediction of children’s scores on internalizing behaviors as indicated by the log likelihood test, x2(1) = 4.766, p < .05. Furthermore, the variance explained with time as a fixed efi'ect was not significantly less than explained with a random effect for time (LR £0) = 1.18, p =.28). The negative sign of this coefficient indicated that internalizing scores declined over time. The adequate modeling of time as a fixed effect indicated that, although children’s scores on internalizing behaviors declined significantly over time, the change was at a similar rate across children. Thus, ' subsequent analyses were conducted using Model I-2a, with time as a fixed variable. For Models I—l, 1-2, I-2a see Table 7. The amount of variance accounted for by the baseline model with random intercept and fixed linear time can be computed using the intercept variance and residual variance of Model 1 [intercept variance/(intercept variance + residual variance)] (Snijders, 1996). The amount of variance explained by the random intercept term (Model H) was 57.09%. Estimates of the amount of variance explained by subsequent models 39 can be generated by dividing the residual variance in the comparison model by the total variance in the original baseline model (intercept variance + residual variance of Model I- 1). This number is subtracted from one to get the amount of variance explained by the comparison model. Using this method, Model 2a, which adds a fixed effect for time, accounts for 58.58% of the variance in children’s score on children’s internalizing behaviors. To ascertain whether there was an acceleration or deceleration in the rate of change on children’s scores on internalizing behaviors, a quadratic (squared) effect for time was added in the next step. This model, Model I-3, did not result in a significant improvement in the amount of variance explained by the model. Neither the coefficient for time squared, nor the likelihood ratio test was significant. See Table 6 for Model I-3. Therefore, the quadratic was dropped from all subsequent analyses on children’s internalizing behaviors. To summarize, based on these analyses, there was no evidence of a deceleration or acceleration in the rate of change over time. Thus, it appears that children’s scores on internalizing behaviors declined significantly over time, that the change did not vary significantly across children, and that the rate of change was constant across the three time points. Insert Table 7 about here To statistically remove the effects of potentially confounding variables, the control variables were added in Model I-4. These included children’s experience of abuse during the three periods of data collection (a time-varying predictor) and the effect of 40 experimental assignment to intervention condition over time. (Children had received no intervention at pre. Thus, a condition effect on baseline internalizing scores not included in the model). For Model I-4, see Table 8. The addition of the control variables did not significantly contribute to the variability explained significantly beyond the linear model, L.R. x2(2) = .610, p =. 13. However, because these variables were conceptualized as control variables, all subsequent analyses for children’s internalizing behaviors included these variables. In the next set of models, the explanatory variables were added. First, mother’s experience of psychological abuse was added to Model I-4. As illustrated by Model I-5, the addition of mother’s experience of psychological abuse was positive and significant; indicating that, across time, higher scores for children’s internalizing were associated with higher levels of mothers’ experience of psychological abuse. Both the coemcient for mother’s experience of psychological abuse, [5 = .120, p < .001, and the log likelihood test were significant, x20) = 25.256, p < .001. Further, the addition of psychological abuse into the model increased the percentage of variance accounted for fi'om 58.74% to 63.1%. The results of Model I-5 are presented in Table 8. In Model I-6, another explanatory variable, mother’s experience of physical abuse, was added to the model with the control variables (Model 14). In this model, both the coefficient for mother’s experience of physical abuse, [3 = .092, l-tailed p < .001, and the likelihood ratio test were significant, x2( 1) = 12.918, p < .001 . Compared with Model 1-4 (model with control variables), the addition of mother’s experience of physical abuse increased the percentage of variance accounted for from 58.74% to 66.3%. The positive sign of the coefficient indicated that, across time, higher scores for children’s 41 internalizing were associated with higher levels of mothers’ experience of physical abuse. See Model I-6 presented in Table 8 for the estimates for the relationship between mother’s experience of physical abuse and children’s internalizing behaviors. Insert Table 8 about here To summarize, results fi'om this set of modeling indicated that there was no evidence that the two proposed control variables, children’ experience of physical abuse by the assailant and the effect of the condition over time, had an effect on children’s internalizing behaviors. However, there were main effects for both of the explanatory variables. Mother’s experience of psychological abuse and physical abuse were both positively associated with children’s internalizing behaviors and explained a significant amount of variance in children’s internalizing scores over time. Next, the proposed moderating effect of gender was examined. Mum According to Barron and Kenny (1986), a moderating effect can be represented as an interaction between an explanatory variable and a factor that specifies the conditions for its effect. Separate analyses were conducted for the moderating impact of gender on the relationship between mother’s experience of psychological abuse and the child’s internalizing behaviors and the relationship between physical abuse and the child’s internalizing behaviors. To test for the existence of gender as a moderator of the impact of mother’s experience of psychological abuse on the child’s internalizing behaviors, child’s gender 42 and the interaction between gender and mother’s experience of psychological abuse were added to Model I-5. Results of this analysis are presented in Model L7. The moderator hypothesis was not supported. Neither the coefficient for the interaction term between gender and mother’s experience of psychological abuse ([3 = -.029, l-tailed p = .27) nor the likelihood ratio test was significant, 260) = .504, p =.39. See Table 9 for Model L7. The results from the analysis examining the moderating impact of child’s gender on the relationship between mother’s experience of physical abuse and children’s internalizing behaviors followed a similar pattern. Model I-8 presents the results of the addition gender and the interaction between child’s gender and mother’s experience of physical abuse were added to Model I-6. Neither the coefficient for the interaction term between gender and mother’s experience of physical abuse ([3 = -.O65, 1-tailed p = .16) nor the likelihood ratio test were significant, x2(2) = 2.110, p =.18. See Table 9 for Model I-8. Insert Table 9 about here These results indicate that there was no evidence to support the hypothesis that gender moderated the relationship between children’s exposure to psychological and physical abuse against mother and children’ internalizing behaviors. Next, disruptions in the child’s life will be examined as a mediator of the effects of children’s exposure to abuse against the mother on children’s internalizing behaviors 'a ' ect: ' ti ' chi ’ life. All previous analyses have only included children’s internalizing behaviors as the 43 outcome variables. Because the test for mediation requires disruptions in child’s life as the outcome variable in order to test relationship a, the first step for these analyses was to establish a model for disruptions over time. To that end, the baseline model for disruptions, Model D-l, was generated. This model included only the random intercept. The results of Model D-l indicated that the intercept was significantly different from zero (B = .636, l-tailed p < .001. The relatively large coefficient for the random effects variance, which at .131 was more than three times the magnitude of its standard error (.041), indicated substantial variability in level of disruptions across individuals (i.e., some children had higher levels of disruption while others had lower levels of disruptions. The intraclass correlation, equal to.28, indicated that some between- individuals effects were present, although the majority of the variance was related to within-individual change over time. In Model D-2, the linear effect of time was added as a random explanatory variable. For this model, coefficients for both the random intercept (B = .890, l-tailed p < .001) and random time effects (B = -.065, l-tailed p < .001) were significant. This model was significantly more predictive of disruptions than Model D-l, x2(3) = 35.508, p <00] and accounted of a total of 75.07% of the variance in disruptions in children’s lives. The negative coefficient for time indicated that, on average, disruptions declined across time. In Model D-3, the quadratic effect of time was added in order to test the explanatory significance of acceleration or deceleration in the slope of disruptions over time. The quadratic effect of time was originally added as a random explanatory variable; however, due to missing observations for a few individuals and to the narrow range of possible scores on the dependent variable, the quadratic could not be estimated as a random effect. Thus, the quadratic was added as a fixed explanatory variable. This model was a significant improvement over Model D-2. For this model, coefficients for the intercept (B = .949, l-tailed p < .001), time (B = -.160, l-tailed p < .001), and time squared (B = .012, l-tailed p < .05) were significant. This model was significantly more predictive of disruptions than Model D—2, L. R. x20) = 7.828, p < .05 and accounted for a total of 77.35% of the variance in disruptions in children’s lives. See Table 10 for Models D-l, D-2, D-3. Although the addition of time squared as a fixed variable did significantly increase the amount of variance explained, this increase was small (2.28%). The inclusion of time as a quadratic variable increased the complexity of the model and introduced substantial multicollinearity that could cause model instability and minimize the explanatory contribution of other variables. Given that the coefficient for time squared and the increase in percent of variance accounted for was small relative to the magnitude of the effect of linear time, and the lack of specific hypotheses involving deceleration, a decision was made to conduct further analyses without time squared in the model. To summarize the basic model of disruptions over time, results indicated that the level of disruptions in children’s lives changed significantly over time. On average, the change reflected a general decline, although there was significant variability in individual trajectories. These analyses suggested that there was a deceleration in the rate of change over time, although it was small relative to the magnitude of the general decline. In light of the complexity and multicollinearity introduced by the quadratic term, it was decided that further analysis would model change in disruptions as a linear function of time. 45 Insert Table 10 about here Next, the control variables were added to Model D-2. As illustrated in Table 11, with the addition of the control variables (child’s experience of abuse across time and the interaction between condition and time, reflecting the impact of experimental condition on disruptions at Times 2 and 3), Model D-4 was significantly more predictive than Model D-2, x2(2) = 8.596, p <.05. Using a one tailed test, the control variables condition by time (B = -.028, l-tailed p < .05) and abuse that the child experienced (B = .143, 1- tailed p < .05) were both significant. These coefficients indicate that across time, children in the control condition and children who experienced higher levels of abuse had higher levels of internalizing behaviors. Psychplpgjcal abuse. To test the mediating relationship of disruptions in children’s lives between mother’s experience of psychological abuse and children’s internalizing behaviors, mother’s experience of psychological abuse must significantly predict disruptions in children’s lives (Barron and Kenny’s (1986) criterion a). Analyses presented in Model D-5 indicated that mother’s experience of psychological abuse was significantly related to disruptions in children’s lives (B=. 130, 1-tailed p<.01). Further, the likelihood ratio test was significant, 260) = 12.112, p <.001. See Table 11 for the results of Model D-5. As previously noted in Model I-5, there was a significant relationship between psychological abuse and children’s internalizing behaviors (B=.120, l-tailed p<.001), supporting Barron and Kenny’s criterion b of a relationship between the predictor and the outcome variable. 46 Insert Table 11 about here Given that all of the direct pathways between predictor and mediator and predictor and outcome were significant, the next step was to add disruptions to child’s life to the model predicting children’s internalizing scores fi'om mother’s experience of psychological abuse (Barron & Kenny, 1986). Results presented in Model I-9 did not support the hypothesis that disruptions in child’s life would mediate the relationship between psychological abuse and children’s internalizing behaviors. The relationship between mother’s experience of psychological abuse and children’s internalizing behaviors remained significant (B=.115, l-tailed p<.001) and nearly identical in magnitude to the relationship without the efi‘ect of disruptions in the model. See Table 12 for the comparison of Model I-5 and Model I-9. Therefore, it appears that the relationship between mother’s experience of psychological abuse and children’s internalizing behaviors is not mediated by the number of disruptions in the child’s life. Insert Table 12 about here Ehfiigal abuse. To test the mediating relationship of disruptions in children’s lives between mother’s experience of physical abuse and children’s internalizing behaviors, mother’s experience of physical abuse must significantly predict disruptions in children’s lives (Barron and Kenny’s (1986) criterion a). To test the relationship between exposure to physical abuse and disruptions in children’s lives, mother’s experience of 47 physical abuse was added to Model D4 (D4 contained only control variables). Although the coefficient for mother’s experience of physical abuse was only approaching significance in Model D-6 (B=.067, l-tailed p= .068), according to the likelihood ratio test this model was a significant improvement over Model D-4, x2(1)=2.832, p< .05. See Table 11 for results of Model D-4 and Model D—6. As previously noted in Model I-6, physical abuse was a significant predictor of children’s internalizing behaviors. The next step was to add disruptions in children’s lives to Model L6 to see if the magnitude of the effect of exposure to mother’s physical abuse on children’s internalizing behaviors is reduced. As shown in Model I-lO, the addition of disruptions in child’s life did not affect the relationship between exposure to mother’s physical abuse and children’s internalizing behaviors. The coefficient for mother’s experience of physical abuse remained significant and its magnitude remained nearly identical with disruptions added to the model. See Table 13 for Model I-6 and I- 10. Therefor, there was no evidence that disruptions in children’s lives mediated the relationship between exposure to physical abuse against mother and children’s internalizing behaviors. Insert Table 13 about here To summarize, both variables measuring children’s exposure to abuse -— mother’s experience of psychological abuse and experience of physical abuse -- were significantly related to the child’s internalizing behaviors and to disruptions in the child’s life. However, the efi‘ects of psychological and physical abuse on child internalizing remained 48 3429‘7"" 3;!2'" significant after the proposed mediating variable (disruptions in child’s life) was added to the model. Thus, analyses on children’s internalizing behaviors indicated that although there were significant effects of mother’s experience of psychological and physical abuse on children’s internalizing behaviors, there was no evidence that a) gender moderated this relationship, or b) disruptions in child’s life mediated this relationship. Extemalizing Behaviors Shape pf the date pver time The analysis of externalizing behaviors followed a process similar to internalizing behaviors. The baseline model was generated first. The results of Model E-l indicated that the random intercept or grand mean across all three time points was significantly different fi'om zero (B=.775, l-tailed p<.001). The relatively large coefficient for the random effects variance, which at .131 was more than five times the magnitude of its standard error (.025) indicated substantial variability in the level of externalizing behaviors across individuals (i.e., some children had higher level of externalizing behaviors while others had lower levels of externalizing behaviors. Further, the intraclass correlation was equal to .655 indicating that a substantial portion of the variability was between individuals, although a sizeable portion (34%) reflected within-individual change over time. In the next model, Model E-2, time was added as a random variable in order to examine the linear effect of time. According to the log likelihood test, Model E-2 provided a significant improvement in the variance in externalizing behaviors over Model 18, LR x2(3)=12.402, l-tailed p<01. However, the s10pe variance parameter was equal to -.016 which was more than twice its standard error, thus, this model was run again using 49 time as a fixed variable. Like Model E-2, Model E-Za was a significant improvement over the model with only the random intercept, x2(1)=8.534, p<.01, and further, there was not a significant difference between Model B-2 and E-2a x2(2)=3.868, p=.07. Because the coemcient for random variance of time was very small and there was not significant explanatory value added by allowing time to vary randomly, the remaining analyses were conducted with Model E-2a, which included time as fixed variable. This model accounted for 67.63% in the variability in children’ score on externalizing behaviors. The negative sign of this coefficient indicates the externalizing behaviors declined over time. Thus, children’s score on externalizing behaviors decreased over time, but decreased at similar rates. See Table 14 for Models E-l, B2, and E-2. Insert Table 14 about here Next, a quadratic (squared) effect for time was added to Model E-2. This term was added to see if there was a deceleration in the rate of change of children’s externalizing behaviors over time. The addition of the quadratic as either a random or fixed effect did not result in a significant improvement in the amount of variance explained by the model or a significant coefficient for time squared. As a result, time squared was dropped from all other analyses. See Table 14 for Model E-3. This resulted indicated that on average, children’s scores on externalizing declined over time, but their individual trajectories declined at similar rates. In Model E-4, the control variables, children’s experience of abuse and the effect of condition over time, were added to Model E-2. See Table 15 for the results of Model 50 E-4. With the addition of the control variables, the model explained 68.23% of the variance and did not significantly contribute to the variability explained beyond Model B- 2a, x2(2)=3.614, p=.08. Again, because they were conceptualized as control variables, children’s experience of abuse from the assailant and condition over time remained in all subsequent analyses. The explanatory models were added in the next set of models starting with mother’s experience of psychological abuse. In Model E-5, the addition of mother’s experience of psychological abuse did not result in significantly more variance explained by the model x2(l)=.876, p=.l7 and the coefficient for psychological abuse against mother was not significant (B=.O36, l-tailed p=.08). See Table 15 for Model E—S. In Model E-6, mother’s experience of physical abuse was added to the model with the control variables, Model E4. The results of Model E-6 indicated that the coefficient for physical abuse (B=.043, l-tailed p<.05) was significant, while the likelihood ratio test, x2(1)=2.662 was very nearly significant at p = .05. The addition of mother’s experience of physical abuse was positive and significant; indicating that, across time, higher score for children’s externalizing behaviors were associated with higher levels of mothers’ experience of physical abuse. Further, the addition of physical abuse to Model E-4 increased the variance accounted for by 6.8%. See Table 15 for Model E-6. Insert Table 15 about here Results from these analyses indicated that exposure to psychological abuse did not appear to be significantly related to children’s externalizing behaviors. However, the 51 relationship between exposure to physical abuse and children’s externalizing behaviors was significant by at least one criterion (the test of the coefficient) and very nearly significant by the other. In the next set of analyses, the moderating effect of gender was examined. Mpderating effect of gender The moderating effect of gender was examined according to Barron and Kenney (1986) as described in the previous section of moderator analysis. Separate models were generated for gender’s effect on the relationship between mother’s experience of psychological abuse and children’s externalizing behaviors and the relationship between mother’s experience of physical abuse and children’s externalizing behaviors. To test the moderating effect of gender on the relationship between exposure to psychological abuse and children’s externalizing behaviors, gender and the interaction between gender and psychological abuse were added to Model E-S. For results of Model E-7 see Table 16 There was a significant increase in the amount of variance explained by Model E-7 according to the log likelihood test, x2(2)=4.922, p<.05. Also, the coefficient for the interaction between mother’s experience of psychological abuse and gender was significant, (B=-.091, 1-tailed p<.05), indicating the moderating effect of gender on the relationship between psychological abuse and children’s externalizing behaviors. This model explained 67.19% of the variability in children’s externalizing behaviors. To interpret the nature of the interaction, simple slopes for the effect of psychological abuse on externalizing were computed within gender, as outlined by Aiken and West (1991). For boys, the coefficient was .088, l-tailed p < .05, indicating that boys’ externalizing behaviors were associated with higher levels of mother’s experience of psychological 52 abuse. For girls, the coefficient was .0005, l-tailed p = .37, indicating that there was essentially no relationship between exposure to psychological abuse and girls’ externalizing behaviors. Insert Table 16 about here To test the moderating impact of gender on the relationship between children’s exposure to physical abuse and children’s externalizing behaviors, gender and the interaction between gender and physical abuse were added to Model E6. The results of Model E-8, presented in Table 17, indicate that gender did appear to moderate the relationship between children’s exposure to physical abuse and their externalizing behaviors. The log likelihood test indicated that Model E-8 explained significantly more variance than Model E-6, x2(2)=5.040, p<.05 and explains 69.19% of the variance. For boys, the coefficient was .104 which indicates that the effect of exposure to physical abuse on boys’ externalizing behaviors was significantly positive; indicating that boys’ externalizing behaviors were associated with higher levels of mother’s experience of physical abuse. For girls, the coefficient was -.027, but not significant; indicating that there was not a significant relationship between exposure to psychological abuse and girls’ externalizing behaviors. Insert Table 17 about here To summarize, gender appeared to moderate the relationship between exposure to 53 abuse against mother and children externalizing behaviors. The interaction terms between mother’s experience of psychological abuse and gender and experience of physical abuse were significant indicating that gender moderates the relationship between exposure to domestic violence and children’s externalizing behaviors. For boys, increased exposure to psychological and physical abuse against mother was related to increased externalizing behaviors. However, there was no evidence of a relationship between exposure to psychological and physical abuse against mother and scores on externalizing behaviors for girls. Mediating effect: Disruptions in child’s 11 fe. WM. To test the mediating relationship between disruptions in child’s life, the process outlined previously for the outcome variable internalizing behaviors was followed. As noted in Model E-5, the mother’s experience of psychological abuse was not a significant contributor to the amount of variance explained in externalizing behaviors among the entire sample of children. Since this relationship was not significant, there was no evidence that disruptions in children’s lives mediates the relationship between psychological abuse and children’s externalizing behaviors. There was a significant relationship between psychological abuse and externalizing behaviors for boys. According to Barron and Kenny (1986), the mediating relationship could be examine using the sub-sample of boys to see if there was moderated mediation. However, in this case the p for boys was equal to 36 which was too small to conduct further analyses. Ehfiisal abuse. For the mediating effect of disruptions in children’s lives on the relationship between exposure to physical abuse against mother and children’s 54 externalizing behaviors, three relationships were examined. First, as previously noted, physical abuse was significant in the explanation of variability in children’s externalizing behaviors. See Table 15 for Model E-6 and these results. Second, also noted previously, mother’s experience of physical abuse was a significant contributor to the variability in the disruptions in children’s lives. See Table llfor Model D6. The third step in this process was to see if the relationship between exposure to mother’s abuse and externalizing behaviors remained significant or declined in magnitude after the disruption in child’s life was added to the model. In Model E-lO, disruptions in children’s life was added to Model E6. The coefficient for mother’s experience of abuse declined slightly in magnitude, fi'om .045 to .041, when the mediator was added to the model; in the presence of the mediator, the coefficient for mother’s experience of abuse was just beyond one-tailed significance (B=.041, l-tailed p= .052). The results of these analyses provide evidence that disruptions in children’s lives partially mediated the impact of exposure to abuse against mother on children’s externalizing behaviors. Insert Table 18 about here In sum, the results of the analyses on children’s externalizing behaviors indicated that children scores vary at the intercept (i. e., some children had higher levels of externalizing behaviors while others had lower levels of internalizing behaviors) and change, at similar rates, over time. After controlling for the effects of children’s experience of physical abuse fi‘om the assailant and assignment to intervention condition 55 over time, there was no evidence of an overall effect for exposure to psychological abuse. However, there was evidence that the effect of psychological abuse was moderated by child’s gender; for boys, exposure to higher levels of psychological abuse against mother was associated with higher scores on externalizing behaviors. On the other hand, exposure to physical abuse had an overall effect on children’s externalizing behaviors; this effect also appeared to be moderated by child’s gender in that boys who were exposed to higher levels of physical abuse had higher scores on externalizing behaviors. Further, the relationship between exposure to physical abuse and children’s externalizing behaviors was mediated by disruptions in children’s lives. Analjgis 5mm These analyses indicated that mothers in the project experienced a considerable amount of violence, and that exposure to that violence significantly impacted aspects of children’s behavior. Specifically, results demonstrate that some children exhibited higher levels of internalizing behaviors while other exhibited lower levels of internalizing behaviors, and these scores declined, at a constant rate, over time. After removing the effects of potentially confounding variables (children’s experiences of abuse and the effect of experimental assignment to intervention condition over time), there was evidence that mother’s experience of abuse, both psychological and physical, impacted children’s internalizing scores over time. Specifically, higher score for children’s internalizing behaviors were related to higher levels of mother’s experience of psychological and physical abuse. However, no evidence was found to support the hypotheses a) child’s gender moderated the relationship between exposure to abuse either psychological or physical, and children’s internalizing scores, or b) disruptions in child’s 56 life mediated this relationship. Children’s scores on externalizing behaviors also varied across individuals. Further, children’s scores on externalizing behaviors decreased, at a consistent rate, over time. After removing the effects of children’s experiences of abuse and the effect of experimental assignment to intervention condition over time, the results demonstrated that higher scores for children’s externalizing behaviors were associated with higher levels of mothers experience of physical abuse across time. Although there was no direct main effect of psychological abuse on children’s externalizing behaviors, there was evidence that gender moderated this relationship. Specifically, boys’ externalizing behaviors were associated with higher levels of mother’s experience of psychological abuse. For girls, scores on externalizing behaviors were not related to mother’s experience of psychological abuse. Child’s gender also moderated the relationship between exposure to physical abuse against mother and children’s externalizing behaviors. For boys, higher levels of externalizing behaviors were associated with higher levels of mother’s experience of physical abuse while for girls there was no relationship between externalizing behaviors and exposure to physical abuse against mother. Also, the results indicated that the relationship between exposure to physical abuse against mother and externalizing behaviors was partially mediated by the number of disruptions in children’s lives. 57 Discussion The findings fiom this study support previous research suggesting that exposure to domestic violence has a negative impact on children’s behavior. Specifically, children’s exposure to the psychological and physical abuse of their mothers influenced the degree to which they exhibited internalizing and externalizing behaviors. However, it is important to note that the degree to which children exhibited these behaviors declined over time. In addition to confirming previous research regarding the negative effects of exposure to domestic violence, this study also expands our current understanding of how and when children are affected by domestic violence. First, the findings fi'om this study suggest that internalizing and externalizing behaviors are affected by exposure to domestic violence through different processes. That is, there are direct relationships between exposure to psychological and physical abuse and children’s levels of internalizing behaviors while the relationships between exposure to psychological and physical abuse and children’s externalizing behaviors are more complex. In particular, disruptions in children’s lives, such as household moves and switching schools, partially explained the relationship between exposure to abuse and the degree to which children exhibited externalizing behaviors. Second, the influence of children’s exposure to abuse is dependent on gender. Both boys and girls were more likely to exhibit internalizing behaviors when exposed to psychological abuse, but only boys were more likely to exhibit externalizing behaviors when exposed to psychological or physical abuse, while girls’ externalizing behaviors were not affected by exposure to either form of abuse. 58 Children’s Behaviors met Time The findings from this study show that there was variability in children’s problematic behaviors and that, in general, these behaviors declined over time. However, there was not variability across children in the rate of decline and, in addition, the rate of change remained constant over time. In other words, children’s scores on internalizing and externalizing behaviors appeared to decline at similar, and unchanging rates over time. This finding is consistent with the few previous studies that have followed children exposed to domestic violence longitudinally. These studies have found that time away from exposure lessens the negative effects of that exposure (Holden, Stein, Ritchie, Harris, & Jouriles, 1998; Wolfe, et al., 1986). For example, Holden and colleagues noted that time away from a violent home was associated with decreases in children’s problem behaviors in the six-month period after leaving the batter. These findings suggest that while exposure to abuse is often harmful to children, time free fiom witnessing abuse against their mothers is a factor which contributes to children’s resiliency over time. Holden’s study and the findings fi'orn this study provide important information regarding concerns about the long-term consequences of witnessing abuse and provide some evidence of children’s resiliency over time. Exmsure to Psychological Abuse Agaigfi Mpther As illustrated, there is evidence that time may ameliorate the negative effects of exposure to domestic violence on children. However, there are other factors that contribute to children’s outcomes. One of these factors is the level of psychological abuse to which the child is exposed. The findings from this study demonstrate that exposure to psychological abuse influences both the degree to which children exhibit 59 internalizing, and for boys, externalizing behaviors. In particular, the level of psychological abuse mothers experienced was positively related to children’s level of internalizing behaviors. Children exposed to higher levels of psychological abuse exhibited higher levels of internalizing behaviors. Interestingly, there was no evidence that exposure to psychological abuse affected boy’s and girl’s level of internalizing behaviors differently. Regardless of their gender, children’s internalizing behaviors were influenced by their level of exposure to psychological against their mothers. The impact of exposure to psychological abuse has been often overlooked in the literature. However, the few studies that have examined the relationship between psychological abuse and children’s functioning have found that exposure to psychological abuse has a negative effect on children’s functioning (F antuzzo, et al., 1991; Jouriles, et al., 1996; Levendosky & Graharn-Bermann, 1998), and this study provides some confirmation for these findings. An interesting contribution of the current study is the finding that exposure to psychological abuse affects the externalizing behaviors of boys and girls differently. Unlike the findings for internalizing behaviors, children’s externalizing behaviors were not uniformly affected by exposure to psychological abuse. These findings showed that exposure to psychological abuse of their mothers influenced boys’ level of externalizing behaviors, but there was no evidence that girls’ externalizing behaviors were affected in the same way. Based on these findings, it appears that the relationship between exposure to psychological abuse and gender varies according to the type of children’s behavior being examined. One explanation of these findings may be that socialization impacts the 60 behaviors that children exhibit when faced with difficult live circumstances. For example, boys and girls may receive different cues about the appropriateness of aggressive or externalizing behaviors, and thus, learn to display different behaviors (Laumakis, Margolirr, & John, 1998). Another possible explanation for these findings is that girls may be resilient against the negative effects of exposure to psychological abuse on their externalizing behaviors. For girls raised in a patriarchal society, verbally abusive comments may seem more normative, and thus, have less of an effect on their acting-out behaviors. In either case, these findings suggest that interventions should focus on girl’s internalizing behaviors when dealing with their exposure to psychological abuse and both internalizing and externalizing behaviors in boys. Otherwise, because children’s internalizing behaviors may be more diffith to discern and because psychological abuse does not leave physical scars, the harmful effects of exposure to psychological abuse could be missed. In particular, boys who are acting out might be the only ones likely to receive help, while girls might mistakenly be assumed to be doing fine. Djspsptipps as s mflisp’pg variabls for exposure tp psysholpgical abuse It seems clear that exposure to psychological abuse is related to some negative aspects of children’s behaviors; however, how exposure to psychological abuse affects children remains unclear. The findings of this study did not support the hypothesis that disruptions in children’s lives meditated the relationship between exposure to psychological abuse and children’s behaviors. Although higher levels of psychological abuse were associated with higher levels of disruptions in children’s lives, there was no evidence that exposure to psychological abuse affected children’s internalizing and 61 externalizing behaviors by creating disruptions in the lives of children (e. g., household moves, moves to shelter, and changing schools). That is, disruptions in children’s lives did not mediate the relationship between exposure to psychological abuse and children’s internalizing behaviors. Previous research has cited disruptions in children’s lives as a variable that may affect the effects of exposure to abuse and children’s behaviors (Margolin, 1998; Peled, 1998), yet there was no evidence of this relationship for children’s exposme to psychological abuse. One reason for this may be that while disruptions in children’s lives may have some negative effects on children’s access to social support and feeling of comfort, for children exposed to domestic violence these negative effects may be outweighed by positive effects such as decreased exposure to psychological abuse against their mothers. Another possible reason is that examination of this hypothesis was hampered by the limited number of indicators of disruptions in children’s lives that could be examined. Only three items were included as indicators of the number of disruptions in children’s lives. These items included the number of household moves, shelter stays, and the number of times the child switched schools. Because this study was a secondary data analysis and the original data collection did not include an instrument specifically designed to measure the number of disruptions in children’s lives, it was not possible to include other items that would represent children’s diverse experiences of disruptions in their lives. Further, these items may have also been indicators of women’s attempts to end the violence against them, and thus, limit their children’s exposure to violence thereby acting as a protective factor for some children. Exppsure to Physical Abuse Agm' s1 Mother 62 These findings from this study also indicate that exposure to physical abuse affects children’s internalizing and externalizing behaviors. Similar to psychological abuse, children’s outcomes were differentially related to exposure to physical abuse. Specifically, children’s internalizing behaviors were influenced by their exposure to physical abuse against their mothers, and there was no evidence that this relationship was dependent on gender (i.e., effects were the same for boys and girls). On the other hand, boy’s and girl’s externalizing behaviors are differentially impacted by exposure to physical abuse against their mothers. Specifically, the degree to which boys were exposed to physical abuse against their mothers affected their levels of externalizing behaviors, but there was no evidence that girl’s externalizing behaviors were affected by their exposure to physical abuse against their mothers. Again, these findings imply that practitioners may need to attend to internalizing behaviors to explore the negative effects of witnessing physical abuse on girls and on both externalizing and internalizing behaviors to understand the efiect on boys. internalizing behaviors, there was no evidence that the physical abuse against their mothers affected children’s internalizing behaviors by creating disruption in children’s lives. However, there was evidence that exposure to physical abuse against mother affected children’s externalizing behaviors by creating disruptions in children’s lives. These disruptions may be creating inconsistencies in children’s lives and causing children to be more insecure in their surroundings. Further, continued disruptions may be exacerbating children’s feelings of isolation and may limit children’s access to buffers such as social support which ultimately leads to more acting out behaviors. Future 63 research should include other areas to better capture children’s experience of disruptions in their lives and the nature of these disruptions to explore the relationship between disruptions and children’s internalizing behaviors. Surnmgy Taken together, these findings make important contributions to our knowledge about how domestic violence affects children’s behaviors. First, exposure to psychological abuse against their mothers is often neglected in the literature. The findings from this study support the few studies that have examined the effect of exposure to psychological abuse and have found that it is related to children’s behaviors (F antuzzo, et al., 1991; Jouriles, et al., 1996; Levendosky & Graham-Bermann, 1998). Evaluations of batterer’s intervention programs have found that while treatment may be successful at reducing the assailant’s use of physical abuse, the majority of men continued to be psychologically abusive (Edleson & Syers, 1990; Gondolf, 1993). Thus, for children in households characterized by domestic violence, exposure to psychological abuse may continue even if the physical abuse ends. The findings from this study suggest that the effects of such abuse should not be minimized, but that they must be understood in the context of children’s gender. Second, findings from this study further illuminate the debate over the relationship between gender and the behavior of child witnesses of domestic violence. Specifically, it appears that boy’s externalizing behaviors are more adversely affected by exposure to physical abuse against their mother’s than girls. For boys participating in this study, higher levels of exposure to psychological and physical abuse were associated with higher levels of externalizing behaviors. However, there was no evidence that girls externalizing behaviors were impacted by their exposure to psychological and physical abuse. This supports previous research which has found the exposure to abuse is associated with more negative behaviors for boys than for girls (J ouriles et al., 1986; Jouriles & Norwood, 1995; Wolfe et al., 1985). However, these results are far fi'om conclusive. Previous research on the differential impact of domestic violence on boys and girls has been inconclusive. Some studies have indicated that boys are more adversely affected by exposure to abuse (Jouriles et al., 1986; Wolfe et al., 1985) while other studies report that girls are more adversely affect than boys (Spaccarelli, et al., 1994; Stemberg et al., 1993). The results of this study and previous research (Jouriles et al., 1986; Jouriles & Norwood, 1995; Spaccarelli, et al., 1994; Stemberg et al., 1993; Wolfe et al., 1985) indicate that there is enough evidence to suggest that boys and girls exhibit problems through different types of behaviors. Future research in this area is wananted to ascertain gender difference with respect to the effect of exposure to domestic violence on children’s behaviors. Lipp'tatiog There are a number of limitations associated with this study. First, because the current study was part of a larger project examining the effects of emotional and physical abuse over time on women and their school-age children, the sample is somewhat small. This may have affected the ability to examine all of the hypothesized relationships. Second, children in this study were only followed for eight months. While this is an important step in examining the impact of domestic violence on children over time, firture work should follow children over a longer period of time. Third, because the present study was a secondary data analysis, there was no way to add measures 65 specifically designed for the research questions examined in this study. This was limiting especially for the hypothesis related to disruptions. The variable ‘disruptions in child’s life’ was a composite of three separate items from the interview with mothers. It would have been more advantageous to administer a reliable questionnaire specially focused on assessing the recent disruptions in the child’s life. Fourth, although this study looked at children’s outcomes associated with both psychological and physical abuse, women were recruited only if they had experienced physical abuse from an intimate partner. All women in the project experienced physical abuse in the four months preceding their recruitment into the study. These women also experienced a substantial amount of psychological abuse but there were no women in this study who experienced only psychological abuse without physical abuse. To truly examine the differential effects of psychological abuse, comparison groups of women who experience psychological abuse only would be essential. Fifth, most of the women in the study were recruited from a shelter. Although all of the interviews took place after women left the shelter, there are still some problems associated with recruitment from shelters. First, as previously mentioned, shelter residents are typically women with fewer economic and social resources. Second, a shelter stay is an obvious disruption in children’s lives. It is imperative that future research incorporate participants who did not use shelter services. Conclusipn The findings of the current study expand our understanding of the processes by which domestic violence affects children’s behaviors. In particular, this research addresses how and when domestic violence affects children. Specifically, child’s gender 66 affected the degree to which exposure to abuse influenced children’s externalizing behaviors. Also, to some extent, there were indications that exposure to physical abuse against the mother was mediated by the number of disruptions in children’s lives. Although this research provides an important step, further research should include larger, more diverse samples and examine other potential mediating factors such as children’s level of social support and coping skills. Also, future research should incorporate other indicators of children’s functioning. Domestic violence could be affecting children in ways that are not captured by the degree to which they exhibit internalizing and externalizing behaviors. Gathering such information is critical to our understanding of how we should intervene and support children who are exposed to domestic violence. 67 Tables 68 bus” . E 9 A25 ~ 25 “meanest .25 :5 a sane £52.. 2: no 28¢. owes; 2: a. :8 5 325:3 ”.56—20 Sm Augu— .o=u€ 32. 8.5a .85» a . . a no eon—SE .82 2: Lo 0953 of a :88 06 230 5:33 08% Romain—q 8m financing» 3.89 as counaflao .895:— owega 2: e. :88 2: on: £56.... . . . o E C Q . . . o... .56 8m was". 8ro “.8398 more? me .355: .82 on. v.8 68%"— eenov vows—Eu «Eu: 23% 28.8.9:— 3:588 no omfiotfi 05 S fiche :38 05 Omsfiw —momm%£n~ 50....— .CDMHOHV .mDECDEOmHM .k—Ohflh "N JD>0§ “w Sag Dmflpfl ~a0mwO—O£O%mg :0 GHOOm Dwmhvxrfl 0:“ Cu aOMDh EQDE 05 .qu—Jfl ~wOmMO~O-D%mQ MOH— .Ducz we. we. mv. 2.. ma. mm. EeSEB meg—2:25 revue—EU NV. on. :.. R. 8. we. eases ”seizes 9:225 L —. «co. 2. So. em. a: 2.50 35%“ 8.5a Rowen—m s. S. s. we. mo. a. a: ygeoaaéaama m. 8. S. S 8. m _ .a use: use; 08% .852; c. o: E. . . cm.m Ema—Lama 33a iowwo—efiewmlnmr El ma ; s. : 3.3.; 9.33me llwwmlllllll )/ 2:: $5 moEung .«o 32333 c5986 98 «:82 \ e3 69 / P $3 £9 288 Se.» 75:8 8 35. $8 XNM sewn so» tan 8 as :2. an ss...§si§§§Damages. assesses in @333. 32 a 2 no 8 masses 8 mE>E we so gm 3% s3 32 .a _ a so» 8 x3 8 333— o\o$ $3 seem /l.|ull|\\\\ll Avenzv 9-30:8 AmbuZV “woe Swuzv Pa an 83% Eomwe—o o» a 8 5:83 me .x. 3 5:83 .3 .x. e m .«e near—t 5:53 mo o\e lllllllllllll )fl I"‘ l/ 70 . Han: «woe—“080m: a. hfiu saga: was . $3 $3 o\...<< 020...; @283th $2 {omm $3 05o: 95.2 8 so» 35fl S no.5. $3 $2 $3 £55322 2: 25 8 5325. Xv gm $3 88: 8m 8 as: o: 5.3» 8023088 no» c3 XNm foam aim 33a mg. 8.5 8 532:... $2 $3 gem so» ta: 3 Etc 5 an: 2: @938 8 3555 $3» e\ow e\.. 2 33 So» ta: 8 vow—88:: .8 @322 3% $2 $8 .8» E: a sec 5 a: 2 35.88. 3.52:8 8.8m .x. a £3 £0; >=E£ So» ta: 2 35.82% $3 $2 Xmm 3E5 5:5 88:8 33.5035 x3 x: 3% ”205.5 bgtzi Egan: £6 £8 3 _ w a; @835 s 8323 $3 $2 a $55 8 >225 :8» 35250 L \x\\\‘ X 33:50 N 035 71 3m AXb— $3 $2 $3 $3 .xé 5qu 9-26:8 3 5:83 .«o .x. {oi fow— =o983 5E vow—nag .8 comma? 53> 5385. $2 $3 3:3me go a: 32. $2 $3 33:8 33% 380m .38 :3 803° a? a: E a 52:5 80.30 cam— .wim BEE 686:9 6869 done—Sm Xmm 33 333m 8 €265 625ch 60:25 :\\\\I! mafia? ES .5 mommflw gov—8m Avmnzv Hmon— AowuZV 2a “a so 595 .8 : “a 5an «o o\.. 00:20? no 095. \‘! 1/ 72 “XL c\c© 0mm .EmofiE .8 acoumozqfioo 35:35 .xL £0 £3 c\om 850% @882me gm XE $2 35% 8 3.83m foo e m o a \..m was? 3835 Q fl a XL .XL mohswomt .8 8:0: Exam Q\ _ o ..\o m {cm 508 coy—05 8 803 9 \% €92 gm £55 we 8 Sam °\om {:3 R. 3 3.38 we £839 £258 £80 fin H $3 $3 8&2: 3923 mmocfiom <1 3qu a: -30 o a AE‘HZV $qu = a H 5825 me .x. 38 a 5:83 ‘8 .x. 8a a 8an «o .x. Danae on? )I/ ){ LI\\\‘ 73 go 6% gm Oi. c\oo X: l/ 74 .xfi £5 o\c~ 85? 25 mat—mom Eva—actom 2:83 855% .5 0.39 Ill/l Essa E3 |I||Ilul||||||lll‘ll .x. ~ _ gem $3 852$: 822$ o 85 $2 852 sen X R :5. v. $3 AU\ mm exam 200:8 womafio O < 65qu A: E swuzv - u o o a a: 30:3 8 56:5 .«o .x. «mom 3 56:5 ea c\.. 89 «a :23: a x. 205956 mo 259 go Ax; o\em oi. oi ill‘nl‘ll‘“ § .8. .x; :0 £1 Ambnzv 3-26:3 an :2ch .8 .x. Ax; .XLN few $2 {av comma? 5:5 83—23% 8 8983 as? 5385. Bo So 5:5 z; 8 3 5585 .830 BEE 88—93 .3on 62355 comma—m 8 696%. 625an dogma wage? ES 8 832m 8885 /\\\\\\\\‘ 9&qu 68 «w :DHU—mfio 1w 0 c\c 8qu 0a 8:203 .«o 0&5. a =2sz / 76 E. V m .32. .3. v m: .mo. v m... . . ”8. #8 v3.9 N? as we we _ 85.38 woo. . . . . $0. woo. . aoo woo one fl 5 «8 oocmtg =32QO moo. Soc. occur—«>8 one—n-30825 So. «co. 2:: Hoop—mtg vac—w omo. woo. 08. was. m8. woo. o8. coo. 85:9» 3028.: 88:0 83:3— Noo. moo. 638933 Baas—um 08E. .58. may 80. .35.. So. as; as; . . . . «@0825 8o 1....ch 35. 1.58. So. :38. 8 tinmn @850 038sz £083 voxml (\W a .m.m 2256 mm 225% .m.m 255% .m .m 325m 3382 m Exam 3 05:. £028.. mm 25% 2 .082 «.2 .252 N _ 38.2 E 322 \..I“ ‘/r I \\l‘ l/ N e3 77 |l|\|‘ .25. V a ...:.:.. .5. V a: .mo. V m... 2.3.2 323 £22 8535 woo. 8o. :5. So. woo. o8. 85:: .228”. a _ c. was. m _ o. moo. 85:9, .8885 $8»? 885m «No. :38. 9.8.3233 0255 3032a no 85:35 «.5532 c8. .1252. Abouacaaxov 33% 303205939 mo 35598 35502 35. So. «no. moor m3. m8. 28:83 83% Rosana mo 35598 Macaw—EU ace. 8.... ace. coo; o5. moor 28:88 08: B>o EVE—Eco .8. 8o. :c. So. do. So; 08:. m8. Ema. m8. Ema. a3. ...:..a. 30225 \! acute omega; flout». cog .m.m 89:33 .m.m 223m .m.m Sgnmm 8.25:?“ c._ E52 2 .322 .1 Eco: P $53.23 m a3. 78 wmo. 1&2. 3835—38 025m 3233 .3 85:36 98:82 mvo. amor Arcana—gov 025m 32%—8&3? ooaotonxo 9.559: x $930 08. 32.9“ 2. oOBSSExoV 83¢ Rommo—onoxmn mo 35588 $8502 to. m8; m8. Re... Abosgaé 38¢ i=6 3o. m 5. v8. moor 28:83 8.5m 32993.8 855%... whet—EU moo. moor moo. Soor 98:83 08: ~26 =36on _ _o. moo. o5. ooo. oEC. o8. :18. woo. :58. Roses Afloobo 03.523 floobo coin .m .m 0383mm .m.m 0355mm 53:33am wA E52 h-“ 3252 Sagas Ea 55:8 3% as. use a . 8.25 585. 79 .25. V a .3...“ .5. v at. dam“. manage 8mg mg E 8:3? fisgmom . o. So. So. moo mo 85:? .8285 . . w s. . 2o moo owe gumbo 883M Aboasaaaxov 83%.. 32929 «a 85598 E 55. moor |||\\\\\\\\‘ 2328 Q ”E 80 .\§.v my: .SV 9:. .3. V m... @933. «N. . . m . . EN «0 8 guaran— omo o oo. Na: . fl . - ooo. 2o - BEES. Rzfimum . . . 823330 25 3388 . co co 8 _ 5 50 N . . 08: ”vocab; 32m $9. as S 2 so. 5. coast?» 30835 muoobo 885m woo. 35. 33:3 25% wmo. ***O©~.u :o. *fwmcor 055‘ E. gag. E. :3: we 25. 332 20% E “80889“ Dagmawm 81 38%» 885: s Q 95. So. . D So 1.02 . A Eat? baa—8.93 83a .afimoBonauo oozotoqxo {0:82 moo. "fl. So. 1 2 So *2; 28:58 8.5m 329330 822398 REED to. mg: m S. 08.- 3. $8.- 352:; 8558 2:: X 52980 So. So. N8. 25.. N3. .8. as: a v8. 3. m8. .336 moo. 3&3. £8.25 ‘|||I\\.\\.\.\.‘ Gaucho owfiozs E v wouoEaHanw E .5 32...... md .832 YD Eco; l|\\|l.||.|||||. f and.» am .93 658 mg .mwd. Hos mo>l= FEE—Eu .3. award—ma mama 3% .05—232 _ ~ 033. Joyce 08:. 6098. m a one. moo. Nos. moo. echo; one. :3 flaw. to. wow. to. woo. aoo. moor ace. So. So. So. So. 03. coo. ow _. one. wand; mmm. ooor Aoc. V a .3... .5. V at. .mo. V m... /l 3:259 02893 3:28.“ 8:356“. acacia—0835 2:: ”uncut? ego—m 028:? 38.2.: ||||.I|||III.I.|||||.. E28 \ N ”E 83 “I 35. mmo. 38b“. 885m 23% 32?? 9». Q a m .3 85.598 @8534 .vmo. coor VMO. Ncor 83% Roofing? cocotomxo FEED . oo. . .- ooo N ooo ooo 08: X 53350 So. So. :o. _oo. 08C. woo. 3&3. mmo. ***mwm. 30825 Amount» omega acute ooxml . . 85m “Seaman .m; as; - 252 S 38} . .Qflfid a 226.56 .«lo “8&0 wag—ulna mfl mid 11.83% onomwooonommm 83— m.: . . \l... 1.. a mad job alldlofi :3 no? a 2088 $5332 49% 88:6 05 mad. . 2 93o 84 woo. 3o. Aoo. v m .3... .5. v at. .mo.vlm* oo§_>om 893.? 3.28% unsung 30825 1|\\.\..\\\\|| l\\\\\\\\\ 83: and: «co. Foo. Noe. coo. moo. moo. Essa o ”E 85 .SoV my; .SV m: «NINA. . moo. . woo . moo o2§>ofl moo . . . o. 2o c8 o—o. co uncut—3 .«aommom 855.» Rue»: . «tube 885% so 25 E22: _ xoo oo.: $2.50 E £8.39:me . . m o. «as. mmo *Iwwo N 1.... Abooncsqxoo 83a 339:3 .3 85598 «.8532 vmo. So. v8. 2o. 29:83 3.3% 3893“— oo 85:35 FEED ooo. Soot ooo. moor 28:83 2:: X 52980 ~ 8. 80. :o. 8o. 08:. . . £988:— omo. {12%. m8 1%? Quote ammonia acute 3le 68883 .m .m 93:35 2-. 322 mafia . z wane—Eu a. gag ole “out... magma ma ma 39% «1‘ III I .\\.I 26 o 32 Em: OE: map—on hom>a£o£ "a flfifimflma m—ofioea — ~ 2 03: II \ .05 .dl all I o :8 o a gob: 0 ocean 0:5 cox... a o 0885 E o 86 2.3: 5o. «8. woo. 39 «m #. mNo. moo. So. So. *mmo... moo. avo. *vaw. 95. .Sov my: .SV m: .3. V m... 03.05 . o. amo woo. Q 8535 3o. 8 as . moo. Noe... 8:33.» 3660M SEES". 08330285 vooo. ~oo. 08.5 “859$ coo—m v2. o8. o3. . go o m fl. 8:35» .3285 900:0 EOUGGM 338355 oo.—mag 085. 06C. 2.29. coo. 3.05.- . Q So :38 as. 33$. 3o. 1%: a o 3 Amount». owfiogo 308.8 3me l\\ \‘ Bonanza 23:5...“ \\\\.\\1 Enamm mam .832 .m .m 225mm .m.w \\\\\.\\.l\.||“ aw... 9.8 m._.oooxm ooxu mm 065. Soc: 8 o 96:. «flu .322 m Tm 352 - m.\|ll.\| w\\\\a|o. 08... 26 _ a law all m .982 \om 803209 553636 8.662 o o o r: 2 u BS .3];— 3 5852 a o\:a mme 03¢ 87 abo moon-Go Eougm mN . “QENNQNOV on: o .33. 3.53 on Eofiééo Bangkok» MEN‘S: . 205.393 85.5qu mzofioE . . . Noo- ma. ms. mmo w go «no 20.53% 823 30532.? 85:35 Macho—EU moo. *o oo. ooo. .55. o5. .& S. 22:58 0:5 ~26 55680 S. \- _o.- :o. . . on: _ .55- :o. .58- o- . on. omo. E088 omo .Iim 138m. omo. 3.1.8. 5 Amouobo omega 38:0 35...“ ll/ “Sagan {new 0385mm .m.m 0685mm .m .m 0385mm o-m 632 Wm 332 v-m $32 -\..\-\\-\\‘ 2 £3 88 . 03. fl: onw o2 NMWNE . moo. woo. woo. . woo woo moo. . mg. vmo. mm . . mNo ~ «No 3:. .50. V m .12.. .8. V at. .mo. VP.” 85.55 02899 _w=v_mom 85%.» 30825 Eggs 2 was 89 )3: . :4“ 30.503 / 3 So. Loo: cmo. on: Awake 8&5on {238 X 5.59 So. *wmo. cNo. cm NmumwoBfiAV-éo unanimous @53on 3o. coo: $250 mmo. woo. woo. moor 83cm 329:3? 85.598 9:22.50 ovo. N _o. ace. .5 S. 2:: ~95 85200 NS. .58.. :o. .53.- 0:5. mo. Ego. a8. 2:63. 30225 39.50 omfloéo 303-co 3me / “2082mm .m .m uEEcmm .m .m cams—cam Tm Eco—z m-m 632 flaw-gm.» 1. E8- 0. 18. alto 015d 3 m < \m 6on unsound oE m-fi odd .095“ _8_wo_o:omma 3&3 . g 0» 736:3 MCI-mm: mung mm and MEN-mama 156%on £255 _o>o:.==2.- 2 £3 41884.: 90 91 .50. V a .12.. .5. V a: .8. V m... boo. mmo. wmbcg mco. o2. boo. vno. cch: mco. o2. 85300 35:: .223 occur-9, 30885 28.28% SEES c ~ mid. 33¢ 55 . ~8- . fie 5o .1 . . *cVO. 080.5%. @8508 X h2:50 omo. to“. moo booowxmfimbako 820.530 @3502 So. ooo: 8280 So. go. mmo. m S. 0.3% 339393 85:35 9:96.30 ooo. *Mzo. ooo. .35. 2:: ~96 52980 :o. . 08:- N- S.- Zo. 2o - E0835 coo. ***wcw. omo. *iwoh. Amoco-co omega 98b» 851ml BEE-«Em .m .m 0685mm .m .m 0883mm mo $82 c-m 322 mama» Ho. duo-om «Quill-7.8088 olfi all a .\-.-.l %§%m§2§2§2 1' I. l..- n g B. . . a fling; 35:00 goo-av o t wove a «Wave/BE g Em: mad-fl awe-.0 2 03$ 92 $63 2..de so. so. so. go. <8. as. m8. «2. .50. V a .31. .5. vat. 6%... 382.59 ass; 35 occur? E0835 885M < woscucoo E 03mg. 93 :3 q .3... .av a: .8. v u... 3.03 562 . cc. 8535 So. So. So m a. 0 ms. o8. m8. 6... 8:25 is; a 8:323 Rookefi $5. $250 8852 08. ... 38.8.:on wéfiéoev a: $230 5 8:955 m8. So. 2o. 25. . _._ Abosaa—axov 8.5a Bambi .8 85598 985.92 m8. . 2o m8. 2 o. eggs 8.5“ 3333 Mo 85598 £220 2 . o *Nmo. . . 28:83 «E: X 5:650 ooo *o—o . . . _ S. . to- :o. to; 08:. 39885 one. flows. amo. 3&2. Each» owflogv 98:0 35% ”Boa—Sam .m.m ougamm .m.w Ban—5mm \1‘ IX}. on E52 \ I M‘s S 2: ml. .mslm own a 3d \: m can—Eu 5 \Eodufiflvdo 88% .u b 5 ~ 9 _ . 5 mo? . .| 1!, \% 8323 jaglaauoio «56623.88. 32%;: 98 .888 auto 0 .3 m \m can 2:: «moouoalfi EI‘ov:...«. u u: 2 nigh 94 Figures 95 939 Eu» 9 S :82 59%”? wficouoasm £28220 £5 FEED E maounama ... ... .- c. . ~— 9. . -. o. .. . .- o p .- . .o.. o .o. .- .u. ..-. u. .a. o o. .- .- ... .... .... I... - cu. .... .. oo. ..- .c. o- a. .u. .- - ... -. ow. .u. . u o .a. . on. .n .. u u. .- - o.- n .- n a Bowmsofifim. 30.6.95. .5502 aEQw< 82?. o. 2335 £2.50 M 555325. m u. 22... m " 35%.“ 327... m ._ u .. moi—33> " m .8300 m 96 Appendix A Assallant’s Psychological Abuse of b cher 97 a list of some of the motional things men might do to annoy or 5)” hurt tbez’rpzmw M115; #0“ art?” to the ”Stafyo 3.: 1 ,2 ...3 4 8 I have N363. Using ‘his card (SHOW 5‘ ELLOW CARD #2) could you tell me, 11:0“, many times in the last 4 months did my of these things to annoy or hurt you? (IF SHE WA NOT INVOLVED LAST 4 MONTHS, SKIP THE MARKED WITH ** AND MARK 8) NEVER ........... - ' RARELY ........ SOMETIMES . _ , . . OFTEN ........ . . - ° (Not Applicable) ' a. HOW often has he refirsed to talk to YOU? ................. ' . . a. How often has he accused you 9f having or wanting other sexual relationshlp(5) ......... - ‘ ......... ° b. Told you about other sexual relationships he wanted orwashavinginordertohurt you?.... .- _ c. Tried to control your money? .............. - d. Triedtocontrol yomacfiVifieS?---......... - - . - - e. Lied to you or deliberately nlisled you? , , , , , , - g. Calledyounames?,,...- - -. ** m. Discouraged your contact With family or friends? h. Triedmhmxiliateyou‘? - i. lgnotfid 0‘ made light Of Your feelings? . . . le (kindled you family or fiends to you? . . . . 1, Harassed your family 01' friends in 5°” my? .......... ........ O n. Threatened to hurt Your family or mmds? .............. 0. Broken Or destroyed something important to you (CONTINUE USING YELLOW CARD #2) 98 WI’IflASS .......... / .......... / ........... / . ... . ooo --- ‘ '..\ t \ \ \ \ NEVER .............. - , 21 RARELY .................... - - 3 80m [[th .................. - : ‘ 4 OFTEN _' ,. s (NOtAppucabne) .....:::::: .......................... p. DO you have or have you had pets in the last four months ........... - I (IF YES,) Abused or threatened to abuse pets to hurt you? ................ Q. Punished or deprived the children when he Was angry at you? ............ r. Threatened to take the child“?11 aWay fiom you? ....................... “5’ Left you somewhere With no way to get home? . - ....................... "1‘ Threatened to end the relationship if you didn’t do what he wanted? ........... u. Tried to force you to leave your home? . . . . . - - - ' . . _ v Threatened to commit suicide when he was mg: at you? ......................... 99 Appendix B Assailant’s Physical Abuse of Moth er 100 and First, Ihave a list different types 0f Physical violence that women have eKPeuenccdjom etc-km CX-partnm, I wonder if you coulfil tell me, to the best of your recollection, Whether M)__../ ever done any of the following things ‘0 You, In the last 4 months: N/A Has he: YES 2N0 . _ - , , 8 a. Broken your glasses or torn you Clothing? 1 ' O : 2 . . . . ' . - - . 8 b- Pushed, grabbed, shoved, or slapped you? 11 . , .. 2 ..... - - . -' 8 C- Punched, kicked, choked or burned 3’0“? X . 2 . . , , . . . - - 8 d'mwnswmttm ‘F"°"‘° $223? s Wlthanobject,orhttyouwrthan0J 1W?- ca1:01'cedsexualactivity? 2...”... - 8 I: Tied You up or physically restrained you X ' ' ' 8 msomeway? \....7- g. Threatened you with a gun or knife 0r used a gun or knife against you 101 Appendix C Assailant’s Injury of Mother 102 8 (IF SHE HAS BEEN HARlVIED AT ALL IN THE LAST 4 MONTHS). Now I’m going core 335%” 1iStofmlysical injuries 311d 351‘ you yes or no ifyou Suffered these injun'cs 50m M25353, 1” t” “'0th. Did you ever suffer: YES . . - 2 ..... - 3' soreness without bruises 8 1 _ . . . 8 8 1" Cuts, scrapes, bruises, red maIkS, 01' welts C- Bums, including rug burns \ . (1. Loose or broken teeth , X . ' . . _ . , 8 , 8 Broken bones or fractures 9 f' Internal injuries g. Smmsorspmns "”8 h Dislocated joints l . . . . i- Pregnancy complications or miscarriage (NO PREGNANCY IN THE LAST 4 MONTHS=8) . 2 ....... 3. Knife or gunshots wound 2 k. Permanent scarring \. Any other physical injuries I haven’t mentioned 8 (Specify) l . . . . 2 ....... 103 Appendix D Disruptions in Child’s Life 104 1. Have you movedinthe last four months? ‘ . - / (WRIrEmEEXAcrNUMBERTflVIES) N0 2. Have you been a shelter residtsflt in the last four months? 3- How many times has C) _ changed schools in the last four, months because of moving, bemg expelled, anything like that? I don t mean natural progression from Head Start to Kindergarten, or elementary to middle school 5) (WRITE THE EXACT NUMBER TIME 105 Appendix E Child’s Experience of Abuse Mother’ s Report 106 Now 1 would like to ask you Whether (A) has ever done any of the following th jugs to (C) . You can just tell me yes or no. Did (A) ever: . . YES NO N/A 1. Break (C’s) ________ glasses or tear his/her clothmg ................... 1 . . 2 - - . 8 2. Push, shove, grab or slap (C) ............................... 1 . . . . .. 2 . 8 3- Puncmdchchoke,orbUIn(C)____ ............................. 1 2 .. , 3 4oThrowsomethingat(C) ,uytohit(C)__ ...................... 1.. 2 .- . 8 With an object or hit (C) __ with an object .......................... 1 ...... 2 . . . . 8 5- Threaten (C) __With3 Weapon .................................. 1 ...... 2 . - . . 8 107 Appendix F Mother’s Report of Children’s Intemalizing and Extemalizing Behaviors 108 Below is a list of items that describe children and youth For each item that describes Your child Within the past 4 months, please circle the 2 if the item is very true or often true of yew Child ngvv or 1 if the item is somewhat or sometimes true of your child If the item is not true of y Our child, Cir ircle the Please answer 311 items 35 well as You can, even if some do not seem to apply to your chfld c e the 0' o‘°.°°>19~.v~.es»~ 10. ll. 12. 13. 14. 15. l6. 17. 18. 19. Argues alot ........................................... 0 ............ 1 .. . - . . 2 Can't concentrate, can't pay attention for long ................. 0 ............ 1 . . . . . . . . 2 Can't sit still, restless, or hyperactive ........................ O ............ 1 - - - . 2 Clingstoadultsortoodependent .......................... 0 ............ 1 2 Complains of loneliness .................................. O ............ 1 ........ 2 Day-dreams or gets lost in his/her thoughts ................... O ............ l ........ 2 Disobedient at home .................................... O ............. 1 , 2 Fears he/she might think or do something bad ................. 0 ............. 1 , - 2 Feels or complains that no one loves him/her ................ . 0 ............. 1 . . . . . . - - 2 Feels worthless or inferior ............................... . 0 ............. 1 , . . . . . . - 2 Getsinmanyfights .................................... _ 0 ............. 1.......-2 Would rather be alone than with others ..................... . 0 ............. 1 ....... . 2 Poor school work ...................................... . 0 ............. 1 ....... ~ 2 Poorly coordinated or clumsy ............................. 0 . . - - . . 1 ....... . 2 Stubborn, sullen, or irritable .............................. 0 ............. 1 ,,,,,,, . 2 Tempertantrumsorhottemper ............................ o - - l ....... . 2 Underactive, slow moving, or lacks energy .................... o . . , , . . - . . 1 . . 2 Unhappy, sad, or depressed ................................ 0 . . . . . . - . 1 . Worries ............................................... O ............. 1 . ° ~ 2 ...... - 2 109 REFERENCES 110 REFERENCES Adamson. L. A., & Thompson, R. A. (1998). COping with interparental Verbal conflict by children exposed to spouse abuse and children fi'om nonviolent homes. Journal of Family Violence, 13(3), 213-232. Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage. Baron, R. M. & Kenny, D. A. ( l 986). The moderator-mediator variable distinction in. social psychological research: Conceptual, strategic, and statistical consrderations. Journal of Personality and Social Psychology, 51 (6), 1 173-1182. Carlson, B. E. (1984). Children’s observations of interparental violence. In. A. R. Roberts (Ed), Battered women and their families: Intervention strategies and treatment programs (147-167). New York, NY: Springer Publishing CO. Christopoulos, C., Cohn, D. A., Shaw, D. 8., Joyce, 8, Sullivan-Hanson, 1., Krafi, S. P., Emery. R. E. (1987). Children of abused women: Adjustment at time of shelter residence. Journal of Marriage and the Family, 49, 611-619. Davis, L. V., Carlson, B. E. (1987). Observation of spouse abuse. What happens to the children. Journal of Interpersonal Violence, 2(3), 278-29, Edleson, J. L. (1999). Children’s witnessing Of adult domestic violence - Journal of Interpersonal Violence, 14(8), 839-870. Emery, R. E. (1982). Interparental conflict and the children of discord and divorce Psychological Bulletin, 92(2), 310-330. ' Eisikovits, Z. & Winstok, Z. (2001). Researching children’s experience violence: Toward a multidimensional conceptualization. In S. A. & J, L. Edleson, (Eds). Domestic Violence in the Lives of Children: 1713 Research, Intervention, and Social Policy. Washington, D. C.: Anierican Psychological Association. ofimelparental .Belm FUIure Qf Fantuzzo, 1., Boruch, R., Beriama, A-s Atkins, M's Marcus, S. (1997). Dom and children: Prevalence and risk in five major US. cities. Journal Academy of Child and Adolescent Psychiatry, 36(1) 116-122. offhe Americ.a'2 Fantuzzo, J.M., Mohr., W. K. (1999). Prevalence and effects of child exposure to domesti violence. Future of Children, 9(3), 21-32. C Fincham, F. D., nyoh, J. H., & Osborne, L.N. (1994). Does marital conflict cause child 111 maladjustment? Directions and challenges for longitudinaI research - J Family Psychology, 8(2), 128-140. ournal of Garbarino, J., Dubrow, N., Kostelny, K., & Pardo, C. (1992). Children in Danger: Coping with the consequences of community violence. San Francisco, CA: Jossey-Bass, Inc. Garmezy, N. (1993). Children in poverty: Resilience despite risk. Psychiatry, 56, 1 27- 1 36. Gore, S. & Eckenrode, J. (1994). Context and process in research on risk and reSilience. In R. J. Haggerty, L.R. Sherrod, N. Garmezy, & M. Rutter (Eds), Stress, risk, and resilience in children and adolescents: Processes, mechanisms, and interventions. Cambridge: University Press. Graham-Bermann, S. A. & Levendosky, A. L. (1998). The social functioning of preschool- aged children whose mothers are emotionally and physically abused. Journal of Emotional Abuse, 1( 1), 59-84. Grusznski, R J ., Brink, J. C. & Edleson, J. L (l 988). Support and education groups for children of battered women. Child Welfare, 67(5), 431,444. Grych, J. H. & Fincharn, F. D. (1990). Marital conflict and children’s adjustment: A cognitive-contextual framework. Psychological Ballerin, 108(2) 267 .290. Grych, J. H., Jouriles, E- N» Swank, P.R., McDonald, R- & Norwood, W. D. (2000). Patterns of adjustment among children 0f battered Women. Journal of Consulting and Clinical Psychology, 68, 84-94. Henning, K. Leitenberg, H., Coffey, R, Turner, T., & Bennett, R. T. (1 996) Lo psychological and social impact of Witnessing PhySicaJ conflict bet‘éeenng-term Journal of Interpersonal Violence, 11, 35-53. par ents. Hershom, M. & Rosenbaum, A. (1985). Children of marital violence: A Closer look unintended victims. American Journal of OrthOpsyciatry, 58(2), 77_90 at the Holden, G. W. (1998). Introduction: The development of researCh into aHOther consequences of family violence. In G. W. Holden, R. Geffiier, and E. N Jo . (Eds), Children Exposed to Marital Violence. Theory Research, and Appfiedumes Issues. American Psychological Association: Washington, DC, Holden, G. W., & Ritchie, K. L. (1991). Linking extreme marital diseord, Child r . g, and child behavior problems: Evidence from battered women. Child Developm en t, 62, 331'327. 112 Holden, G. W., Stein, J. D., Ritchie, K. L., Harris, S. D., & Jouriles, E. N_ (1998) parenting behaviors and beliefs of battered women. In G. W. HOIden, R. Cieflirer and E. N. J ouriles (Eds), Children Exposed to Marital Violence. new), Research and Applied Issues. American Psychological Association: Washington, DC. ’ Hughes, H. M. (1988). Psychological and behavioral correlates of family violence in child witness and victims. American Journal of Orth0psychiatry. 58, 77-90- Hughes, H. H., Graham-Bermann, S. A., & Gruber, G. (2001). Resilience in children exposed to domestic violence. In S. A. Graham-Bermann & J. L. Edleson, (Eds). Domestic Violence in the Lives of Children: The Future of Research, Intervention, and Social Policy. Washington, D. C.: American Psychological Association. Hughes, H. M. & Luke, D. A. (1998). Heterogeneity in adjustment among children of battered women. In G. W. Holden, R. Geflt’ner, and E. N. Jouriles (Eds), Children Exposed to Marital Violence. Theory Research, and Applied Issues. Amaican Psychological Association: Washington, DC. . Hughes, H. M., Parkinson, D., & Vargo, M. (1989). Witnessing spouse abuse and 4 experiencing physical abuse: A “double whammy”? Journal of Family Violence. , 197-209. Jaffe, P. G., Wolfe, D. A., & Wilson, S. K. (1 990). Children of Battere d Women. Newbury Park, CA: Sage Publications. maladjustment among child victims and witnesses to family vi 01 en c e. American Journal of Orthopsychiatry, 56(1), 142-146. Jouriles, E.N. & LeCompte, S. H. (1991). Husbands’ aggression toward Wives d mothers’ and fatherS’ aggression toward children: Moderating efl‘ects 0:11 hi gender. Journal of Consulting and Clinical Psychology, 59(1), 190-192 c M J ouriles, E. N., McDonald, R., Norwood, W. D., & Ezell, E. (2001). Issues and controversies in documenting the prevalence of children’s CXposure to do . violence. In s. A. Graham-3mm & J. L. Edleson, (Eds). Doniestic Vlnfsnc the Lives of Children: The Future of Research, Intervention, and Social golence in Washington, D. C.: American Psychological Association. 0 my Jouriles, E. N., McDonald. K., Stephens. N-. Norwood. W., Spiller, L. c., & Ware H. s (1998). Breaking the cycle of violence: Helping families departing from baitter’s- women’s shelters. In G. w. Holder, R. Geffner. & E. N. Jouriles (Eds), Children Exposed to Marital Violence: Theory, Research, and Applied Issues (pp. 337—369). 113 Washington, DC: American Psychological Association. Jouriles, E.N., Mehta, P., McDonald, R. & Francis, D. J. (1997). Psychometric properties of family members’ reports of parental physical aggression toward clinic-referred children. Journal of Consulting and Clinical Psychology, 65(2), 309-318. Jouriles, E. N., Murphy, C.M., & O’Leary, D. (1989). Interspousal aggression, marital discord, and child problems. Journal of Consulting and Clinical Psychology, 57(3), 453-455. Jouriles, D. N. & Norwood, W. D. (1995). Physical aggression toward boys and girls in families characterized by the battering of women. Journal of Family Psychology, 9(1), 69-78. Jouriles, E. N., Norwood, W. D., McDonald, R., Vincent, J. P., & Mahoney, A. (1996). Physical violence and other forms of marital aggression: Links with children’s behavior problems. Journal of Family Psychological, 10(2), 223-234. Jouriles, E. N., Pfiffner, L. J., & O’Leary, S. G. (1988). Marital conflict, parenting, and toddler conduct problems. (1988). Journal of Abnormal Child Psychology, 16(2), 197-206. Kashani, J. H., Daniel, A. E., Dandoy, A. C., & Holcomb, W. R (1992). Family violence: Impact on children. Journal of the American Academy of Child and Adolescent Psychiatry, 31, 181-189. Kolbo, J. R. (1996). Risk and resilience among children exposed to family violence. Violence and Victims, 11(2), 113-128. Kolbo, J .R., Blakely, E., & Engleman, D. (1996). Children who witness domestic violence: A review of empirical literature. Journal of Interpersonal Violence, 11(2), 281-293. Levondosky, A. A. & Graham-Bermann, S. A. (1998). The moderating effects of parenting stress on children's adjustment in woman-abusing families. Journal of Interpersonal Violence, 13(3), 383-397. Laumakis, M. A., Margolin, G., & John, RS. (1998). The emotional, cognitive, and coping responses of preadolescent children to difl‘erent dimensions of marital conflict. In G. W. Holder, R. Geffiier, & E. N. J ouriles (Eds), Children Exposed to Marital Violence: Theory, Research, and Applied Issues (pp. 337-369). Washington, DC: American Psychological Association. Maker, A. H., Kemmelmeier, M., & Peterson, C. (1998). Long-term psychological 114 IIill111111111121