DOMESTIC VIOLENCE AS A PREDICTO R OF BULLYING AN D VICTIMIZATION BEHAVIORS IN RURAL ADOLESCENTS By JENNIFER SIVINSKI A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2006
Copyright 2006 by Jennifer Sivinski
This document is dedicated to my grandmother for her love and support.
iv ACKNOWLEDGMENTS I would like to thank my family and frie nds for their love a nd support through this process. Specifically, I w ould like to thank my father for his ongoing assistance and motivation, my mother for her love, and my brother for his patience. I would also like to thank my mentor, Brenda Wiens, for her support above and beyond what I had expected. Without her assist ance, it would not have been possible to complete this manuscript.
v TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES............................................................................................................vii ABSTRACT.....................................................................................................................vi ii CHAPTER 1 INTRODUCTION........................................................................................................1 Witnessing Adult-to-Adult Violence in the Home.......................................................1 Bullying....................................................................................................................... .4 Victimization................................................................................................................9 Methodological Considerati ons in the Assessment of Bullying/Victimization..........13 Unique Challenges of Rurality...................................................................................14 Purpose of the Current Study......................................................................................15 2 METHODS.................................................................................................................17 Participants.................................................................................................................17 Measures.....................................................................................................................17 Violence in the home:..........................................................................................18 Bullying:..............................................................................................................18 Peer victimization:...............................................................................................19 Procedure....................................................................................................................19 3 RESULTS...................................................................................................................22 Prevalence of Bullying, Victimi zation, and Family Violence....................................22 Preliminary Analyses..................................................................................................23 Primary Analyses........................................................................................................24 Examination of Hypothesis 1: Pred iction of Bullying Behaviors from Witnessing Violence in the Home...................................................................26 Examination of H2: Prediction of Peer Victimization from the Witnessing of Violence in the Home......................................................................................28 Examination of H3: Moderated Regression Analyses.........................................30
vi 4 DISCUSSION.............................................................................................................32 Summary of Principal Findings..................................................................................32 Research Implications.................................................................................................36 Limitations..................................................................................................................38 Future Directions........................................................................................................40 Conclusion..................................................................................................................42 LIST OF REFERENCES...................................................................................................43 BIOGRAPHICAL SKETCH.............................................................................................48
vii LIST OF TABLES Table page 3-1 Prevalence of Bullying a nd Victimization (Percentages) and Gender Comparisons............................................................................................................22 3-2 Summary of Hierarchical Logi stic Regression Analysis (H1).................................27 3-3 Outcome Variables in Final Step of Logistic Regression Model (HI).....................28 3-4 Summary of Hierarchical Logi stic Regression Analysis (H2).................................29 3-5 Outcome Variables in Final Step of Logistic Regression Model (H2)....................30
viii Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science DOMESTIC VIOLENCE AS A PRED ICTOR OF BULLYING AND PEER VICTIMIZATION AMONG RURAL ADOLESCENTS By Jennifer Sivinski May 2006 Chair: Brenda Wiens Major Department: Clini cal and Health Psychology As posited by the Social Cognitive Theory, behavior exhibited by parents may be learned and mimicked by their children. Speci fically, in the context of the current study, children who witness adult-to-adult violence in the home may display violent behavior or may be more likely to be victimized by peer s. The objective of the current study was to determine if witnessing adult-to-adult violen ce in the home predic ted adolescent bullying or peer victimization in a rural sample. Adolescents in grades 6-12 (N = 1, 539) from Columbia County, Florida, were given two se lf-report scales in order to measure both physical and relational bullying and victimi zation. In addition, students were asked questions regarding their experi ence of violence in the home. Logistic regression models were constructed to determine if witnessi ng violence in the home was a significant predictor of bullying or vict imization behaviors. Afte r controlling for demographic factors and child abuse in the home, results indicated that witne ssing domestic violence was a significant predictor for bullying beha viors but not peer victimization. In
ix conclusion, these findings have implications for anti-bullying cu rriculum development and for interventions with children who have witnessed adult-to-adult violence in the home.
1 CHAPTER 1 INTRODUCTION According to Albert Banduraâ€™s Social Cogni tive Theory (1973), individuals learn by imitating the actions of others (termed â€œmodeli ngâ€). Consistent with this concept, one way in which children learn is by observing ot herâ€™s behavior. Following from this, it is possible that children and adol escents who witness violence in the home may learn that it is acceptable to commit violent acts themselv es. In a similar manner, children might observe and mimic the role of the victim in a given relationship. Occurrences of school violence and increased reports of bullying behavior have led to focused efforts to determine what factors are rela ted to violent behavior in you th, especially since there are multiple short and long-term consequences of youth bullying and victimization. The current study attempted to add to the knowledge base regarding factors related to violent behavior in youth by looking at the impact of witnessing ad ult-to-adult violence in the home on adolescent bullying and victimization behaviors Witnessing Adult-to-Adult Violence in the Home Severe violence is thought to be present in almost 13% of all marriages in the United States (Straus, 1977). Children are often the unintended vict ims of adult-to-adult violence in the home, and the effects of witn essing this violence have been increasingly well-documented over the past two decades. Findings suggest that youth who witness violence in the home are more likely to have increased levels of de pression, externalizing and internalizing disorders, decreased self -esteem and social competence, and higher
2 levels of child psychopathology when compared to peers who have not witnessed such violence (Levendosky, Huth-Bocks, & Semel, 2000). The effects of witnessing violence in the home are not limited to childhood. According to a review conducted by Widom ( 1989), up to 70% of violent adults have a history of exposure to violence in childhood, either as direct victims or as witnesses of interparental violence. A nationally admini stered survey by Straus (1975) found that adults who had observed violence as childre n were more likely to approve of violent means of conflict resolution as adults. Similarly, Kalmuss (1984) reported that observing interparental hitting was strongl y related to severe marital aggression in adults. In fact, observation of interparental violence had str onger effects on future marital aggression than did parent-child violence. A more re cent study by Ehrensaft and Brown (2003) also found that children who witnesse d adult-to-adult violence in the home were significantly more at risk for adult partner violence than children who had not witnessed this type of violence in the home. However, several other studies have found that children who have witnessed adultto-adult violence in the home may not be negatively affected. Jaffe and Wolfe (1990) found that interparental violence may not al ways have negative behavioral effects on children. Instead, they posited that, in some s ituations, it actually had protective effects on those who were exposed. They based this conclusion on findings that some children in the study who were exposed to interparen tal violence had higher self-concepts than children not exposed to violence in the home. Song, Singer, and Anglin (1998) found that being a victim or witness of violence at home was not a predicto r of violent behavior for boys, but was a predictor of violent behavior s in girls. In gene ral, it appears that
3 childrenâ€™s responses to witn essing adult-to-adult violen ce in the home may vary depending on a number of factors, some of which may still be unknown, and it should not be assumed that all children will react similarly. Some research in the area has documented that children who were abused and also witnessed adult-to-adult violen ce in the home were more severely affected than children who were abused but did not witness violen ce in the home, or children who witnessed violence in the home but were not physica lly abused. A study by Miller and Gilner (1991) found that adolescents with a history of both abuse and witnessing violence had significantly higher scores on a measure examin ing effects of child abuse (Child Abuse Potential Inventory) than those who had not b een a victim or witne ss of violence in the home or children who had been abused but did not witness interparental violence. Considering that there can be overlap between child abuse and adult-to-adult violence in the home, such findings speak to the importance of examining the effects of witnessing adult-to-adult violence in the home separately fr om child abuse in orde r to control for the influence of child abuse on behavior. The majority of studies in this area have been conducted with younger populations, and thus do not address the effects of observi ng such violence on adolescents, who are at a unique developmental stage. Adolescents are more independent and have formed many of their own opinions, but still live within the rules and instructions of their caregivers. Straus (1992) has suggested there might be as many as 10 million teenagers exposed to parental violence each year. While results of a national survey conducted in 1987 suggested that approximately 3.3 million children are exposed to domestic violence each year (Fantuzzo & Mohr, 1999), the number of adolescents exposed has not been
4 empirically studied. A few st udies have looked at the impact of family violence on adolescent relationships. Wekerle and Gelles (1998) found that witnessing or experiencing family violence was the most significant predictor of adolescent male abusive behavior in an adolescent roman tic relationship. Also, witnessing or experiencing family violence was found to be a significant predictor for male or female victimization when in a close relations hip with a member of the opposite sex. Levendosky and colleagues (2002) found that adol escents who witnessed violence in the home were more likely to have an avoidant a ttachment style and less likely to be securely attached. Based on their results, they sugge sted that adolescents might internalize the relationship difficulties they witness in the home. While this literature documents the effects of family violence on adolescentsâ€™ romantic relations hips, less research has been done on how witnessing such violence affects adolescent peer relationships. Thus, the unique effects of witnessing adult-to-adult vi olence in the home on peer relationships during adolescence are still largely unknown. In general, research in the area of witn essing adult-to-adult violence in the home has determined that there are a number of c onsequences for those th at witness violence, both in childhood and adolescence. While research determini ng the effects of domestic violence on adolescent social functioning is more limited, there is evidence to suggest that family violence may influence adol escent aggressive behavior in romantic relationships. Bullying Bullying has been defined in the literatur e as repeated behavi or, both physical and verbal, that occurs over time a nd entails an imbalance of strength and power, so that the individual being bullied has difficulty defending him or he rself (Craig, 1997). An
5 increasingly common way that researchers cla ssify bullying behavior has been to identify categories of bullying. These categories identi fy individuals as bulli es, bully-victims, or victims. Those that fall into the role of bu llies are aggressive with other students, but are never themselves the victims of bullying be haviors. Victims are consistently acted aggressively upon, but are not considered to consistently behave aggressively towards others. Bully-victims alternate between the role of bully and/or victim in a given situation (Austin & Joseph, 1996). The emerging theme from the most commonly utilized definitions in the fiel d is that bullying consists of group, as opposed to individual, mentality/brutality and can be physical or psychological in nature. Bullying and victimization most likely affect almost all stud ents in a school to some extent, including bystanders. Due in part to a recent surge of literature in the area, bullying is now widely regarded as a serious problem for the health and well-being of children. Estimates of bullying prevalence vary gr eatly depending on a number of factors including survey instrument utilized, research location, ages of students surveyed, and number of participants surveyed. Smith and colleagues (2000) stated that bullying affects between 7 and 35% of students in the United St ates, Canada, Australia, and Japan. Seals and Young (2003) surveyed 430 students in California and found that 24% of the sampled students reported recent bullying invol vement. Data regarding the relationship of bullying to gender, age, ethnicity, and ot her demographic information have also been reported in the literature. Seals and Y oung (2003) found that significantly more boys than girls were involved in physical bullying behaviors a nd that there was a drop in bullying between seventh and eighth grades. Other studies (Smith et al., 2000; Xin Ma, 2002) have also shown an age drop in bully ing during middle school. In terms of the
6 effect of ethnicity on bullyi ng behavior, Seals and Young (2003) did not find differences in rates of bullying when comparing Cau casian and African American students. Bosworth and colleagues (1999) also found no significant association between ethnicity and bullying behaviors. Additionally, Sourander and Helstela (2000) found no significant associations between parental education or soci o-economic status and bullying persistence from ages eight to sixteen. Physical bullying has been shown to be pe rsistent and have negative consequences for bullies. A longitudinal study by Sourander and Helstela (2000) found that bullying at age eight was associated with bullying at age sixteen. They also found that bullying at both time points led to a wide range of psyc hological problems for bullies and referral to child mental health services. Kumpulainen and Rasanen (2000) found that those students who were labeled as bullies at age eight were more likely to have significant deviant and illegal behavior seven years la ter at age fifteen. Other findi ngs suggest that bullies are at greater risk for depression a nd social anxiety. Students in grades five through eight who were bullies reported increas ed levels of depression on the Childrenâ€™s Depression Inventory or anxiety on the Franke and Hymel (1984) Soci al Anxiety Scale (Craig, 1998). Interestingly, a 2001 school-based st udy (Haynie, Nansel, & Eitel, 2001) found that bully-victims had higher rates of depressi on and behavioral problems than children who were only victims or bullies. In anot her study conducted in Greece, Andreau (2001) found that bully-victims also had lower leve ls of social acceptance and problem-solving abilities than bullies, victims, or those not involved in any related behavior. These two studies suggest that bully-victims have unique traits not shared by those who are only victims or bullies.
7 Crick and colleagues (1999) have posited that bullying behavior is not always physical in nature. They use the term â€œrela tional aggressionâ€ to refer to more subtle aggressive acts that utilize th e power of relationships to vic timize others. Specifically, as opposed to physical aggression that encompasse s physical threat of harm or actual harm, relational aggressive acts â€œare those in which damage to relationships (or the threat of damage) serves as a means of harmâ€ (C rick, 2002, p. 599). Relational aggression generally does not include physic al aggression; instead, it refe rs to socially manipulative behaviors such as spreading rumors or social exclusion. Relational aggression is frequently seen in American schools and rates have been found to differ by gender. Crick (1996) a nd other researchers (Bosworth, Espelage, & Simon, 1999) have found girls to display hi gher amounts of relational aggression than boys, while boys displayed a higher frequency of direct aggressive behaviors. As with physical bullying, researchers have found rela tional bullying to be associated with negative outcomes. Craig (1998) found that relational aggressors suffered from more depression and social anxiety than a co mparison population of fifth through eighth graders who did not endorse relationally aggressi ve behaviors. Furthermore, relationally aggressive children evidence more externaliz ing symptoms and are more disliked than other middle-school aged students (Prins tein, Boergers, & Vernberg, 2001). Potential predictors of bullying behavior are coming to light as a result of recent studies and appear to be multiply derive d. For example, it has been shown that aggressive attitudes predic t bullying behavior in middl e-school students (McConville & Cornell, 2003). Also, it has been shown that the company one keeps (i .e., an adolescentâ€™s peer group) effects adolescen t bullying behavior (Espelag e, Holt, & Henkel, 2003).
8 Through peer nomination, Espelage and colleag ues (2003) found that students (both male and female) generally affiliated with other students who had comparable levels of selfreported bullying behaviors. In addition to the peer group, the home environment has also been shown to play a role in the development of bullying behavior s (Farrington & Ely, 1993) . In fact, violent homes are one of the highest risk factors fo r anti-social behavior in general (Seals & Young, 2003). The effect of the family environment on bullying behaviors has been studied both in the United States and abro ad. Several studies have found associations between specific parenting styles and bullyi ng behavior in childhood. For example, it has been found that the parents of bullies often ha ve conflictual relations hips, lack warmth in their relationships, are authoritarian, a nd use power-assertive me thods for discipline which include physical punishment (Lowen stein, 1978; Olweus, 1994; Smith & MyronWilson, 1998). Bowers and Binney (1992) found th at bullies were more likely to live in non-cohesive families as measured by the Family System Test, which examined how children ages eight to eleven perceived th eir families on dimensions of Power and Cohesion. Baldry and Farrington (2000) co mpared children who were self-reported bullies to those who were reportedly delinque nts (i.e., committed illegal acts) and found that bullies had authorit arian, punitive, and non-supportive parents who tended to disagree with each other. Bullies were also more likely to disagree with their parents than non-bullies. This particular study found that these parental characteristics were unique to bullies. Those stude nts classified as delinquents were more likely to have conflictual and low supportive parents as opposed to the puniti ve and non-supportive parenting associated with bullies.
9 The unique impact of witnessing adult-to-a dult violence in the home on childrenâ€™s bullying behaviors has only been documente d in one previous Italian study (Baldry, 2003). This study, conducted in Rome, Italy, ex amined the impact of domestic violence on childrenâ€™s bullying behavior in middle a nd high schools. The researcher utilized a translated version of the bullying questionnaire designed by Olweus (1993) to measure bullying behaviors. Exposure to domestic violence was measured with a modified version of the Conflict Tactic Scale. Baldry (2003) f ound that exposure to domestic violence was associated with increased bu llying behaviors. Ov erall, she found that bullies were 1.8 times more likely than nonbullies to have been exposed to domestic violence. Also, there was a highly significan t relationship between bullying and maternal verbal aggression against the father figure in the home, as well as a slightly less significant correlation between paternal phys ical aggression in the home and bullying behaviors. In these cases she found that boys and girls exposed to th is type of violence were significantly more likely to be involved in both physical and relational bullying at school. Girlsâ€™ bullying behavi or appeared to be somewh at more influenced by the witnessing of domestic violence than boys. In general, this study provided evidence that bullying and witnessing adult-to-adult violen ce in the home are related phenomena. Victimization Prevalence rates of victimization from bullying are widely disputed in the literature. In a large, school-based sample , Duncan (1999) reported that 25% of students were victims of bullying. However, Perry, Kusel, and Perry (1998) found that only 10% of students reported peer victimization, while a retr ospective study by Hoover and colleagues (1992) found that 77% of adults reported being victims of bullying before the age of eighteen. Whether or not a consiste nt prevalence rate can be determined in
10 American schools, peer victim ization is clearly a significan t problem that affects many children. Like bullying, peer victimization has been found to vary along demographic characteristics, especially age and gender. A retrospective study by Eslea and Rees (2001) found that adults genera lly remembered being bullied be tween the ages of eleven and thirteen, with less risk for peer victim ization before and after this age range. A Canadian study conducted by Xin Ma (2002) in a rural school dist rict found gender and physical condition to be the most importan t characteristics associated with being victimized in grades six and eight; boys were more likely to be victimized as well as those individuals in poor physical condition (including children who were overweight, disabled, etc.). Prevalence rates for victimization by relati onal aggression have been reported less in the literature. Storch and Masia-Warner (2004) found that 17.5% of the females in their sample reported being victims of re lational aggression. Crick (1996) has also reported high levels of victimization from relational aggression (above 20%), which was found to be linked to psychosocial adjustment problems for children. Similar to the commission of relational aggression, being the victim of relational aggression has been found to be more commonly reported by girl s than boys. Crick and Grotpeter (1996) found that boys were less likely to be victim s of relational aggression and more likely to be victims of physical aggression. Baldry a nd Farrington (2000) al so found that girl victims suffered more from relational bullyi ng, while boy victims were more affected by physical bullying. However, several recent stud ies have suggested that male and female students are equally likely to be the targ et of relational aggression (Paquette & Underwood, 1999; Roecker-Phelps, 2001). Roecke r-Phelps (2001) suggested that studies
11 that have found gender differences may be overlooking relational victimization in boys because physical victimization receives more attention and may be perceived as more harmful to them. Research suggests that victims of physical bullying are at risk for suicide (Tanaka, 2001), medical and psychological conditions (Bond & Thomas, 2001), and other negative outcomes. Craig (1998) found that victims of bullies reported si gnificantly higher anxiety and depression than ch ildren who had not been victimized. When these results were examined by gender and age, it was dete rmined that female victims reported more depression than male victims, and that olde r victims also reported more depression than younger victims (Craig, 1998). In this study, victims reported a higher level of anxiety than bullies or students who did not report ei ther bullying or peer victimization. Relational aggression can also affect victims in various wa ys. Victims of relational aggression have been found to be more depres sed, anxious, and to have lower self-esteem than children who do not report such peer victimization (Crick & Grotpeter, 1996; Ladd & Ladd, 2001). Yoon and colleagues (2004) s uggested that relational aggression may have a greater impact in adolescence because of the importance of social relationships during this developmental period. Victims of relational aggression also are more likely than those who are not victimized to feel that there is some thing they have done to cause the maltreatment by others. These negative attr ibutions may lead to additional decreases in self-esteem and other difficulties, su ch as depression (Ladd & Ladd, 2001). Victims of both physical and relational bully ing often fail to report that they have been bullied (Smith, Morita, Junger-Tas, Ol weus, Catalano, & Slee, 2001). Because of this, school authorities may be unaware of the extent of viol ent activity occurring in their
12 educational environments and may not take necessary action. Newman, Murray, and Lussier (2001) found that girls were more willing to seek help than boys, although in general decisions were influenced by the stud entsâ€™ self-concepts and emotions, as well as their perceptions of how teachers and cla ssmates might respond to their help-seeking behavior. Another study by Unnever and Corn ell (2004) also looked at factors that may influence a studentâ€™s decision to report schoo l bullying. Reporting was found to increase with the chronicity of vic timization. Students who percei ved their parents as using coercive forms of discipline (such as threat ening language and other forms of aggression) were less likely to report bei ng bullied at school. Past re search indicates that helpseeking can provide health related bene fits (Kaukinen, 2002) and can reduce the probability of being victimized in the future (Ladd & Ladd, 2001). General effects of the home environment on peer victimization have been examined in several studies. A Norwegian study by Fosse and Holen (2002) surveyed adult psychiatric outpatients and f ound that men who were verbal ly or physically bullied in childhood were more likely to have been raised without a biological fa ther. Furthermore, women who were victims of physical or rela tional bullying as children were more likely to have had less care from their fath ers when younger and to have experienced significantly more emotional a nd physical neglect and abuse growing up. Bully-victims have described their home lives as cohesive , but enmeshed (Bowers et al., 1992). Also, Rigby (1994) reported that female bully-victi ms had poorer family functioning than nonvictims, but that there was no difference between male victims and non-victims when they were in families with both the mother a nd father present. In a retrospective study on the effects of the home environment on p eer victimization, Duncan (1999) surveyed
13 college freshman in order to assess frequenc y of abuse, peer vict imization, and current psychological distress. She found that victims of bullying were significantly more likely to be victims of emotional and physical maltr eatment by their parents than non-victims. Victims of bullies were significantly more likel y to have been slapped or kicked by their mothers and were also more likely to have experienced childhood sexual assault. Finally, it has been found that children with coercive pa rents are less likely to report being bullied (Unnever & Cornell, 2004). Baldry (2003) examined the relationship between exposure to domestic violence and peer victimization and found a significan t relationship between witnessing domestic violence and being victimized at school. In fact, 71% of children who were exposed to domestic violence were also victimized at school. In contrast, 56.9% of children who were not exposed to domestic violence report ed being victimized at school, which was a significant difference. Additionally, while expos ure to maternal physical violence against the father was not significantly associated w ith victimization, patern al physical violence against the mother was positively associated wi th victimization. Results of this study suggested a significant relationship between witnessing violence in the home and being the victim of bullying. However, this study did not examine potential moderating factors and focused mainly on children ages eight to fifteen. Methodological Considerations in the Assessment of Bullying/Victimization As noted earlier, definitions and preval ence rates of bullying and victimization differ in the literature. Some of these vari ations likely result from differences in the assessment of these behaviors. Pellegrini and Bartini (2000) compared peer and selfreport measures to identify middle-schoolers with different levels of aggression and victimization. They found that both peer and self-report measures were associated in
14 their identification of aggressive students. They also found that student diary entries were significantly related to the self-report meas ure. In general, they found that the selfreport measure was an adequate way to a ssess bullying behavior. However, they suggested that, optimally, several measures s hould be used to assess both bullying and victimization behaviors. In another study, Olweus (1993) f ound childrenâ€™s self-reports to be valid indicators of bullying behaviors, pe rhaps even more valid than assessments by school officials who sometimes do not realize th e extent of violence in their schools. Unique Challenges of Rurality The specific challenges that rural communities face are often underrepresented in literature concerning social probl ems. Individuals living in ru ral environments often have less access to health-care, jobs, education, and general necessities. These challenges could potentially make redressing social problems such as family violence and bullying/victimization all the more difficult for rural communities. However, it is vitally important to study and address these issues in rural communities, especially since previous research has indicated that rates of domestic violen ce and bullying/victimization may be as high or higher in rural communities as compared to urban communities. Studies conducted in rural areas suggest that approximately one in five women report being the victim of violence by a nother adult. A study of 136 women presenting for care in rural medical clinics found th at between 20-28% of these women had experienced recent physical abuse by anothe r adult (Johnson & Elliot, 1997). In an examination of abuse against women in rura l Minnesota, Kershner and colleagues (1988) found that 21.4% of the women surveyed ha d experienced abuse ( physical, emotional, sexual) in the past 12 months. They found that women who were single, separated, or divorced were significantly more likely to be in an abusive relationship than married
15 women. In another study, Logan and colle agues (2003) compared rural and urban victims of domestic violence and found that rural female victims reported significantly less social support, income, and education. Th ey also reported wors e overall health and mental health. In a shelter-based study l ooking at rural domestic violence, 22% of the women interviewed had experienced physical, se xual, and verbal abuse (Krishnan et al., 2001). Only half of the women at the shelte r had reported the abuse to law enforcement, which was the most common help-seeking be havior reported, and only one-third had sought counseling. These findings suggest that family violence is a significant problem for rural communities. Bullying is also a significant problem in many rural communities. In the past, school violence has been associated with ur ban communities. However, several recent studies suggest that bullying may be equally pr evalent in rural school settings. Pellegrini and Bartini (2000) found that the mean num ber of bullying/victimization incidents reported by sixth graders in a rural setti ng was 3.57 incidents reported over a one-month period. Stockdale and colleagues (2002) surveyed 739 fourth, fifth, and sixth graders from seven rural elementary schools and found that approximat ely 34% of students agreed that they had been bullied in the pa st week. The study also found that bullying and victimization were correlated with aggr ession and studentsâ€™ approval of violent means of conflict resolution. These findings und erscore the presence of bullying in rural schools at levels at least equi valent to, if not higher tha n, those found in urban schools. Purpose of the Current Study The purpose of the current study was to examine whether adolescent reports of witnessing adult-to-adult violence in the home were related to self-reported bullying behaviors (both physical and re lational) and reports of expe riencing victimization (i.e.,
16 being bullied, both physical and relational). In addition, this study aimed to examine potential moderators of these relationships. Since there are relativ ely fewer studies of these issues with rural popul ations, the current study focused on a rural sample that may help shed further light on the prevalence of , and the relations be tween, witnessing adultto-adult violence in the home and bu llying/victimization in rural areas. There were several main hypotheses for the current study: 1. It was hypothesized that witn essing adult-to-adult violen ce in the home would be predictive of adolescent reports of bully ing behaviors when controlling for the potential effects of child abus e in the home (against adoles cent reporter or sibling). 2. It was hypothesized that witn essing adult-to-adult violen ce in the home would be predictive of adolescent reports of bei ng victimized (bullied) by peers when controlling for the potential effects of child abuse in the home (against adolescent reporter or sibling). 3. It was believed that age, gender, and ethnicity could be moderators of the relationship between witnessing adult-to-adul t violence in the home and reports of bullying and/or victimization. It was hypothesized that younger children might show more of a relationsh ip between witnessing viol ence and bullying and/or victimization. Based on past research in the area (Baldry, 2003), we also theorized that girls may demonstrate more bullyi ng behavior/peer victimization following exposure to violence in the home. Ther e was no specific hypothesis regarding the direction or impact of ethnicity on the relationship between witnessing violence in the home and bullying and/or victimization behaviors.
17 CHAPTER 2 METHODS Participants The study population consisted of student s in grades 6-12 attending all public middle and high schools in Columbia County, Florida. All students in attendance on April 12, 2005 were asked to complete th e study questionnaire by their classroom teacher, resulting in a total sample size of 3,004. For this study, a sub-sample of 1,539 students who answered questions about past 30-day behavi ors (procedure described in more detail below) was utilized. The gender distribution was 52.5% female and 47.4% male. The majority of students were Cau casian (70.6%), 19.6% we re African-American, 3.1% Hispanic, and 5.9% were of other ethnic origins. The distri bution of students by grade was as follows: sixth grade, 17.2% of th e overall students sampled; seventh grade, 18.6%; eighth grade, 19.6%; ninth grade, 13.4% ; tenth grade, 12.8%; eleventh grade, 12.0%; and twelfth grade, 6.1%. Fifty-seven percent of the students sampled were twelve or younger, 17.9% were thirteen-years-o ld, 18.3% were fourteen, 15.6% were fifteen, 13.5% were sixteen, and 16.5% were seventeen. Measures The data for this study originated from a subs et of items included as part of a larger questionnaire designed for the evaluation of a federally-fund ed prevention and intervention program (Columbia Acting Togeth er for Children â€“ CATCh) for youth risk behavior in Columbia County, Florida. Project CATCh was initiated in 2003 and focuses on decreasing substance use, violence, and othe r youth risk behaviors in this rural region.
18 As part of the evaluation plan for Proj ect CATCH, the Risk Incidence for Schools Inventory (RIScI) was developed to assess the frequency of vari ous youth risk behaviors. The RIScI was designed by University of Flor ida staff to be an anonymous questionnaire to examine a range of risk behaviors in the areas of substance use (alcohol, tobacco, marijuana, and other drug use), violence/aggr ession at school and in the home, and peer victmization, as well as querying students about their attitudes towards Columbia Countyâ€™s school drug-testing polic y. The RIScI also includes questions regarding basic demographic information, including sex, grade, age, ethnicity, and living situation. The yearly survey administration is one facet of ongoing program evaluation conducted by the University of Florida National Rural Behavior al Health Center (N RBHC) in collaboration with the Columbia County School District. The 2005 RIScI survey contained 122 items. For the purposes of the present study, que stions addressing vi olence in the home, bullying, and victimization be haviors were examined. Violence in the home: The question addressing adult violence in the home was, â€œHave you witnessed one or more of your parents/guardians harm another adult in your home?â€ The question concerning child abuse in the home was, â€œHave you witnessed one or more of your parents/guardians harm you or your siblings?â€ The responses were Likert scale format and consisted of the response options â€œnever, almost never, sometimes, almost always, and always.â€ Bullying: The Aggression Scale developed by Sc hwartz and colleagues (1998) was embedded in the RIScI to measure the physic al and relational bullying behaviors of students. The scale consists of five highl y loaded items (Cronbachâ€™s alpha = .75) that
19 assess studentsâ€™ bullying behavior on and off campus. The items on The Aggression Scale are presented in Appendix 1. Res pondents could choose one of the following response options for frequency of bullying be havior: 1) never, 2) almost never, 3) sometimes, 4) almost all the time, or 5) al l the time. A reliability analysis was conducted for the present study that also demonstrated st rong internal reliabi lity with our sample (Cronbachâ€™s alpha = .85). Peer victimization: The Social Experience Questionnaire-Self -Report (SEQ-SR; Cr ick & Grotpeter, 1996) was embedded in the RIScI to measure p eer victimization. The SEQ-SR consists of three subscales (Overt Victimizatio n, Relational Victimization, and Prosocial Behaviors) with questions that are answered on a five-point Likert scale: 1) never, 2) almost never, 3) sometimes, 4) almost all th e time, or 5) all the time. Items from two subscales, Overt Victimization and Relational Victimization, were embedded in the RIScI (see Appendix 2). Items from the SEQ-SR were modified slightly to reflect the age of the students being surveyed, as adolescents ma y prefer not to be addressed as â€œkids.â€ The term â€œkidsâ€ was changed to â€œstudentsâ€ in order to make the survey more amenable to the adolescent population that was queried fo r this study. The SEQSR has been shown to have good test-retest reliability over four weeks ( r = .90; Crick & Masia, 2001) and strong internal reliability for each subscale wi th coefficient alphas ranging from .77 to .80 (Crick & Grotpeter, 1996.) A reliability analysis was conduc ted for the present study that also demonstrated strong internal reliabil ity with our sample (Cronbachâ€™s alpha = .90.) Procedure Classroom teachers administered the RISc I to students during their first period class on April 12, 2005. Staff members from th e NRBHC were at each school to provide
20 directions over the morning announcements a nd assistance with any questions that arose during administration. While all responses we re anonymous, teachers and staff remained in student classrooms to ensure that a ll students were completing the surveys individually. Care was taken to ensure that the students were seated so that they could not see the responses of their classmates. Su rveys were collated such that one-half of the students surveyed in each classroom were admi nistered a 30-day version of the RIScI that asked them to report frequency of behaviors th at had occurred over the last 30 days. The other half of the students we re administered a 12-month version of the RIScI that asked them to report frequency of behaviors that ha d occurred in the past 12 months. Questions on both versions were identical except for th e time frame on which they were queried. The surveys were distributed in this manner in order to gather 30-day and 12-month data while ensuring that the process was not overly exhaustive for the students completing the surveys (i.e., not asking students to repor t on both 30-day and 12-month frequencies). Surveys were collected by NRBHC staff im mediately once all stude nts in a classroom had completed their surveys. Only the surv eys that queried stude nt behavior over the past 30 days were utilized in this studyâ€™s an alyses to obtain the most recent portrayal of student behavior. Data collected with the RIScI were check ed by several research staff, including a statistician, using five sepa rate mechanisms, which resulted in a total of 321 RIScIâ€™s being discarded prior to analys es for the following reasons: 1. An item inquiring about use of a false dr ug named â€derbisolâ€ was included in a series of questions about use of different drugs. If a student reported using this drug, their entire questionnaire was discarded. 2. Questionnaires were checked for invalid response patterns (e.g., entering the same response choice on every item or sets of items, creating patterns - â€œChristmas
21 treeingâ€ of responses). If a questionnaire appeared to contain an invalid response pattern, the entire questionnaire was discarded. 3. Questionnaires were discarded if the st udent endorsed â€œDailyâ€ on any of the questions asking about the av erage occurrence within the last 30 days of cocaine, hallucinogen, or ecstasy use, or being physic ally hurt by a student with a knife. 4. A question was added to the 2005 survey th at stated, â€œHave you answered the items on this questionnaire truthfully.â€ Ques tionnaires on which th e response â€œnoâ€ was selected were discarded. 5. Following the above mentioned data check ing, the statistician examined all forms that were scanned in order to ensure th at forms were properly entered into the database system.
22 CHAPTER 3 RESULTS Prevalence of Bullying, Victimiz ation, and Family Violence Overall, 22.6% of students reported regularly bullying other students. Similarly, 23.5% of students reported being regularly vi ctimized by others. While overall gender differences in bullying and peer victimiza tion were small, frequencies of behavior differed by gender on individual items. Freque ncy distributions by gender for individual items on bullying and victimization subscales are shown in Table 3-1. Table 3-1. Prevalence of Bullying and Victimization (Percentages) and Gender Comparisons All Students (N = 1537) Girls (N = 807) Boys (N = 729) Bullying Prevalence: Bully/pick on 17.8 14.9 20.9 Tease/make fun of 29.5 28.0 30.8 Hit/push 20.5 16.2 25.1 Gossip, spread rumors, or say mean things 18.6 24.8 13.8 Hurt another studentâ€™s feelings by not including them 10.2 10.9 9.5 Overall Bullying: 22.8 21.3 24.2 Victimization Prevalence: Leave you out on purpose 16.8 19.8 13.7 Yell and call you mean names 18.6 19.0 18.3 Not let you in group anymore 13.2 17.8 8.1 Tell lies about you 21.8 25.6 17.7 Have to do what they say 10.1 12.5 7.5 Says mean things about you 18.1 22.3 8.1 Say will beat you up if donâ€™t do what want 6.2 4.2 8.5 Gossip/spread rumors about you 24.2 30.3 17.3 Threaten to hit, slap, kick you 5.3 4.0 6.4 Hit, slap, or kick you 13.3 10.2 16.6 Overall Victimization: 23.7 25.6 22.5
23 When asked if they had witnessed one of their parents or guardians harm another adult in the home at least a few times in th eir lives, 7.8% of student s reported witnessing this type of violence in the home. Regular witnessing of this behavior was much lower; 2.5% of students reported witne ssing inter-adult violence in the home on a regular basis. With regards to child abuse, 5.1% of stude nts had witnessed one or more of their parents/guardians harm them or their siblings at least a few times in their lives, with 4.0% of students reporting more regu lar child abuse in the home. Preliminary Analyses After examining existing research in the field it was expected that both bullying and peer victimization would be associated with gender and age, but not necessarily ethnicity. No a priori hypotheses were form ulated concerning the relationship between demographic variables and the witnessing of violence in the home. Preliminary analyses were conducted to determine the relationship be tween primary variable s of interest (peer bullying/victimization and violence in the hom e) and demographic variables. There was no significant association observed between gender and overall bullying behaviors ( = 1.87, p = .18). While literature in the area t ypically finds boys to demonstrate more physical bullying behaviors than girls, the scale utilized in this study examined both physical and relational bullying behaviors, wh ich allowed for an examination of gender differences by type of bullyi ng. Males were significantly more likely to engage in physical bullying ( = 18.91, p < .001) and females were found to engage significantly more in relational bullying ac tivities, such as gossiping ( = 49.78, p < .001). No significant relationship was found between peer victimization and gender. However, when physical and relational peer victimiza tion were examined separately, it was found
24 that males were more likely to be physically victimized ( = 17.6, p < .001) and females were more likely to be relationally victimized ( = 24.76, p < .001). The relationships between age and bullying behaviors and peer victimization were f ound to be significant, with children between the ages of thirteen and sixteen reporting both more bullying ( = 22.18, p < .01) and peer victimization ( = 16.77, p < .05). Additionally, there was a significant relationship found between bullyi ng and ethnicity, with African-Americans reporting proportionally more bullyi ng behaviors than Caucasians ( = 19.38, p < .01). However, the relationship between peer vic timization and ethnicity was not significant ( = 4.35, p = .63). In regards to witnessing adult-to-adult violence in the home, there were no significant gender differences in the obser vation of this type of violence ( = 2.64, p = .76). Furthermore, there was no significant ag e difference in the reporting of witnessing such violence ( = 45.06, p = .27). However, significant differences emerged between ethnicities ( = 51.36, p < .01), with more African-Americans and Hispanics reporting witnessing adult-to-adult violence in the home than Caucas ians. Furthermore, there were significant differences in reports of witnes sing adult violence in the home based on living situation ( = 83.06, p < .001). Adolescents living in a home with a single mother, or a mother and a stepfather/boyfriend were more likely to report witne ssing adult violence in the home than those living in a home with both parents. Primary Analyses The relationships between bullying/victim ization behaviors and witnessing adultto-adult violence in the home were assessed w ith logistic regressi on models. Logistic regression is commonly utilized when de pendent variables ar e dichotomous or
25 dichotomized by researchers. In this case, bullying was dichotomized into those students who regularly bullied others (sometimes, al most always, always) or did not regularly bully others (never or rarely). Peer victim ization was dichotomized similarly into those students who were regularly victimized by peer s and those who were not. Past research suggests that children who are classified as regular bullies have char acteristics that set them apart from children not considered to bully regularly (Olweus, 1993). Dichotomizing the variables in this fashion has been done in previous research and is considered legitimate for bullying/peer victim ization variables by othe r researchers in the field (see Olweus, 1993). Logistic regression is also appropriate in skewed and kur totic samples. While data from the bullying and peer victimization it ems in the present study approached normality, data from items asking students to report adult viol ence in the home were found to be highly skewed and kurtotic (skewness = 5.35, kurto sis = 35.87). In this situation, logistic regression is an appropriate analysis because it is not highly reliant on the normality of the data. Furthermore, logistic regression doe s not assume homoscedasticity or a linear relationship between independent and depende nt variables. Like other regression techniques, logistic regression can be used to predict categorical or continuous variables. The most commonly reported statistic is the O dds Ratio (OR), which we utilized in our analyses. For our sample, logistic regression was utilized to determine the impact of witnessing violence in the home on bullying behaviors and peer victimization while controlling for demographic variable s and child abuse in the home.
26 Examination of Hypothesis 1: Prediction of Bullying Behaviors from Witnessing Violence in the Home A logistic regression model was constructe d to identify whether witnessing adultto-adult violence in the home was predictive of bullying behavior while controlling for demographic variables and child abuse in the ho me. For the construction of this logistic regression model, variables were entered in three blocks. Demographic variables including ethnicity, gender, and grade were ente red into the first block. Also included in the first block was a variable querying students whom they live with the majority of the time. Responses were categorized according to whether they lived with both parents, one parent singly or one parent along with a si gnificant other or st epparent, or another individual. The second block in cluded the item assessing st udent report of harm of a child in the home. The third block include d the item that asked students to report whether or not a parent in the home had harmed another adult in the home. The dependent variable consisted of a dichotom ized score from the Schwartz Aggression Scale, with the midpoint of the scale being th e split point. Collinea rity of independent variables was examined through the tolerance and variance in flation factors. It was determined that multicollinearity was not a problem for this sample. The Hosmer-Lemeshow test indicated an adequate model fit ( p = .12). As predicted, witnessing adult-to-adult violence in the home was predictive of bullying behaviors. Each step contributed signifi cantly to the model at the .05 level, and the overall model was significant ( p < .001). Table 3-2 presents a summary of the hierarchical logistic regression model.
27 Table 3-2. Summary of Hi erarchical Logistic Re gression Analysis (H1) Model Step -2LL Model Test Block Test (df) Nagelkerke R Family Demographics 2055.59 14.67* 14.67* .013 Child Abuse In Home 2041.77 28.49*** 13.82*** .025 Witnessing Adult Violence 2036.03 34.23** 5.74*** .030 Notes: * p < .05, ** p < .01, *** p < .001. Results of this analysis indicated that ethnicity was the only demographic factor significantly associated with bullying behaviors in this sample. Other minorities (Hispanic, Asian/Pacific, Native American, and those of mixed/ot her origin) were found to be significantly more likely to engage in bullying behaviors than Caucasian students (OR = 2.03). Even after accounting for a number of de mographic and other related variables, witnessing or being a victim of child abuse in the home was signi ficantly related to overall bullying behavior (OR = 1.25). The odds that a student who endorsed child abuse in the home engaged in bullying behaviors in creased 25% at each step on our six-point Likert scale. In other word s, students who reported that an adult had harmed a child in the home a few times a week were considered to be 1.25 times more likely to engage in regular bullying behaviors than students who never witnessed child abuse in the home. As predicted, witnessing a parent harmi ng another adult in the home was also significantly associated with bullying behavior (OR = 1.37) . Students who witnessed adult-to-adult violence in the home were found to be 37% more likely to engage in bullying behaviors at each step on the Likert scale. Therefore, according to our model, the odds that a student who witnessed adult-to -adult violence in the home at least a few
28 times weekly engaged in regular bullying beha viors was 1.85 times that of a student who never witnessed such violence in the home. Table 3-3 presen ts data pertaining to beta weights, standard errors, odds ratios and 95% confidence interval s for each predictor variable in the final step of the logistic regression model. Table 3-3. Outcome Variables in Final Step of Logistic Regression Model (HI) SE OR (95% CI) African-American .13 .24 1.15 (.72-1.82) Hispanic/Other Male .71** .19 .26 .13 2.03 (1.223.38) 1.21 (.94-1.55) Lives with one parent .15 .26 1.16 (.70-1.91) Lives with foster parent/ Adopted parent/other .09 .25 1.10 (.67-1.78) Grade .03 .03 1.10 (.97-1.11) Child abuse in home .22* .10 1.25 (1.031.51) Witnessing Violence in home .32** .12 1.37 (1.081.74) Notes: * p < .05, **p < .01 For race and ethnicity variables, the reference group was Caucasian. For these variables, the referen ce group was living with both parents. Examination of H2: Prediction of Peer Victimization from the Witnessing of Violence in the Home The second logistic regression model wa s constructed to determine whether witnessing adult-to-adult violence in the home was predic tive of adolescen tsâ€™ reports of being victimized by peers. Firstly, the same demographic variables as utilized in the logistic model examining bullying were inserted into the first step of the model. These included ethnicity, sex, grade, and with wh om the student lives. The second block included the item assessing student report of ch ild abuse in the home. The third block included the item that asked students to repor t whether or not a parent in the home had harmed another adult in the home. The depe ndent variable consiste d of a dichotomized
30 The presence of child abuse in the home contributed significantly to the final model (OR = 1.53). The calculated odds ratios demonstr ated that, according to the model, those adolescents who reported child abuse in th e home were 2.5 times more likely to be victimized than those who had not re ported child abuse in the home. However, contrary to our hypothesis, it wa s determined that witnessing adult-toadult violence did not contri bute significantly to the model (OR = 1.23). Although not significant, odds ratios sugge sted that adolescents who reported regularly witnessing adult-to-adult violence in the home were 1.5 tim es more likely to be victimized by peers than those students who did not re port witnessing this type of violence. Table 5 presents data pertaining to beta weights, standard er rors, odds ratios and 95% confidence intervals for each predictor variable in the final step of the logistic regression model. Table 3-5. Outcome Variables in Final St ep of Logistic Re gression Model (H2) SE OR (95% CI) African-American .25 .93 1.29 (.77-2.13) Hispanic/Other .29 .29 1.34 (.76-2.36) Male .31* .14 1.36 (1.041.78) Lives with one parent -.54* .25 .58 (.36-.94) Lives with foster parent/ Adopted parent/other -.58* .24 .56 (.35-.89) Grade -.97* .04 .91 (.84-.98) Child abuse in home .43** .10 1.53 (1.251.87) Witnessing Violence in home .21 .13 1.23 (.95-1.59) Notes: * p < .05, **p < .01 For race and ethnicity variables, the reference group was Caucasian. For these variables, the referen ce group was living with both parents. Examination of H3: Moderated Regression Analyses In order to determine whether gender, age, or ethnicity were moderators of the relationship between witnessing adult-to-adul t violence in the home and bullying/peer
31 victimization, moderated regr ession analyses were conducte d. We first looked at the main effects of demographic variables, witnes sing of child abuse, and witnessing of adult violence. Next, interaction by product intera ction terms were put into the logistic regression models in order to test the possible inte raction effects of the above mentioned demographic variables with the relationshi p between witnessing a dult-to-adul t violence in the home and bullying or peer victimizati on. For both bullying and peer victimization, results revealed that neither gender, age, or ethnicity were significant moderators.
32 CHAPTER 4 DISCUSSION Summary of Principal Findings Of the students sampled, over 10% reported witnessing adult-to-adult violence in the home at least once in their lifetime. Approximately 3% of our sample reported witnessing this behavior on a regular basis. The prevalence of reported violence in the home according to results from the present st udy is lower than what might be expected given national averages for the prevalence of domestic violence (i.e., domestic violence is reported to be found in approximately 10 to 20% of American households). However, it is possible that adolescents may have underre ported witnessing this ty pe of violence due to the sensitivity of the question. Also, the inclusion of only one question assessing this type of violence may have contributed to the low rates found in this study. In our sample, approximately one-quarter of all students reported regular bullying behavior. This finding is similar to rate s reported in other studies examining the prevalence of bullying behavior (Seals & Young, 2003; Smith et al., 2000). However, bullying prevalence in our study was lower th an that found by Stoc kdale and colleagues (2000), who found that over 70% of a rural sa mple reported being involved in bullying behaviors. They suggested that perhaps bully ing behaviors were more prevalent in rural samples than urban or suburban samples. Ou r results are more cons istent with national samples that tend to find bullying behaviors to occur in one-quarter to one-half of all students. When physical and relational bully ing were examined separately, it was found that physical bullying was significantly more prevalent in males and relational bullying
33 was more prevalent in females. Crick a nd Grotpeter (1996) and Crick and Bigbee (1998) also found gender differences between re lational and overt bullying behaviors. Similar to bullying prevalence, 23% of students reported being physically or relationally victimized on a regul ar basis. This finding is also consistent with other research from the United States which sugge sts that between onequarter to over onethird of students are victimized by peer s on a regular basis (Seals & Young, 2003; Smith et al., 2000). Overall, males and females repor ted similar levels of peer victimization. However, when physical and relational victim ization were looked at independently, our results indicated that boys were more commonly physically victimized, while girls experienced more relational pe er victimization. Similar fi ndings have been reported by Baldry and Farrington (1999) and by Lagerspe tz, Bjorkqvist, and Pelt onen (1988). Boys reported higher levels of dire ct victimization in regards to both the threat of physical harm and being the victim of actual physical ha rm. Girls reported relational victimization such as peer exclusion and being the subjec t of gossip more often than boys in our sample. Past research, incl uding that by Crick and colleague s (1996), has indicated that girls are more likely to inflict relational harm on other girls, with the intent of causing psychological damage. In accordance with our first hypothesis, resu lts indicated that witnessing adult-toadult violence in the home wa s predictive of bullying, alt hough our logistic regression based models did not fully predict bullying be havior, based on low Nagelkerke R values. Although logistic regression doe s not directly calculate a variable accounting for explained variance, the Nagelkerke R estim ate, a measure of the pseudovariance (see Tables 2 and 4), suggests that the explained variance for the overall model was low. This
34 low Nagelkerke R estimate indicates that ot her factors play a signi ficant role in the prediction of bullying behavior s in addition to witnessing ad ult-to-adult violence in the home. Accordingly, the Odds Ratios found in th is study should be in terpreted with this in mind, as the overall model likely explains less than 10% of the variance in the prediction of bullying behaviors. The odds that a child who had witnessed violence in the home reported bullying other st udents was significantly high er than for those students who had not witnessed violence in the home. However, contrary to our second hypothesi s, results of the present study did not indicate a significant predictive relationshi p between witnessing adult-to-adult violence in the home and peer victimization after co ntrolling for demographi c variables and child abuse in the home. Initially, we hypothesized that peer vic timization at school could be an indicator of individual difficu lties as a result of exposure to violence in the home. For example, one might argue that exposure to violence in the home could reduce an adolescentâ€™s ability to be a ssertive if victimized at school . However, this relationship was not significant, indicati ng that perhaps witnessing vi olence in the home does not directly predict an increased risk for peer victimization. On the other hand, this result may be a function of our singular question on the witnessing of violence in the home, which may not have been comprehensive e nough to reveal a predictive relationship between adult-to-adult vi olence in the home and peer victimization. Our finding, that witnessing adult-to-adult vi olence in the home did not predict peer victimization, is contrary to the results found in Baldryâ€™s (2003) Italian study. She found domestic violence to be related to peer vi ctimization both in and out of school. She utilized similar criteria for peer victimi zation, although she measured peer victimization
35 with the Olweus Bullying Scale (1993). Similar to the present study, she also controlled for the effects of child abuse in the ho me on adolescent victimization. Questions concerning domestic violence in the home were more specifically delineated than the questions utilized in our analyses. As me ntioned above, it is possi ble that our singular question was not sensitive enough to reveal a predictive relationship between adult violence in the home and peer victimiza tion. The populations examined in the two studies were very different as well. Baldry examined these behaviors in an urban, Italian sample, whereas we looked at victimization a nd violence in the home in a rural American sample. It is thus possible that the imp act of witnessing violence in the home has a cultural component, which may also account fo r the differing results between these two studies. Child abuse in the home was a significan t predictor of both bullying and peer victimization in our adolescent sample. These findings are consistent with other literature in the field, incl uding the Baldry (2003) study th at found child abuse in the home to significantly predict adolescent bullying behaviors. Finally, analyses looking at the potential moderating effect s of gender, age, and ethnicity did not reveal any significant moderators, which was contrary to our third hypothesis that gender and age may be significant moderators. The Social Cognitive Theory developed by Bandura (1973) posits that behavior is transmitted through observation by one individual to another. The results of this study, indicating that witnessing adult-to-adult violence in the home predicted bullying behavior, could be explained according the Soci al Cognitive Theory. In other words, it is possible that the adolescents in our sample who witnessed violence in the home learned
36 to express anger through aggression, and that this anger was later manifested in bullying behaviors. Research Implications The results of this study cont ribute to literature in the ar ea because this is the first such study to examine the relationship between adult-to-adult violen ce in the home and adolescent bullying and peer victimization in an American sample, as well as in a rural setting. Furthermore, because the literature base in the research areas of bullying and victimization contains a number of studies utilizing European samples, studies such as this one conducted with American samples are important contribu tors to our knowledge on the prevalence and contributi ng factors of these dangerous behaviors that have such negative consequences for youth. In addition, th is study brings atten tion to the potential impacts of adult-to-adult viol ence for American youth. A practical implication of th is studyâ€™s findings is that the discovery of adult-toadult violence in the home should prompt observation and enquiry concerning the adolescentâ€™s behavior at sc hool, because adolescents who wi tness violence at home have been shown to be more likely to be perpetra tors of aggression them selves. Professionals in the community, including teachers, social workers, doctors, and staff in shelters for battered or abused women and their families should be made aware of the potential connection between family violence and children â€™s behavior. As outlined in the literature review for this study, childrenâ€™s violent beha vior at school results in multiple negative outcomes for both the aggressors and the victims of this behavior. Thus, it is important to take steps to identify risk factors for thes e behaviors and intervene to decrease them. Alternatively, if school officials notice th at an adolescent is bullying others at school they should enquire abou t the adolescentâ€™s home environment. An adolescent
37 who is physically or relationally aggressive towards other students may be learning that behavior from what is seen or experienced by the adolescent in the home environment. As with other risky youth behavi ors, it is important to inte rvene early, both in regards to preventing and stopping adult vi olence in the home and in intervening with youth to prevent bullying and peer victimization. While this study did not find a significant relationship betw een witnessing adult-toadult violence in the home and peer victimiza tion, results indicated that child abuse in the home significantly impacted peer victimizati on, as well as bullying. Therefore, similar practical implications apply in regards to notifying relevant child professionals in the community that one risk factor for peer vic timization and bullying may be child abuse in the home. Therefore, those children who ar e victimized at school should also be asked about their home environment. In addition to the practical implications outlined above, the findings from this study may assist in the development of school prev ention programs. The results of this study suggest that adolescents may imitate learned behavior, as implicated by Albert Banduraâ€™s Social Cognitive Theory. Accordingly, students may be able to learn more appropriate behaviors. Anti-bullying programs, such as the Olweus anti-bullying program, are currently being implemented in schools. Th is school-based program has been shown to reduce school bullying by as much as 50% (Olweus, 1993). The program utilizes aspects of the Social Cognitive Theory, such as obs ervational learning, to influence student behavior. Interventions such as this one may be able to help students learn different, more adaptive ways of interacting sociall y. In addition, early intervention for younger children could help teach children to change a ggressive or antisocial behaviors and show
38 them how to express their anger more c onstructively. Prevention and intervention programs should be informed by professionals who work with children both inside and outside the school environment. Social work ers, doctors, and individuals working in the area of domestic violence may all be able to contribute to the cr eation of intervention programs for children who may have b een exposed to violence in the home. Limitations The current study had several limitations. First, a causal re lationship between witnessing adult-to-adult violence in the home and bullyin g/peer victimization could not be established within the confin es of this study. However, it is less likely that bullying or victimization at school cause adult violence to start in th e home, although child behavior problems may exacerbate parental conflict. Following from Banduraâ€™s Social Cognitive Theory, it is more likely that violence obser ved in the home precedes and is learned by children and adolescents, who then bully others or become victims of bullying. However, witnessing adult violence in the home and/ or child abuse in the home does not fully explain the complex phenomenon of bullying or peer victimization. Other factors, such as parent-child relationship, neighborhood viol ence, and peer group in fluences are also thought to contribute to bully ing and peer victimization. Another limitation stems from the nature of the Schwartz Bullying Scale and Social Experience Questionnaire-Self Re port. These measures are se lf-report scales that were completed by the adolescents during school. Id eally, corroborating data would have been collected from other sources, such as parent s, teachers, or peers. However, other investigators have found that childrenâ€™s sel f-report data is among the most reliable method of disclosing bullying (Solberg & Ol weus, 2003), victimization, and aggressive behavior ( Khatri, Kupersmidt, & Patterson, 2000). In fact, it has been suggested that
29 total score from the Social Experience Qu estionnaire-Self-Report (which assessed both physical and relational aggression ), with the midpoint of the scale being the split point. Collinearity of independent variables was ex amined through the tole rance and variance inflation factors. It was determined that multicollinearity was not a problem for this sample. The Hosmer-Lemeshow Test indicat ed an adequate model fit ( p = .32). It was determined that each step in the model cont ributed significantly at the .05 level. The overall model proved significant ( p < .000). Table 3-4 presents a summary of the hierarchical logistic regression model. Table 3-4. Summary of Hi erarchical Logistic Re gression Analysis (H2) Model Step -2LL Model Test Block Test (df) Nagelkerke R Family Demographics 1413.49 20.17** 20.17** .022 Child Abuse In Home 1381.52 52.13*** 32.97*** .056 Witnessing Adult Violence 1379.12 54.53** 2.4 ..058 Notes: * p < .05, ** p < .01, *** p < .001. Several of the demographic variables proved to be significant additions to the final model. Boys were significantly more likel y to be victimized than girls (OR = 1.36). Furthermore, children in lower grades were mo re likely to be victimized than children in higher grades (OR = .91). Finally, living with one parent (OR = .58) or foster/adoptive parents (OR = .56) significantly decreased th e reporting of peer vi ctimization in our sample.
39 child self-report on bullying and peer victimizat ion may be more accurate than the reports of parents and teachers who may not be aware of the prevalence of these behaviors both in and out of school. Another limitation of this st udy was that the question rega rding witnessing of adult violence in the home was limited in its scope and breadth and may not have captured all violence in the home. Our question was limited to harm afflicted by a parent or guardian on another adult in the home. This question did not capture violence that may not have caused harm or abuse that was not inflicted by a parent or guardian. However, since the question on adult-to-adult violen ce was part of an existing lengthy questionnaire used for evaluation purposes, it was not possible to a ssess all potential forms of violence in the home. This question involved self-report on th e part of the adoles cent, which introduces the possibility that students may have unde r-reported this sensitive information. Furthermore, social desirability may have a ffected student ratings. However, it was not possible to determine whether so cial desirability impacted student reports, as it was not measured in the current study. Another limita tion is that our question regarding violence in the home did not assess at what age the adolescent witnessed the violence. Past research (Wolfe, 1986) has indicated that r ecent witnessing of domestic violence may result in poorer outcomes for children than the witnessing of violence in the more distant past. Another study by Hughes (1988) suggested that witnessing violence in preschool may have unique effects on child behavior compared to witnessing violence at later developmental stages. An additional limitati on of the question assessing violence in the home was that it utilized the word â€œharmâ€ but did not include the word â€œviolence.â€ It is possible that adolescents may have interprete d the meaning of â€œharmâ€ in different ways
40 when responding to this question. For exampl e, adolescents participating in this study may have witnessed an adult hitting another ad ult, but not interprete d that action as harm if the other adult did not sustain any injury. Finally, another limitation is that the ques tions utilized were embedded in a larger evaluation instrument examining adolescent risk behaviors. Because the questions were not asked independently of ot her questions concerning risky behaviors (such as drug and alcohol use), it is possible th at these other questions may have influenced how students responded to the questions used for this study, which may have impacted their validity. Future Directions The limitations associated w ith this research highlight the importance of further research in this area utilizing additional re liable and valid measurements of both child abuse in the home and adult-to-adult violence in the home. Instruments such as the Conflict Tactic Scale (Straus, 1979) have been developed for this purpose and could be used for follow-up analyses of these students. Utilizing reliable a nd valid measurements of violence in the home could provide more information on the prevalence of such violence and itsâ€™ association with bullying a nd peer victimization. Given the results of our present study, it is hypothesized that O dds Ratios would be larger with better measurement of family violence. Furthermore, for future analyses we also intend to assess other potential moderators of the association between witnessing violence in the home and bullying/peer victimization. Potential moderators include participation in an after school activity, attendance at church, participati on in athletic activities, and ov erall parental involvement. In addition, further research should assess child -on-child violence, or sibling violence, in
41 the home. Sibling aggression may interact with, or complicate th e relationship between, the witnessing of adult-to-adult violence and bullying or peer victimization. This study provides initial support for a rela tionship between witnessing violence in the home and bullying in a rural American samp le. Future studies should investigate the impact of other family factors on bullying and victimization, specifically in rural areas. Past research has indicated that family vari ables such as cohesion are related to violent behavior in youth (Bowers et al., 1992). Interpersonal st yles of relating may also contribute significantly to bullyi ng behavior. In addition to fam ily factors, future research should look at the impact of ot her variables that may be rela ted to bullying, such as the witnessing of neighborhood violence and school clim ate. Other variables of interest that have been shown to be associated with bul lying behavior include the contextual peer group (Espelage et al., 2003) and exposure to media violence. Assessing these types of variables in future research is important , as they may explai n additional variance associated with bullying behaviors. Finally, future research could identify if there is a true cause and effect relationship between witnessing adult-to-adult violen ce in the home and bullying or peer victimization by incorporating longitudinal designs. Children who are known to have been exposed to violence in the home could be tracked in order to determine if these children later engage in bullying behaviors or are victimized by their peers. Such research would represent a novel and important step towa rds acknowledging the specific influences of witnessing adult violence in th e home on bullying and peer victimization. Future research should be conducted with both urban and rural samples. If causal
42 connections are made between witnessing vi olence in the home and bullying in rural areas, it may turn more attention to the needs of these communities. Conclusion In conclusion, our findings demonstrate th at witnessing adult-to-adult violence in the home is predictive of bullying behavior, a lthough it is clearly not the only influential variable. Results of the curre nt study highlight a need to acknowledge the influence of the home environment on bullying behaviors. W ith the recent increase in attention being paid to school bullying and itsâ€™ dangerous out comes, both for perpetrators and victims, it is apparent that future resear ch should be directed towards identifying predictors of these behaviors and curbing them th rough well-crafted interventions. Furthermore, it is our hope that rural areas will not be neglected in either research or intervention, as the current study demonstrates that these dangerous phenomen a exist in rural areas at similar rates as urban and suburban areas. In general, fi ndings from the present study offer further insight into the correlates of bullying and peer victimization, and may help to frame solutions to this problem in the future.
43 LIST OF REFERENCES Andreou, E. (2001). Bully/Victim Probl ems and Their Association with Coping Behaviour in Conflictual Peer Inte ractions among School-Age Children. Educational Psychology, 21(1), 59-66. Austin, S., Joseph & S. (1996). Assessmen t of Bully/Victim pr oblems in 8-11 YearOlds. British Journal of Educa tional Psychology, Vol 66(4), 447-456. Baldry, A., & Farrington, D. (2000). Bulli es and Delinquents: Pe rsonal Characteristics and Parental Styles. Journal of Comm unity & Applied Social Psychology, 10(1), 17-31. Baldry, A. (2003). Bullying in Schools and Exposure to Domestic Violence. Child Abuse and Neglect, 27, 713-732. Bond, L., Carlin, J., & Thomas, L. (2001). Does Bullying Cause Emotional Problems? A Prospective Study of Young Teenagers. British Medical Journal, 323, 480-484. Bosworth K, Espelage, D., and Simon T. (1999). Factors Associated with Bullying Behavior among Early Adolescents. J ournal of Early Adolescence, 19, 341-362. Bowers, L., Smith, P., & Binney, V. (1992). Cohesion and Power in the Families of Children Involved in Bully/Victim Problem s at School. Journal of Family Therapy, 14(4), 371-387. Craig, W. (1998). Bullying, Victimizat ion, Depression, Anxiet y, and Aggression in Elementary School Children. Journal of Personal and Individual Differences, 2, 123-130. Crick, N. (1996). The Role of Overt A ggression, Relational Aggression, and Prosocial Behavior in the Prediction of Childre n's Future Social Adjustment. Child Development, 67(5), 2317-2327. Crick, N. & Grotpeter, J. (1996). Children' s Treatment by Peers: Victims of Relational and Overt Aggression. Developmen t and Psychopathology, 8(2), 367-380. Crick, N. and Bigbee, M. (1998). Relationa l and Overt Forms of P eer Victimization: A Multi-informant Approach. Journal of Consulting and Clinical Psychology, 66, 337-347.
44 Crick, N., Werner, N., & Casas, J. ( 1999). Childhood Aggressi on and Gender: A New Look at an Old Problem. In Bernstei n, Dan (Ed.), Gender and M otivation (pp. 75141). Lincoln NE: University of Nebraska Press. Duncan, R. (1999). Maltreatment by Parent s and Peers: The Relationship Between Child Abuse, Bully Victimization, and Psychol ogical Distress. Child Maltreatment: Journal of the American Professional So ciety on the Abuse of Children, 4(1), 4555. Ehrensaft, M., Cohen, P., & Brown, J. (2003). Intergenerational Transmission of Partner Violence: A 20-year Prospective St udy. Journal of Consulting and Clinical Psychology, 71(4), 741-753. Eslea, M. & Rees, J. (2001). At What Ag e are Children Most Likely to be Bullied at School? Aggressive Behavior, 27(6), 419-429. Espelage, D., Holt, M., Henkel, R. (2003) . Examination of Peer-Group Contextual Effects on Aggression. Child Development, 74 (1), 205-220. Fantuzzo, J. & Mohr, W. (1999). Prevalence and Effects of Child Exposure to Domestic Violence. Domestic Violence and Children, 9(3), 21-32. Farrington, L. and Ely, C. (1993). Social Conf lict Theories of Family. Sourcebook of Family Theory and Methods: A Contextual Approach. New York: Plenum Press. Fosse, G. & Holen, A. (2002). Childhood Environmen t of Adult Psychiatric Outpatients in Norway Having Been Bullied in School . Child Abuse & Neglect, 26(2), 129137. Hughes, H. (1988). Psychological and Behavi oral Correlates of Family Violence in Child Witnesses and Victims. American Journal of Orthops ychiatry, 58(1), 77-90. Jaffe, P., & Wolfe, D. (1990). Children of Battered Women. Thousand Oaks, CA: Sage Publications. Johnson M., and Elliot, B. (1997). Domes tic Violence Among Family Practice Patients in Midsized and Rural Communities. Journal of Family Practice, 44, 392-400. Kalmuss, D. (1984). The Intergenerational Transmission of Marital Aggression. Journal of Marriage & the Family, 46(1), 11-19. Kaukinen, C. (2002). The Help-Seeking D ecisions of Violent Crime Victims: An Examination of the Direct and Conditi onal Effects of Gender and the VictimOffender Relationship. Journal of In terpersonal Violence, 17(4), 432-456. Kershner, M., Long, D., & Anderson, J. (1998). Abuse Against Women in Rural Minnesota. Public Health Nursing, 15(6), 422-431.
45 Khatri, P., Kupersmidt, J. & Patterson, C. ( 2000). Aggression and Peer Victimization as Predictors of Self-Reported Behavioral and Emotional Adjustment. Aggressive Behavior, 26(5), 345-358. Krishnan, S., Hilbert, J., and VanLeeuwen, D. (2001). Domestic Violence and HelpSeeking Behaviors among Rural Women: Results from a Shelter-Based Study. Family Community Health, 24, 28-38. Kumpulainen, K. & Rsnen, E. (2000). Ch ildren Involved in Bullying at Elementary School Age: Their Psychiatric Sympto ms and Deviance in Adolescence: An Epidemiological Sample. Child Abuse & Neglect, 24(12), 1567-1577. Ladd B. and Ladd, G. (2001). Variations in Peer Victimization: Re lations to Childrenâ€™s Maladjustment. J. Juvonen and S. Graham (Eds.) Peer Harassment in School: The Plight of the Vulnerable and Victim ized (pp.25-48). New York: Guildford. Levendosky, A., Huth-Bocks, A., and Semel, M. (2002). Adolescent Peer Relationships and Mental Health Functioning in Families with Domestic Violence. Journal of Clinical Child Psychology, 31, 206-218. Logan, T., Walker, R., Cole, J., Ratliff, S., and Leukefeld, C. (2003). Qualitative Differences Among Rural and Urban Intimat e Violence Victimization Experiences and Consequences: A Pilot Study. Journal of Family Violence, 18, 83-92. Lowenstein, L. F. (1978). The Bullied and Nonbullied Child: II. Home & School, No 13, 3-4. Ma, X. (2002). Bullying in M iddle School: Individual and School Characteristics of Victims and Offenders. School Effectiv eness and School Improvement, 13(1), 6389. McConville, D. and Cornell, D. (2003). Aggressive Attitudes Predict Aggressive Behavior in Middle School Students. Journal of Emotional and Behavioral Disorders, 11, 179-187. Miller, T., Handal, P., Gilner, F. (1991). The Relationship of Abuse and Witnessing Violence on the Child Abuse Potential Inve ntory with Black adolescents. Journal of Family Violence, 6(4), 351-363. Newman, R., Murray, B., & Lussier, C. (2001). Confrontation with Aggressive Peers at School: Students' Reluctance to Seek Help from the Teacher. Journal of Educational Psychology, 93(2), 398-410. O'Brien, M., John, R. & Margolin, G. (1994) . Reliability and Diagnostic Efficacy of Parents' Reports Regarding Children's Exposure to Marital Aggression. Violence and Victims, 9(1), 45-62.
46 Olweus, D. (1993). Bullies on the Playground : The Role of Victimization. In: Children on Playgrounds: Research Perspectives a nd Applications. Hart, Craig H.; Albany, NY, US: State University of New York Press, 85-128. Paquette, J. & Underwood, M. (1999). Ge nder Differences in Young Adolescents' Experiences of Peer Victimization: So cial and Physical Aggression. MerrillPalmer Quarterly, 45(2), 242-266. Pellegrini, A., and Barini, M. (2000). An Empirical Comparison of Methods of Sampling Aggression and Victimization in Sc hool Settings. Journal of Educational Psychology, 92, 360-366. Perry, D., Kusel, S., & Perry, L. (1988). Victims of Peer Aggression. Developmental Psychology, 24(6), 807-814. Prinstein, M., Boergers, J., & Vernberg, E. (2001). Overt and Relational Aggression in Adolescents: Social-psychological Adjustment of Aggressors and Victims. Journal of Clinical Child Psychology, 30(4), 479-491. Rigby, K. (1994). Psychosocial Functioning in Families of Australian Adolescent Schoolchildren Involved in Bully/Victim Pr oblems. Journal of Family Therapy, 16(2), 173-187. Roecker-Phelps, C. (2001). Childrenâ€™s Res ponses to Overt and Re lational Aggression. Journal of Clinical and Child Psychology, 30, 240-252. Seals, D., & Young, J. (2003). Bullying and Victimization: Prevalence and Relationship to Gender, Grade Level, Ethnicity, Se lf-Esteem, and Depression. Adolescence, 38(152), 735-747. Smith, P., Morita, Y., Junger-Tas, J., Olweus, D., Catalano, R., and Slee, P. (1999). The Nature of School Bullying. A Cross-Na tional Perspective. London: Routledge. Smith, Peter K.; Myron-Wilson, R. (1998). Parenting and School Bullying. Clinical Child Psychology and Psychiatry, 3(3), 405-417. Solberg, M. & Olweus, D. (2003). Preval ence Estimation of School Bullying with the Olweus Bully/Victim Questionnaire. Aggressive Behavior, 29(3), 239-268. Sourander, A., Helstela, L. (2000). Persistence of Bullying from Childhood to Adolescence-A Longitudinal 8-Year Follo w-Up Study. Child Abuse and Neglect, 24, 873-881. Stockdale, M., Hangaduambo, S., Duys, D., La rson, K., & Sarvela, P. (2002) Rural Elementary Studentsâ€™, Parentsâ€™, and Teacher sâ€™ Perceptions of Bullying. American Journal of Health Behavior, 26, 266-277.
47 Storch, E. & Masia-Warner, C. (2004). Correlates of Victimization in Chinese Childrenâ€™s Peer Groups. Developmental Psychology, 37, 520-532. Straus, M. (1979). Measuring Interper sonal Conflict and Vi olence: The Conflict Violence Scale. Journal of Ma rriage and the Family, 41, 75-88. Tanaka, T. (2001). The Identity Forma tion of the Victim of â€œshunning.â€ School Psychology International, 22(4), 463-476. Unnever, J. & Cornell, G. (2004). Middl e School Victims of Bullying: Who reports Being Bullied? Aggressive Behavior, 30, 373-388. Wekerle, F. & Gelles, R. (1998). Pare nting and School Bullying. Clinical Child Psychology and Psychiatry, 3, 405-411. Wekerle, C. &Wolfe, D. (1998). The Role of Child Maltreatment and Attachment Style in Adolescent Relationship Violence. Development and Psychopathology, 10(3), 571-586. Widom, C. (1989). The Cycle of Violence. Science, 244, 160-166. Yoon, J., Barton, E. & Taiariol, J. (2004). Relational Aggression in Middle School: Educational Implications of Developm ental Research. Journal of Early Adolescence, 24(3), 303-318.
48 BIOGRAPHICAL SKETCH Jennifer Sivinski was born in Gainesvill e, FL. She attended Barnard College, Columbia University, where she received a ba chelorâ€™s degree in psychology and a minor in education. The following year, Jennifer entered the Peace Corps where she served in Paraguay for two years. Jennifer is currently in a Ph.D. program at the University of Florida where she aspires to receive a doctora l degree in clinical and health psychology.