Fostering healthy teen intimate relationships through an in-school violence prevention program

The sample of the Relationships Without Violence (RWV) Prevention Program consists of ninth- through twelfth-grade students in one Salt Lake City high school health class. The RWV was conducted between the Fall of 2008 to Fall of 2010 academic year. The RWV was designed to fit the curriculum in health classes regarding dating violence, consequences of dating violence, healthy ways to deal with conflict, and healthy relationships. The RWV curriculum consists of four different sessions, which covered the topics of sexual coercion, violent behaviors, violent attitudes, socialization of violence, and substance abuse while promoting prosocial behaviors, positive peer culture, healthy masculinities/femininities, consent, and help-seeking behaviors. The RWV used a quasi-experimental design with a control group and treatment group. Pretests were administered before the program began and posttests were administered 2 weeks to 1 month after the program ended. Participants experienced physical intimate partner violence (33.7%), emotional/psychological violence (46.2%), followed by sexual coercion at 16.5%. Additionally, boys and girls did not differ on intimate partner violence (IPV) perpetration and victimization. Results reveal that there was no significant difference from pretest to posttest when comparing control group and treatment group. However, multivariate analysis found that controlling behaviors significantly predicted violent attitudes, which significantly predicted partner violence and victimization. Additionally, results suggest that controlling behaviors partially mediated the relationship between violent attitudes and IPV perpetration and IPV victimization. Interestingly, when controlling for controlling behaviors, violent attitudes had a negative effect on female IPV perpetration and IPV victimization. Findings from the study reveal the powerful effects of controlling behaviors on adolescents, and the difference between violent attitudes only and violent behaviors to an intimate partner. The study revealed that coercive control did mediate the predictive relationship between controlling behaviors on IPV perpetration and IPV victimization. Interestingly, if the participants had controlling behaviors and only had violent attitudes, their IPV perpetrations reduced. These findings suggest that coercive control not only leads to violent perpetration but also leads to violent victimization. Professionals working with adolescents should be aware of the adverse consequences of dating violence that may negatively affect academic achievement. The findings of this study reveal that controlling behaviors present in adolescent dating relationships must be addressed by social workers working with adolescent populations.

FOSTERING HEALTHY TEEN INTIMATE RELATIONSHIPS THROUGH AN IN-SCHOOL VIOLENCE PREVENTION PROGRAM by Peter Ronald Fawson A dissertation submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Doctor of Philosophy College of Social Work The University of Utah August 2012 Copyright© Peter Ronald Fawson 2012 All Rights Reserved Th e Uni v e r s i t y o f Ut a h Gr a dua t e S cho o l STATEMENT OF DISSERTATION APPROVAL The dissertation of Peter Ronald Fawson has been approved by the following supervisory committee members: David S. Derezotes , Chair 4-9-2012 Date Approved Jannah Mather , Member 4-9-2012 Date Approved Hank Liese , Member 4-9-2012 Date Approved Scott Boyle , Member 4-9-2012 Date Approved Jonathan Ravarino , Member 4-9-2012 Date Approved and by Jannah Mather , Chair of the Department of College of Social Work and by Charles A. Wight, Dean of The Graduate School. ABSTRACT The sample of the Relationships Without Violence (RWV) Prevention Program consists of ninth- through twelfth-grade students in one Salt Lake City high school health class. The RWV was conducted between the Fall of 2008 to Fall of 2010 academic year. The RWV was designed to fit the curriculum in health classes regarding dating violence, consequences of dating violence, healthy ways to deal with conflict, and healthy relationships. The RWV curriculum consists of four different sessions, which covered the topics of sexual coercion, violent behaviors, violent attitudes, socialization of violence, and substance abuse while promoting prosocial behaviors, positive peer culture, healthy masculinities/femininities, consent, and help-seeking behaviors. The RWV used a quasi-experimental design with a control group and treatment group. Pretests were administered before the program began and posttests were administered 2 weeks to 1 month after the program ended. Participants experienced physical intimate partner violence (33.7%), emotional/psychological violence (46.2%), followed by sexual coercion at 16.5%. Additionally, boys and girls did not differ on intimate partner violence (IPV) perpetration and victimization. Results reveal that there was no significant difference from pretest to posttest when comparing control group and treatment group. However, multivariate analysis found that controlling behaviors significantly predicted violent attitudes, which significantly predicted partner violence and victimization. Additionally, results suggest iv that controlling behaviors partially mediated the relationship between violent attitudes and IPV perpetration and IPV victimization. Interestingly, when controlling for controlling behaviors, violent attitudes had a negative effect on female IPV perpetration and IPV victimization. Findings from the study reveal the powerful effects of controlling behaviors on adolescents, and the difference between violent attitudes only and violent behaviors to an intimate partner. The study revealed that coercive control did mediate the predictive relationship between controlling behaviors on IPV perpetration and IPV victimization. Interestingly, if the participants had controlling behaviors and only had violent attitudes, their IPV perpetrations reduced. These findings suggest that coercive control not only leads to violent perpetration but also leads to violent victimization. Professionals working with adolescents should be aware of the adverse consequences of dating violence that may negatively affect academic achievement. The findings of this study reveal that controlling behaviors present in adolescent dating relationships must be addressed by social workers working with adolescent populations. I dedicate this dissertation to my loving wife Rachel Wright, who has given countless amounts of time and energy to make this possible; I could not have done this without you. You have been my steady and calming voice through the complexities of my dissertation. Additionally, I dedicate this dissertation in loving memory to my late father, Ronald Charles Fawson (1950-2006) TABLE OF CONTENTS ABSTRACT …………………………………………………………………………....iii ACKNOWLEDGMENTS …………………………………………………………..viii Chapter 1. INTRODUCTION…..…...………………………………………………… ……..1 Background…..……………………………………………………………………1 Theoretical Perspectives to Intimate Partner Violence…………………………....4 Studies Supporting the Feminist Perspective……………………………...4 Research on Family Violence Perspective………………………………...5 Literature Supporting Johnson's Control Typology of IPV....……………6 The Present Study….……………………………………………………….……..8 2. LITERATURE REVIEW……………………………………………………………. 9 Rates of Intimate Partner Violence………………………………………………..9 Intimate Partner Violence and Controlling Behaviors…………………...10 Intimate Partner Violence among Adolescents………………………......11 Teen Intimate Partner Violence Prevention Programs…………………...13 Reviews of Adolescent Violence Prevention Programs…………………16 Specialized Features of the Relationships Without Violence Program…….........23 Theory of Intimate Partner Violence……………………………………….........24 Family Violence Perspective of Intimate Partner Violence…….………..25 Feminist Perspective on Intimate Partner Violence………………….…..27 Johnson's Control Typology of Intimate Partner Violence………...........28 Purpose of This Study………………………………………………………....34 vii 3. METHODS…………………………………………………….………….................36 Population…………………………………………………………………….37 Participants…..……………………………………………………………….38 Study Design...……………………………………………………………….38 Measures……………………………………………………………………..39 Demographics Page..…………………………………………………….…..40 Revised Controlling Behaviors Scale …………………………………….….40 Justification of Dating Violence Scales…..…………………………….……42 Dating Violence Scale-Victimization …………………………………….….42 Statistical Analysis Methods…………………………………………….......43 4. FINDINGS……………………………………………………………………...........45 Descriptive Statistics ……………………………………………………………45 K-means Cluster Analysis……….............................................................47 ANCOVA…..………………………………………………………........48 Structure Equation Modeling ……………………………………………49 Male Model…………………….………..……………………………….55 Female Model ………………………….………………………………...55 5. DISCUSSION………………………………………………………………..………57 Summary of Findings ……………………………………………………………57 Limitations of the Study…..……………………………………………………...62 Strengths of the Study ……………………………………………………………64 Implications for Social Work Practice…………………………………………...65 Implications for Social Work Policy …………………………………...……….68 Implications for Social Work Research…..………………………………..........69 APPENDIX: PRE- AND POSTTEST INSTRUMENTS…………..………………….71 REFERENCES…………………………………………………………….…………….75 ACKNOWLEDGMENTS I am grateful to the University of Utah PhD Program at the College of Social Work. My committee members' time, guidance, and patience mean a great deal to me. Thank you to Dr. David Derezotes, my chair, and Drs. Jannah Mather, Scott W. Boyle, Hank Liese, and Jonathan Ravarino. I am especially grateful to Michael Tanana from the Utah Criminal Justice Center for his expertise in statistical analysis. This dissertation depended on tireless efforts from community members dedicated to ending and preventing violence, Cindy Taylor and Heather Stringfellow of the Rape Recovery Center in Salt Lake City. I am thankful to Jenne Hamlin from West High School in Salt Lake City, for her endless service promoting healthy relationships with youth and for believing in this project from inception. CHAPTER 1 INTRODUCTION Background Adolescents are dying for lack of violence prevention programs. While crime has declined across the United States, dating violence persists with significant negative consequences for youth, with severe impact on morbidity and mortality (Whitaker et al., 2006). Research also indicates that aggressive behaviors in childhood are considered a risk factor for future violence and criminal behavior in adulthood (Mytton, DiGuiseppi, Gough, Taylor, & Logan, 2009). Violence prevention programs, with skill-building strategies targeting adolescents, such as school-based programs, are considered to be promising interventions (Avery-Leaf,Cascardi, O'Leary, & Cano, 1997; Mytton et al., 2005). For these reasons, most social workers assert that the provision of prevention programs that address partner violence is a form of social justice. The National Coalition Against Domestic Violence (NCADV, 1996) defines domestic violence as the act of one partner utilizing a pattern of behavior to maintain control over another person within an intimate, loving, and dependent relationship. However, intimate partner violence (IPV) expands the definition of violence against a partner to include verbal aggression, physical aggression, and/or sexual violence perpetrated by a current or former spouse, cohabitating partner, or dating partner (Straus 2 & Gelles, 1990). For the purpose of this dissertation, the term IPV will be used when referring to partner violence. There is a plethora of research exposing negative consequences of IPV. An international study concluded that 1 in 4 women report sexual violence from an intimate partner, and 1 in 3 girls report having been forced into the initiation of a sexual encounter (WHO, 2002). The National Crime Victimization Survey further revealed that 70% of women who were victims of sexual violence reported that the perpetrator was someone known to her, either an intimate, a relative, or a friend (Bureau of Justice Statistics, 2005). IPV can result in additional negative consequences, such as physical injury, sexually transmitted infections (STIs), depression, posttraumatic stress disorder, anxiety, and substance abuse (Campbell, 2002; Plichta, 2004; Prospero & Fawson, 2009). In the U.S., the National Violence Against Women Survey purports that an astounding 25% of women have been victims of physical violence (Tjaden & Thoennes, 2000), while another study found that 27% of women experience sexual victimization in their lifetime (Koss, Gidcycz, & Wisniewski, 1987). Violence against women is also high regionally; for example, in the state of Utah, 1 in 3 women will experience some form of sexual violence by a man in their lifetime (Utah Commission on Criminal and Juvenile Justice, 2005). The literature regarding adolescent samples is limited if not nonexistent. Although women are more likely to be IPV victims, men can also be victims of IPV. Consequences of IPV are not limited to women. In the U.S., between 39% and 50% of all IPV instances each year involve male victims (Kessler, Molnar, Feurer, & Appelbaum, 2001). In Utah, it is estimated that 1 in 7 boys will be sexually assaulted 3 before the age of 18 (Utah Commission on Criminal & Juvenile Justice, 2005). These findings suggest that the prevalence rates for both females and males are high. Therefore, prevention programs aimed at preventing IPV may help those suffering the effects of violence. Prevention programs may not only provide skills to prevent future perpetration, but will additionally promote help-seeking behaviors for those who already suffer from IPV (Cornelius & Resseguie, 2006; Whitaker et al. 2005). Research on prevention programs has demonstrated effectiveness in changing violent attitudes among adolescents. Recent efforts have addressed adolescents' knowledge about IPV, violent attitudes, and aggressive behaviors through violence prevention programs (Edelen, McCaffrey, Marshall, & Jaycox, 2009; Whitaker et al., 2006). Notably, researchers have discovered that multiple in-class sessions help students decrease attitudes that justify the use of dating violence as a means to resolve conflict (Avery-Leaf, Cascardi, O'Leary, & Cano, 1997; Foshee, Bauman, Arriaga, Helms, Koch, & Linder, 1999). Dating and sexual violence prevention programs have been found to be effective when administered in health classes (Mytton et al. 2005; Pacifici, Stoolmiller, & Nelson, 2001). In addition, prevention programs that focus on decreasing sexual coercion in dating situations are able to significantly reduce coercive attitudes among adolescent participants (Pacifici, Stoolmiller, & Nelson, 2001). Some suggest that the reason violence prevention programs are well suited to health classes in schools is that these particular classes are mandatory for all students and students are not discriminated against by ability level. The Relationships Without Violence (RWV) program was developed by this author in collaboration with the Rape Recovery Center and the University of Utah, 4 College of Social Work, both located in Salt Lake City, Utah. RWV is a curriculum-based program with the aim of reducing intimate partner violence among adolescents in high schools. The RWV program was also established to explore changes in participants' violent attitudes and behaviors regarding their intimate partners. Finally, the RWV program assessed whether adolescents who participated in the program ultimately changed with respect to endorsing violent attitudes and behaviors over time. Theoretical Perspectives to Intimate Partner Violence Researchers and theorists have developed multiple theories attempting to explain why IPV occurs. Three of these stand out as frequently utilized theories in the current IPV literature: (a) Feminist Perspective, (b) Family Violence Perspective, and (c) Johnson's Control Typology of IPV. The Feminist Perspective asserts that violence is the result of a patriarchal social structure that encourages men's dominance of women through power and control (Bogart, 1988). The Family Violence Perspective purports that violence is a result of society's tolerance of violence as a means of conflict resolution (Straus, 1979). Finally, Johnson's Control Typology maintains that intimate violence is an expression of patterns of interpersonal control and advances four types of IPV defined by the extent to which the perpetrator and their partner use violence in an attempt to control their relationship (Johnson, 2001). Studies Supporting the Feminist Perspective Many studies utilizing a feminist perspective find that men are the main perpetrators of violence. For a more extensive discussion of feminist theory regarding 5 IPV, the reader can review Dobash and Dobash's (1992) seminal work on feminist theory of violence. Additionally, feminist researchers have found that in U.S. police files, more than 90% of family assaults were committed by men against their wives (Dobash & Dobash, 1992; Gaquin, 1978). This phenomenon is partially explained by gender socialization. Men and women are socialized to gender-specific roles by a male-dominated culture wherein males are taught to use power to control women. Therefore, violence is used as a tool to maintain control within a relationship and society (Bogard, 1988; Pagelow, 1984). Research on the Family Violence Perspective Although, many family violence researchers consider themselves feminists, their contributions to the IPV literature conflict with the findings of feminist researchers. Their research indicates that women and men can use violence equally against their intimate partners (e.g., Fiebert, 2004; LaRoche, 2005; Nicholls & Dutton, 2001). For example, the literature has found that men do not always inflict the "first blow" when conflict arises within a relationship. Stets and Straus (1992a, 1992b) found that 43.7% of men reported striking first compared to 52.7% of women (n = 428). Stets and Straus concluded that women engage in comparable amounts of violence and are "at least as likely to instigate violence." Additionally, family violence researchers found that IPV perpetrators were 12.1% of men and 11.6% of women in 1975, and 11.3% of men and 12.1% of women in 1985 (Straus & Gelles, 1990). For these reasons, family violence researchers suggest violence is not simply a result of control (men dominating women) but rather due to lack of conflict resolution skills. 6 Literature Supporting Johnson's Control Typology of IPV Michael Johnson (1995) provided groundbreaking research on different IPV typologies that appears to bridge feminist and family violence theories. Johnson (2001) presents four types of IPV based on a control typology using two factors. These two factors are the violent tendency of the individual and his/her partner, and the motivation of the individual and partner to control the other person. The four types of IPV described by Johnson's Control Typology are intimate terrorism, violent resistance, mutual violent control, and situational couple violence. An intimate terrorist (IT) uses violence in the service of general control over his or her partner while the partner does not. For example, IT is the type of IPV where men use violence as a tool of power to control women. This type of IPV is motivated by power and control, deriving from males' socialization to perceive control as a vital factor of masculinity, and from the patriarchal emphasis of male control of the family. The second type of IPV is violent resistance (VR). An individual is considered to be VR when his or her partner is violent and controlling (an intimate terrorist) and the resister's violence arises in reaction to the attempt to exert general control. VR can be described simply as violence utilized in response to intimate terrorism. The third type of IPV is mutual violent control (MVC), wherein both members of the couple use violence attempting to gain general control over one's partner. Both individuals in the relationship are motivated to control each other through power and violence. Therefore, three of the four types of IPV are organized around attempts to exert or thwart general control. Due to the nature of the violence, these first three types of control appear to support a feminist perspective on IPV. A perpetrator's use of violence is attributed to his or her attempts to 7 control a partner, and violence used by a victim is due to attempts to defend against the controlling behavior. The final type of IPV is situational couple violence (SCV). SVC defines a situation where both partners may use violence. However, neither partner uses violence in an attempt to exert general control. SCV corresponds well with the Family Violence Perspective (Straus, 1979) and posits that escalation of conflict(s) in stressful situations of family life may lead to partner violence. This may be due to a lack of conflict resolution skills. Compared to Feminist and Family Violence perspectives, there is little research on Johnson's Control Typology. This paucity of research might be due, in part, to the introduction of the control typology in just the last 15 years. Johnson's Typology contributes to the IPV literature through findings indicating that only 11% of the general sample was classified as IT, while the court and shelter samples consisted of 68% and 79% IT, respectively (Johnson, 2001). Johnson also found that 97% of IT was male-perpetrated, compared to only 56% of the SCV. Testing Johnson's types of IPV, Graham-Kevan and Archer (2003) found 33% of the general sample violence was IT, whereas the shelter sample revealed 88% IT. Additionally, Graham-Kevan and Archer found that 87% of IT was male-perpetrated, compared with to only 45% of SCV. There are additional studies supporting Johnson's typology (e.g., meta-analyses by Archer, 2000 and Sugarman & Frankel, 1996). 8 The Present Study Relatively few studies have investigated Johnson's Control Typology among adolescents when IPV is present in their dating relationship. As such, the present study will investigate whether the RWV prevention program has an effect on teen IPV attitudes and behaviors in relationship to four types of couple violence: situational couple violence, mutual violent control, intimate terrorism, and violent resistance. The use of violent control by IPV perpetrators may directly or indirectly contribute to further violence in an intimate relationship. The purpose of this dissertation is threefold. The three research questions in this dissertation are the following: (1) This study evaluates a four session RWV program in order to answer the following research questions: (1) Are the violent attitudes and behaviors of RWV participants reduced as measured by the Controlling Behavior Scale (modified), Justification for Dating Violence Scale, and Dating Violence Scale?; (2) Which of Johnson's four typologies are manifested among an adolescent population?; and (3) Are there differences in RWV program participants, particularly between males and females, in regard to violent attitudes and controlling behaviors having a direct predictive effect on IPV perpetration (psychological/emotional, physical, and sexual) and IPV victimization (psychological/emotional, physical, and sexual)? CHAPTER 2 LITERATURE REVIEW Rates of Intimate Partner Violence According to the World Report on Violence and Health (World Health Organization, 2002), both males and females are victims of sexual violence, yet the majority of victims are women and girls who are perpetrated on primarily by men and boys. Additionally, the World Report revealed that 1 in 4 women report sexual violence and 1 in 3 girls report having forced sexual advances from an intimate partner. In the US, the National Violence Against Women (NVAW) Survey found that women were more likely than men to experience intimate partner violence (Tjaden & Thoennes, 2000). Furthermore, the NVAW also found that 25% of women have been victims of physical violence (Tjaden & Thoennes, 2000), while another study found that 27% of women report sexual victimization in their lifetime (Koss, Gidcycz, & Wisniewski, 1987). The fact that a substantial number of IPV victims are females supports the need for programs aimed at preventing sexual violence. In addition, the reduction of female victims is, as stated above, likely to result in both lower rates of victims and future perpetrators. The National Crime Victimization Survey, a national study investigating partner violence among adults indicated that only 30% of female victims were perpetrated by 10 strangers. This leaves an astounding 70% of women that were victims of sexual violence from an intimate partner, a relative, or a friend (Bureau of Justice Statistics, 2005). Consequently, programs focused on reducing intimate partner violence among adolescents are needed in order to better prevent IPV in adulthood. Intimate Partner Violence and Controlling Behaviors Research indicates that controlling behaviors, which may be reinforced through controlling social structures, are predictors of IPV perpetration. Controlling structures in society imposing rigid sex-role expectations can lead to emotional abuse and set the stage for violence to occur in relationships (Tolman, 1989; Tolman & Bennett, 1990). Partners that use controlling tactics such as prohibiting their partner from going to work or school are eight times more common in abusive relationships than non-abusive relationships (Allard, Albelda, Colten, & Cosenza, 1997). Additionally, a study conducted in Massachusetts found that among men arrested for IPV, 38.1% prevented their partners from freely coming and going in their daily routine, 58.5% restricted their partners access to money, and almost one half reported restricting their partners in three or more additional ways (Buzawa, Hotaling, Klein, & Byrne, 1999). A study comparing 224 violent homicides, where women who were killed were compared with similar cases where no death occurred, found that when an abuser was highly controlling, it increased a victim's risk of fatality nine times (Campbell, Sharps, & Glass, 2000). These findings correlating IPV and controlling behaviors were from adult samples. The literature regarding adolescent samples is limited if not nonexistent. Therefore, these findings suggest the need for prevention programs that address controlling behaviors. 11 Intimate Partner Violence among Adolescents Social scientists are becoming more aware of adolescent IPV as a public health concern in the US. Dating violence prevention programs have been designed to address these concerns and prevent IPV among adolescents dating. Researchers claim that adolescence offers an important window of opportunity for breaking the cycle of perpetuation of violence and abuse. Additionally, research has revealed that adolescence is the optimum time to expose students to more adaptive, nonviolent skills and coping strategies (Werkerle & Wolfe, 1999). Prevention efforts that target adolescents are important as this is the developmental period when a teen begins to separate from parental influence and begins to date and establish significant relationships outside of the family structure (Erickson, 1968). Dating is a common experience in adolescence, not isolated to adulthood (Cohall, Cohall, Bannister, & Northridge, 1999). A national survey, using a random-sample design of 650 teens, between 13 and 18 years old, reported that 89% had some romantic involvement with a member of the opposite sex (Kaiser Family Foundation, 1998). A dating violence prevention curriculum delivered to adolescents entering romantic relationships would likely help cultivate nonviolent conflict resolution strategies. Over the past decade, there has been a focus by researchers on IPV and aggression among adolescent students. Studies show dating violence frequently varies among high school students from 9% to 57% (Avery-Leaf, Cascardi, O'Leary, & Cano, 1997; Foshee et al., 1996; O'Keefe, 1998). Prevalence of perpetration and victimization 12 has been found among both males and females ( Malik, Sorenson, & Aneshensel, 1997). Additionally, sexual violence prevalence rates are between 8% and 26%, when considering behaviors such as, forced kissing, forced intercourse, and unwanted touching (Foshee, 1996; Foshee et al., 1996). However, prevalence rates do vary from different studies, which may be due to the widely divergent definitions of violence utilized across studies. These differences include different timeframes for reporting the experience of violence (e.g., lifetime prevalence, current relationship, the previous year, etc.), differences in geographic locations (e.g., rural versus urban populations), and demographic differences in study samples. Literature examining the prevalence of IPV in adolescent populations has conflicting findings as to which gender is more likely to perpetrate violence. Scholars measuring IPV reporting mutual violence consistently report that girls are just as likely as boys to perpetrate and be victims of IPV (Avery-Leaf et al., 1997; Foshee, 1996; O'Keefe, 1997; Schwartz, O'Leary, & Kendziora, 1996). As stated previously, researchers have found that girls are more likely to be victims of IPV than boys. Scholars have tried to explain the difference in these findings. Possible explanations for these findings are that girls are using violence out of self-defense. However, after controlling for violence perpetrated in self-defense, finding reveal that females still perpetrated more violence than males (Foshee, 1996). A more recent explanation for females using violence at higher rates is that there are different types or typologies of partner violence that may make sense of these contradictory research findings. Further explanation of this is detailed in the theory section under Johnson's control typology. 13 The IPV literature is riddled with controversy as to which gender is more likely to perpetrate violent behaviors toward their intimate partners, with studies reporting that women are more likely than men to perpetrate IPV (Stets, & Straus 1990; Straus, 1978). Other studies found that women are as likely as men to be perpetrators of IPV (Staus & Gelles, 1990; White & Koss, 1991). Still other studies on the subject report that women are less likely to use violence than men (Dobash & Dobash, 1978). There are conflicting findings on IPV perpetration and victimization. An area that appears to be absent in the literature on IPV among adolescents are different types of controlling behaviors leading to violence. The IPV adult literature is widely researched on IPV and controlling behaviors; however, the adolescent IPV literature regarding controlling behaviors is sparse. Teen Intimate Partner Violence Prevention Programs Primary prevention programs addressing dating violence have targeted elementary, middle school, and high school students. However, the majority of prevention programs target high school students through mandatory health classes. An important issue arising in literature on prevention programs is how many sessions are necessary to effect change in behavior or attitude. Moreover, literature on dating violence programs has revealed a negative effect of a 1-day session (program presented in one day). Violence prevention programs consisting of a 1-day program reveal that some male students had a backlash effect in which there were significant attitude changes that increased their acceptance of violence (Jaffe, Sudermann, Reitzel, & Killip, 1992). Therefore, these findings seem to suggest that 1-day or 1-session prevention programs 14 might actually increase men's violence against women. Alternatively, multisession dating violence prevention programs can lead to positive change in participants. These positive changes are revealed in a decrease of violent attitudes and an increase in prosocial skills (Avery-Leaf, Cascardi, O'Leary, & Cano, 1997; Cornelius, & Resseguie, 2007). Research findings differ on optimal approaches to delivering prevention programming to students in health classes. For example, some researchers discuss a "train the trainer" approach to delivering the multisession program. This type of delivery would consist of health teachers being trained by violence prevention experts, after which the teacher would deliver the violence prevention curriculum to their students (Avery- Leaf, Cascardi, O'Leary, & Cano, 1997). While this approach has gained attention, some scholars have argued that the best type of delivery for violence prevention programs is for experts, specializing in violence prevention, to facilitate the curriculum in health classes. There are others who found that the best programs are offered by professionals who have ongoing contacts and relationships with students, like teachers. The justification for experts presenting their curriculum is that they have passion and knowledge in the area that may translate to more effective change in their students (Weisz & Black, 2009). Literature on dating violence prevention programs suggests that programs are effective in changing participant's attitudes regarding violence (Foshee, Bauman, Arriaga, Helms, Koch, & Linder, 1998; Jaffe, Sudermann, Reitzel, & Killip, 1992 Macgowan, 1997). While this finding is consistent across studies, researcher also point out that attitude change does not always lead to behavior change (Foshee, Bauman, & Greene, 2000); therefore, assumptions should not be made that changes in attitudes indicate changes in behavior. Furthermore, programs addressing violence among youth 15 have struggled to show behavior changes among the teen participants. Macgowan (1997) found that students participating in violence prevention programs did not change their behavior responses to deal with existing relationship violence or take measures to avoid violent behavior. Based on the above findings, dating violence prevention programs that address violent attitudes, violent behaviors, and healthy relationship skill-based approaches that resolve conflict were more successful for participants when programs were in a multisession format. Research has found that multiple session workshops help students decrease attitudes that justify the use of dating violence as a means to resolve conflict (Avery- Leaf, Cascardi, O'Leary, & Cano, 1997; Foshee, Bauman, Arriaga, Helms, Koch, & Linder, 1998). In addition, researchers who have focused on decreasing sexual coercion in dating situations have been able to significantly reduce coercive attitudes (Pacifici, Stoolmiller, & Nelson, 2001). Health classes are traditionally a mandated part of a students' educational process and are not typically segregated by student ability levels (Avery-Leaf, 1997; Pacifici, 2001). As noted earlier, dating and sexual violence programs are effective when administered in health classes (Avery-Leaf, 1997). These students would likely come from various cultural and ethnic backgrounds and create a mixed-gender setting. The literature on prevention programs discusses a variety of content areas regarding intervention and prevention programs in schools. These content areas address the importance for engaging high school students in programs that address healthy sexuality and dating relationships (Center for Disease Control and Prevention, 2007). Discussion and educational techniques help students develop skills leading to successful 16 conflict resolution (O'Keefe, 1997). Additionally, researchers have provided a set of guidelines that seem to be helpful for rape prevention education in high schools (Pacifici, 2001). Lastly, it is recommended that programs emphasize the legal ramifications for sexual violence in prevention efforts (Center for Disease Control and Prevention, 2007). These findings provide important educational guidance regarding conflict resolution skills, guidelines for rape prevention, and the need to address legal ramifications for violence prevention curriculum. Reviews of Adolescent Violence Prevention Programs Over the past decade, multiple reviews of IPV prevention programs in primary and secondary schools have been conducted. These reviews consist of systematic reviews, meta-analysis, literature reviews, and teen violence prevention programming books. Prevention program reviews (Table 1) provide a synthesis of what is effective and not effective in various prevention programs. Implementing findings from these reviews can help provide best-practice guidelines in an effort to decrease or end intimate partner violence among adolescents. A systematic review conducted by Mytton, DiGuiseppi, Gough, Taylor, and Logan (2006) examined prevention programs searching databases from 1998 to 1999, updated in 2001, and added three new databases in 2003 and found 56 studies to include in their review. The systematic review revealed the effects of school-based violence prevention programs for children found effective programming areas. The first area is that interventions designed to improve relationship or social skills were beneficial. The second area is that interventions targeting students attending primary (elementary) and 17 Table 1 Prevention Program Reviews Authors _ Program Quality____ Curriculum___________ Outcome Data Mytton et al., 2006 Failed to mention Improve relationship or social skills were beneficial Interventions in elementary through high school schools both reduced aggression in participants and were effective for both mixed gender and boys only Limbos et al., 2007 Failed to mention Failed to mention Prevention programs effective at reducing violent behavior Whitaker et al., 2005 Two programs revealed behavior change, each using randomized designs. Most programs reviewed used a curriculum based on feminist and social cognitive theory with considerable differences in duration of programs and the rigor of intervention approaches 9 of the 11 programs reported positive effects for either knowledge or attitudes but not behavior Ting, 2009 Most frequent design used was the pretest-posttest approach (quasi-experimental). Most studies were conducted in high schools Multiple intervention approach is assumed by program designers to be most effective Program participants improved their knowledge and attitudes toward dating violence Cornelius & Resseguie, 2006 Most prevention programs are conducted in a school setting among middle school, high school, or college age students Programs reported using skill-building components but failed to include specific skill-building exercises Lack of outcome research evaluating effects of behavioral and attitude change. Lack of follow-up data on participants' Weisz & Black, 2009 Most programs use a quasi-experimental pre-posttest design which evaluates attitudes and not always behaviors Two dominant theories used were feminist and social learning theory. Multisession programs were effective if they looked at attitude change and skill development Failed to mention ________________________________________________________________________ 18 secondary (middle and high school) schools both reduced aggression. Programs designed for both primary and secondary schools were effective. This finding dispels the myth that we need to intervene at the earliest age possible for the prevention program to be effective. The final area is that programs were effective for both mixed gender and boys only (Mytton et al., 2006). Although these findings suggest intervention effectiveness in reducing aggression at both elementary through high school age students, the findings fall short in identifying differences in curriculum from the programs targeting elementary through high school. Limbos et al. (2007) conducted a systematic review to identify intervention effectiveness of preventing youth violence and collected data from May 2003 to October 2003 and updated in 2006 and they found 41 studies to include in their review. The systematic review revealed that among the 41 studies, 33% were primary interventions (general population), 43% of secondary interventions (identified at increased risk for violence), and 100% tertiary interventions (engaged in violent behavior) were effective in reducing violent behavior in youth (Limbos et al., 2007). These findings suggest an increase in general effectiveness for primary and tertiary interventions with the greatest effectiveness being with primary preventions. A limitation to this review, however, is that it does not provide information regarding the difference between secondary and tertiary prevention programs' curriculum. Whitaker et al. (2005) conducted a systematic review exploring interventions that address primary prevention of partner violence and reviewed 11 programs that met the researcher's inclusion and exclusion criteria between 1990 and March 2003. The review revealed that overall, the quality of reviewed studies was low, 9 of the 11 programs 19 reported positive effects for either knowledge or attitudes but not behavior, and most used a curriculum based on feminist and social cognitive theory with considerable differences in duration of programs and the rigor of intervention approaches (e.g., training, ensuring fidelity, etc.). However, two programs revealed behavior change, each using randomized designs, which may suggest more rigorous method designs are needed to help effective behavior change (Whitaker et al., 2005). This review provided recommendations regarding content and evaluation, which are expanded theory and program development, need for culturally specific programs, target interventions, new settings for interventions, and evaluation design. Ting (2009) executed a meta-analysis exploring effectiveness of dating violence prevention programs in regards to middle and high school students' knowledge and attitudes from 1990 to 2007. This meta-analysis of 13 studies revealed that program participants improved their knowledge and attitudes toward dating violence. Additionally, the meta-analysis revealed that most of the studies were conducted in high schools. The most frequent design was the pretest-posttest approach. A multiple intervention approach was assumed by program designers to be most effective (Ting, 2009). Although this meta-analysis provided some essential information as to overall improvement of attitudes regarding dating violence, the review failed to mention what type of skills were taught in the curriculum to change these violent attitudes. Cornelius and Resseguie (2006) conducted a literature review of primary and secondary prevention programs for dating violence among adolescents. The authors reviewed nine prevention programs in detail for methodology and theoretical concepts. The review revealed the difficulty of comparing different prevention programs due to the 20 vast difference in outcome variables, instrumentation, participant characteristics, and length of treatment. However, the review did capture some themes among the programs reviewed. The first theme was the lack of outcome research evaluating the effects of behavioral and attitude change. The second theme was that most prevention programs are conducted in a school setting among middle school, high school, or college age students. Another theme was that the majority of programs collected data from one partner and then applied the results to the relationship. Additionally, this review revealed that some programs reported using skill-building components but failed to include specific skill-building exercises or did not describe those aspects of the programs. Finally, the literature review revealed a lack of follow-up data on participants' attitudes, knowledge, application of new skills, and help-seeking behaviors (Cornelius & Resseguie, 2006). Weisz and Black (2009) conducted a study compiling their findings in a book titled "Programs to Reduce Teen Dating Violence & Sexual Assault," which reviewed dating violence and sexual assault programs. More specifically, the authors interviewed experienced professionals who either designed or administered dating or sexual violence prevention programs. Among the interviewees were 58 females and 3 males who had a mean age of 37. The authors also reviewed published literature regarding dating and sexual violence to represent both published and professional practice knowledge. The purpose of the book was to discover as much as possible about prevention programming, including what works and what has failed to work in reducing violence. The book revealed that most programs did not mention a theory; however, the two dominant theories used were feminist and social learning theory. Additionally, Weisz and Black discussed the structure of prevention programs being effective if they were multiple 21 sessions looking at attitude change and skill development. They mentioned that most programs fail to offer homework for the students because the students do not often complete the homework. The authors discussed program curriculum and that most program facilitators develop their own curricula and borrow parts from other programs. Findings from their book also mention limited involvement in prevention programs, which is difficult including parents, family, and the community. Regarding program evaluation, they found that most programs use a quasi-experimental pre-posttest design, which evaluates attitudes and not always behaviors. Finally, Weisz and Black discuss that study participants agree educators should have both passion and knowledge in addressing prevention programs with educators being trained and having strong communication skills (Weisz & Black, 2009). These findings provide great insight not only into what the literature has found to be effective among adolescent dating violence prevention programs but sheds a light on prevention practitioners' experiences, knowledge, and perceptions. Although, these different reviews of IPV prevention programs among youth provide encouraging information regarding overall program effectiveness, they do reveal the lack of attention focused on controlling behaviors among adolescents in intimate violent relationships. The research related to whether these adolescent dating violence prevention programs promote change in violent attitudes and behaviors is less straightforward. Only a few prevention programs look into measuring behavior and less reveal behavior change possibly due to not long enough follow-up time when collecting posttest data. While attitude change appears to be more widely measured, research does not provide a direct connection between attitude change and behavior change. 22 Another conclusion from these reviews is that violence prevention programs can be effective for both males and females. Taken together, these studies indicate that programs appear to work for mixed gender and boys only (Mytton et al., 2006), which helps clarify some confusion regarding the need to separate boys and girls into separate rooms for prevention efforts to be effective. These findings are promising for prevention programs conducted in health classes insofar as these particular classes do not typically separated students according to gender, which makes it more practical to conduct interventions targeting boys and girls. Additionally, boys and girls might likely benefit from hearing one another's experiences and perspectives regarding gender socialization. It should be noted that more research is needed to investigate this question and explore a possible interaction effect. Finally, these reviews reveal a lack of research design, rigor, and theory-based interventions. The overall design quality of the programs was also low because most programs either failed to mention how theory was incorporated into their violence prevention curriculum or simply lack a theoretical basis for their program. Moreover, only a few reviews discussed theoretical underpinnings and when mentioned, theories were limited to feminist and social learning theory (Weisz & Black, 2009; Whitaker et. al., 2005). This becomes problematic because the absence of theory can make research or prevention programs less rigorous. Theory attempts to answer a systematic set of interrelated statements intended to explain some aspect of social life and enrich our sense of how people behave (Rubin, 2008). 23 Specialized Features of the Relationships Without Violence Program The Relationships Without Violence (RWV) program has several features that make it different from previously implemented teen violence prevention programs. These features are valuable (Fawson, 2012; Fawson & Prospero, 2009) because the program has potential to be implemented in multiple high schools and statewide institutions. Furthermore, school systems and districts may find the value in implementing the RWV curriculum. Unique features outside of the context of the RWV curriculum are (a) delivery of multiple session/theory-driven intervention and (b) community-university collaboration. The RWV curriculum has multiple components targeting different populations (boys/ girls and different ages from sophomore to seniors in high school). Additionally, the RWV has a strong theoretical base driving the curriculum. RWV draws on feminist theory (e.g., awareness of male dominated society and controlling behaviors) and family violence (e.g., conflict resolution skills). The combination of these theories addresses best-practice guidelines regarding curriculum content and goes a step further to provide skills for participants to effectively deal with conflict in relationships (Mytton et al., 2005). Additionally, the combination of these theories provides a bridge between addressing consequences of controlling behaviors and skills to address these controlling tactics when they appear in relationships. The second specialized feature is the collaboration between the University of Utah and a community agency in Salt Lake City, Utah that provides services for victims of sexual violence. This collaborative relationship guided every aspect of the RWV 24 curriculum, from the development of research design, theory-based curriculum, and program evaluation instruments. The collaborative relationship with the community agency followed feminist thought with curriculum designers continually going to female stakeholders (agency) to help guide the curriculum and research process. The University of Utah, College of Social Work provided the research expertise to ensure best-practice guidelines were infused into the research design. Future research may be necessary to examine this collaborative relationship in more detail through qualitative interviews and focus groups when discussing potential replicability. Theory of Intimate Partner Violence There are several theories that emphasize multiple factors in the etiology of IPV. Among these theories is psychological theory, which posits that psychopathology contributes to IPV perpetration such as borderline personality disorder (Dutton, 1992; Holtzworth-Munroe & Stuart, 1994). Another theory is social learning theory, which posits that conforming behaviors and deviant behaviors are learned through the behavioral processes involved in operant, classical conditioning and social cognition (Akers, 1998). Although both psychological theory and social learning theory offer much to the discussion on IPV, this dissertation will focus on three alternative theories of IPV that have special relevance to addressing the research questions, including family violence perspective (Straus, 1979), the feminist perspective (Bogart, 1988), and the violent control typology (Johnson, 1995). Taken together, these particular theories attempt to explain micro, mezzo, and macro effects of IPV and have been developed 25 primarily from adult populations. In review, this dissertation will attempt to address the lack of research that applies theoretical constructs to adolescent IPV. Family Violence Perspective of Intimate Partner Violence The family violence perspective, most frequently associated with the work of Murray Straus and colleagues, posits that conflict is a natural occurrence in human interactions and is actually needed for progression to occur but does not have to lead to violence (Straus, 1979). This perspective discusses how victimization and perpetration is explained as a result of symmetrical power relations. Family violence researchers advocate for a gender-inclusive approach to violence. Family violence theorists have discovered that male-dominated relationships have the highest rate of abuse (Straus, Gelles, & Steinmetz, 1980). In support of this position, family violence perspective researchers point to the findings that reveal equal perpetration and victimization between husbands and wives, between siblings, and parents toward children, and children toward elders (Straus, 1979, 1980b, 1983). Specific to couples' violence, family violence perspective researchers have found that women were just as likely as men to use violence against their partners during conflict (Stets & Straus, 1990). Additionally, the common perception for the tendency for men to escalate their use of violence against their female partner was not found (Feld & Straus, 1990). Another study investigating partner violence among married couples found that 27% of violent males did not receive or report violent responses from their female partner while 24% of violent females did not report violent reactions from their male partners 26 (Straus, Gelles, & Steinmetz, 1980). Contrary to the belief that males are generally the perpetrators and females are the victims, family violence perspective researchers believe that the most common violence in relationships is where both males and females use violence against each other. A 1975 national study (National Family Violence Surveys) that investigated violence among American families found that 49% of the couples were both violent towards each other (Straus, Gelles, & Steinmetz, 1980). The study (National Family Violence Resurveys) was repeated 10 years later and both studies found gender symmetric partner violence: 12.1% of men and 11.6% of women were IPV perpetrators in 1975; 11.3% of men and 12.1% of women IPV were perpetrators in 1985 (Straus & Gelles, 1990). The findings from the family violence perspective researchers that revealed the prevalence of male-to-female violence, female-to-male violence, parent-child violence, and child-to-child violence has led them to believe that violence is an all-pervasive feature of family life and therefore, the primary solution to intimate partner violence is beyond managing violent males only (Straus, Gelles, & Steinmetz, 1980). Rather, family violence perspective researchers use the family as the central unit of analysis rather than the relationship between women and men. It is theorized that violence is weaved into the family system and social structure (Gelles, 1985; Gelles & Straus, 1988). Strategies for reducing stress proposed by researchers of family violence perspective include reducing the stressors in the family, changing the acceptance of violence in society and the family, and changing the patriarchal family and social structures. Specifically, family violence perspective researchers recommend the following: reducing the stressors of society, such as unemployment and poverty; 27 eliminating the acceptance of violence in society and the family through the reduction of violence in the media, the abolition of the death penalty, gun control, and bringing an end to corporal punishment; and addressing sexism both in the family and in society as a whole (Gelles & Straus, 1988; Straus, Gelles, & Steinmetz, 1980). Feminist Perspective on Intimate Partner Violence Feminist theorists criticize family violence theory as ignoring the context of the patriarchal society in which the violence occurs (Lenton, 1995). The feminist IPV theory does not replace other IPV theories but adds to the body of literature. The feminist perspective argues that IPV is a result of the patriarchal social structure that socializes males and females into gender-specific roles (Bogart, 1988; Pagelow, 1984). Feminist theory differs from the family violence perspective assuming that broader societal factors may influence the balance of power within the family. The feminist perspective assumes that male domination is the underlying cause of violence against women (Dobash & Dobash, 1979). Research supports the feminist perspective by establishing relationships between wife abuse and ideology of patriarchy or structured gender inequality (Dobash & Dobash, 1988). Feminists challenge the social structures that reinforce the patriarchal worldview that male domination over women is not only normal but actually expected and therefore perpetuates male violence in society. Therefore, this perspective assumes that men are perpetrators and females are victims. Feminist scholars argue that the motive to exercise control over their partner stems from men's need to maintain their dominant status in the relationship and society (Dobash & Dobash, 1979, 1992). With that said, multiple control tactics are used to 28 control their partner, with violence being one method. An instrumental tool used by feminists to identify control tactics from a partner is The Duluth Domestic Abuse Intervention Project (Pence & Paymar, 1986, 1993) that describes a variety of control tactics (e.g., isolation and intimidation). They developed the Power and Control Wheel, which illustrates how violent control tactics are a pattern of power and control, which are economic control, threats, use of children against a partner, use of privilege, emotional abuse, sexual control, and isolation. These control tactics were the bases for a ground-breaking theory of IPV, Johnson's (1995) violent control typology, which attempts to explain how the feminist and the family violence researchers were finding such different results regarding which gender causes more violence. Johnson's Control Typology of Intimate Partner Violence Johnson (1995) investigated (feminist and family violence) and found that theorists from these two different perspectives analyzed two different kinds of partner violence data/population. Johnson's IPV typologies addressed the importance of recognizing that there are different types of partner violence that can explain contradictory research findings. In the mid-1970s a study began the debate regarding the nature of violence. It was published in a 1975 National Family Violence Survey (Steinmetz, 1977-78). This study discussed "the battered husband syndrome" which found that men were just as likely to be battered by their female partners as females were by their male partners. This began a 30- year debate over which gender is more violent. These researchers (feminist and family violence) assumed that all partner violence was the same thing and that although 29 violence may occur at different rates and with different severity, IPV was one phenomenon. Family violence researchers studied violence among parents, children, between siblings, and between partners/spouses. They conducted large-scale national surveys and found that family conflict led different members of the family to assault each other in their homes (Gelles, 1980; Staus, 1973). The feminist scholars concentrated their studies on wife abuse. They studied agency data from police, courts, emergency rooms, and shelters, and used qualitative interviews of women who came to these agencies for help. Johnson (1995) found that the main difference between forms of partner violence was the propensity to use violence and the motivation behind the use of violence. Family violence researchers who collected quantitative survey data from random samples were more likely to find that perpetrators use violence to control their situation, while feminist perspective researchers collect qualitative data from areas where the violent perpetrators are trying to control their partners, such as shelters and criminal/divorce courts (Johnson, 1995). Johnson's (1995) results, from family violence and feminist scholars, revealed that two researchers were studying two different kinds of partner violence. Studies that demonstrated the predominance of male violence involved agency data (courts, police agencies, hospitals, and shelters), while studies that showed gender symmetry involve a representative sample of large-scale research. However, both of these research strategies were found to be biased. For example, family violence researchers who used a random sample attained sample bias through participants' nonresponses or refusals. Johnson 30 (1995) revealed the nonresponse rate of 40% where 18% usually report in the National Family Violence Surveys. Johnson argues that most of the respondents would have been classified as (SCV) because nonrespondents were likely to be intimate terrorists, as they would not want to report to the public and their victimized partners would fear retaliation for reporting any IPV within the relationship. The sampling methods used by feminist researchers also reveal sample bias results. For example, women who were more likely to experience severe consequences from a violent partner were more likely found in shelters and hospitals. Therefore, it appears that both camps of researchers (family violence perspective and the feminist perspective) gathered data from different samples using different research methods and thereby obtained results that support their respective perspective regarding the root cause of IPV. These finding suggests there could be two qualitatively different forms of partner violence, one gender-symmetric and overrepresented in general surveys, and the other male-perpetrated and overrepresented in agency samples. Although there are significant differences between intimate terrorism and situational couple violence, Johnson (1995) maintains that the range of frequency, escalation, severity, and mutuality of violence within each of these two types of violence can be large. For example, intimate terrorists can maintain their power and control over their partner without severe violence, needing a simple gaze at their partner to "practice" their control. Yet, an individual who participates in situational couple violence may use severe violence against their partners when confronted by extreme family stressors. The general pattern of power and control is the defining characteristic that differentiates the 31 two types of violence, with intimate terrorism using controlling violence upon the partner and situational couple violence being noncontrolling violence without the intention of general power and control of either partner. Johnson (1995) looked at the sampling argument. He tested whether or not general survey samples tap primarily situational couple violence, while agency samples give access primarily to intimate terrorism and violent resistance. The four types of IPV described by Johnson's control typology are situational couple violence, intimate terrorism, violent resistance, and mutual violent control. Johnson (2001) presents four types of IPV based on a control typology using two factors. These two factors are the violent tendency of the individual and their partner, and the motivation of the individual and partner to control their partner. The four types of IPV described by Johnson's control typology are, intimate terrorism, violent resistance, mutual violent control, and situational couple violence. In intimate terrorism (IT), the perpetrator uses violence in the form of general control over his or her partner while their partner does not. IT is the type of IPV where men use violence as a tool of power to control women. This type of IPV is motivated by power and control, deriving from the socialization of males to perceive control as a vital factor of masculinity and the patriarchal emphasis of male control of the family. The second type of IPV is violent resistance (VR); the partner is violent and controlling (an intimate terrorist) and the resister's violence arises in reaction to that attempt to exert general control. VR can be described simply as violence utilized in response to intimate terrorism. The third type of IPV is mutual violent control (MVC); both members of the couple use violence in attempts to gain general control over their partner. In MVC, both individuals in the 32 couple are motivated to control each other through power and violence. Therefore, three of the four types of IPV are organized around attempts to exert or thwart general control. Additionally, the first three types of control resonate well with the feminist perspective because the nature of the violence is where the perpetrator attempts to control their partner and the victim may use violence to defend against the controlling behavior. The final type of IPV is situational couple violence (SCV); the perpetrator is violent (the partner may be as well); however, neither of them uses violence to attempt to exert general control. SCV corresponds well with family violence perspective (Straus, 1979) and therefore posits that escalation of conflict(s) in the stressful situations of family life may lead up to partner violence due to a lack of conflict resolution skills. Johnson (2001) argues that IPV is not a single phenomenon and that the two groups of researchers are using different types of research methods in different populations and thereby finding different types of IPV. Family violence perspective researchers used quantitative analysis by collecting survey data from random samples whereas feminist perspective researchers used qualitative analysis collecting data from interviews in agency samples, such as shelters, hospital emergency rooms, and criminal and divorce courts. Johnson argues that the opposing family violence and feminist researchers are actually investigating different types of IPV because the two distinct methodologies used to collect data come from two distinct samples that are based on gender differences. Family violence perspective research has largely used survey research on randomized samples that do not necessarily include the population that is likely to be involved in more serious and more frequent violent incidents. Random samples are more likely to find SCV because it is the most common IPV. Additionally, 33 Johnson (1995) found a 40% nonresponse rate in the survey research, which may have eliminated most of the cases of IT. Therefore, this methodology will more likely find SCV. Feminist perspective researchers, on the other hand, generally collect data from areas where more frequent violent and more violent incidents occur and will more likely find IT, such as shelters, hospital emergency rooms, and criminal/divorce courts. Additionally, the qualitative aspect will more likely uncover VR, as individuals who experience IPV will provide contextual information regarding the situation and therefore allow researchers to elicit information regarding the motivation for the violence. Agency samples are more likely to have clients that experience IT because this type of violence is recurring and escalating, which will more likely lead to hospital visits due to injuries, shelters, and/or contact with the court system. Using Frieze's data (Frieze, 1983; Frieze & Browne, 1989; Frieze & McHugh, 1992), Johnson (2001) found that only 11% of the general sample was classified as IT, while the court and shelter samples consisted of 68% and 79% IT, respectively. Johnson also found that 97% of IT was male-perpetrated, as compared to only 56% of the SCV. Testing Johnson's types of IPV, Graham-Kevan and Archer (2003) found 33% of the general sample violence was IT, whereas the shelter sample revealed 88% IT. Additionally, Graham-Kevan and Archer found that 87% of IT was male-perpetrated, compared with only 45% of SCV. A meta-analysis from Archer (2000) found gender-symmetry in IPV among general samples (situational couple violence) but found males to primarily be the perpetrators in agency samples (patriarchal terrorism). Another meta-analysis found a strong relationship between gender attitudes and male-perpetrated IPV 34 in agency samples but no relationship in the general samples (Sugarman & Frankel, 1996). Purpose of This Study The purpose of the present dissertation is three-fold. The first purpose is to evaluate a four-session RWV program to see if violent attitudes and behaviors were reduced from pretest to posttest. The second purpose of this study is to determine if three of Johnson's four typologies are manifested among an adolescent population, which are IT, SVC, and VR. The final purpose of this dissertation is to see if there are differences between males and females regarding violent perpetration and victimization. The present study investigated the effectiveness of the RWV program at reducing violent attitudes and behaviors among adolescent participants. Further, this study specifically looked at the presence of IPV perpetration among adolescents classified as situational couple violence and mutual violent control by Johnson (2001). Johnson (2008) states that future research regarding education prevention efforts needs to be conducted addressing specific violent typologies to prevent IPV. Prevention of IPV has long been a major part of the shelter system that grew out of the violence against women movement. One focus of female victims of IPV services is to provide community education in schools to address social change. Most of this focus is on intimate terrorism. Prevention with regard to situational couple violence takes place largely in marriage preparation classes. Addressing specific violent typologies among adolescents participating in violence prevention programs may lead to more effective change due to targeting specific controlling behaviors. 35 The present study investigated significant differences in the IPV perpetration (psychological, sexual, and physical), and IPV victimization (psychological, sexual, and physical) and controlling behaviors (intimidation & threats). The use of coercive controlling behaviors may contribute to IPV victimization and IPV perpetration. Finally, because male socialization and controlling behaviors may be greater among more traditional cultures (e.g., Asian and Hispanic populations), the present study analyzed the effects of race/ethnicity on IPV perpetration and IPV victimization. As stated earlier in this dissertation proposal, the present study's three research questions are the following: (1) evaluate a four-session RWV program to see if violent attitudes and behaviors were reduced as measured by the controlling behavior scale (modified), justification for dating violence scale, and dating violence scale; (2) determine if three of Johnson's four typologies are manifested among an adolescent population, which are an intimate terrorist, situational couple violence, and violent resister; and (3) find out if there are differences in all participants in the RWV program and differences between male and female participants in regards to where violent attitudes and controlling behaviors have a direct predictive effect on IPV perpetration (psychological/emotional, physical, and sexual) and IPV victimization (psychological/emotional, physical, and sexual)? CHAPTER 3 METHODS Research has revealed that dating violence prevention programs with multiple sessions help students decrease attitudes that justify the use of dating violence as a means to resolve conflict (Avery-Leaf, Cascardi, O'Leary, & Cano, 1997; Foshee, Bauman, Arriaga, Helms, Koch, & Linder, 1998). Dating and sexual violence programs have been found effective when administered in health classes. Health classes are traditionally a mandated part of a students' educational process (Avery-Leaf, 1997; Pacifici, 2001). These students would likely come from various cultural and ethnic backgrounds and create a mixed-gender setting. The sample of the RWV Prevention Program consists of ninth- through twelfth-grade students in one Salt Lake City high school health class. The RWV was conducted Fall of 2008 to Fall of 2010 academic year; all sessions were conducted by one facilitator, with multiple secondary facilitators participating in different sessions throughout the project. This section will describe the methods used to evaluate The RWV Prevention Program with regard to a) population, design, and procedure, b) measures, c) statistical analysis methods, and d) testing theory through computation of variables. 37 Population The RWV Prevention Program is a pilot program that is aimed at teenage boys and girls in a high school health class to help reduce or eliminate dating violence among adolescents. Research has found that both boys and girls report perpetration and victimization of dating aggression in their current or past relationships. Studies revealed that among high school students, 9-57% of dating couples have experienced some form of violence (physical, emotional, and sexual) in their relationships (Avery-Leaf, Cascardi, O'Leary, & Cano, 1997; Foshee et al., 1996). Literature reveals a backlash effect from one-session prevention programs where the boys actually became more violent (Jaffe, Sudermann, Reitzel, & Kellip, 1992). Multiple sessions for dating violence prevention programs have been found to be the most effective in reducing violent attitudes (Cornelius & Resseguie, 2007). Dating violence can be addressed through primary and secondary prevention efforts. Primary and secondary prevention can be conducted in multiple-session dating violence prevention programs in high school health classes (Cornelius & Resseguie, 2007). Primary prevention focuses on stopping the violence before it occurs among high school students in dating relationships or future dating relationships. Secondary prevention focuses on addressing violence that is already occurring in a relationship so that perpetrator(s) cease violence. Early prevention and intervention efforts can be implemented to help reduce and prevent dating violence among high school students. The literature reveals that when exploring effectiveness among prevention program length, programs that consist of multiple sessions were found to be more effective than single-session prevention programs. 38 RWV curriculum includes methods to prevent dating violence, both directly (prevent becoming a victim) and indirectly (prevent others from becoming victims). Topics include addressing sexual coercion, violent behaviors, violent attitudes, socialization of violence, and substance abuse while promoting prosocial behaviors, positive peer culture, healthy masculinities/femininities, and consent. RWV was developed to address cultural issues that are relevant to dating violence. RWV was based on evidence-based practices that have been used nationally in violence prevention (Cornelius & Resseguie, 2007). Participants Participants were boys and girls attending their mandated high school health class. The total sample for the Relationships Without Violence (RWV) project consisted of ninth- to twelfth-grade students. The RWV was designed to fit the curriculum in health classes regarding dating violence, consequences of dating violence, healthy ways to deal with conflict, and healthy relationships. The RWV curriculum consists of four different sessions, which covered the topics of sexual coercion, violent behaviors, violent attitudes, socialization of violence, and substance abuse while promoting prosocial behaviors, positive peer culture, healthy masculinities/femininities, consent, and help seeking behaviors. Study Design The RWV used a quasi-experimental design. The program was conducted in one high school in the Salt Lake City area and a control group was conducted at the same 39 high school when no treatment was given. The program was administered to ethnically/racially diverse high school students during their mandatory health classes in four 90-minute sessions. The control group received no treatment. Pretests were administered 1 week before the program began and posttests were administered 2 weeks to 1 month after the program ended. Instruments measured attitudes related to gender-role violence, behaviors related to violence, and controlling behaviors. The students were also asked to provide written responses to open-ended questions regarding the limitations and strengths of the program at the end of the four-session program. The sample size to provide enough power to receive the desired effect after running a power analysis should be 132 students. The school participating in the study (treatment and control groups) has six to eight classes a semester with 30 to 35 students in each class, which provided a sample size of 1071 (treatment group) and 203 (control group) students participating in the study. The sample size will provide enough power to find the desired effect when using an analysis of covariance (ANCOVA), K-means cluster analysis, and structural equation modeling (SEM). The pretest was administered 1 week before the RWV curriculum begins and posttests were collected 1 month or 2 weeks after the RWV ends. The reason for waiting 1 month or 2 weeks after the program (which is the end of the semester) to administer the posttest is to measure for behavior change. Measures The survey consisted of the following instruments: (a) Demographics page (b) the Revised Controlling Behaviors Scale (CBS-R: Graham-Kevan & Archer, 2003); (c) the 40 Justification of Dating Violence Scale (attitudes) (Shen, 2008); (d) Dating Violence Scale (Shen, 2008). Permission for all standardized instruments was obtained from the appropriate publishing company or authors (e.g., Shen, Graham-Kevan & Archer for CBS-R). Demographics Page The demographics page consisted of 10 questions regarding the personal characteristics of the participant. Following are the nine demographic questions with the appropriate coding in parentheses: sex (male=1; female=0); age (participant wrote age in years); race/ethnicity (1=Asian; 2=African American; 3=Hispanic; 4=Native American; 5=White; 6=Pacific Islander; 7=Other); parents' marital status (1=married and living together; 2= married and living separately; 3=divorced; 4=widowed; 5=remarried; 6=other); what grade are you in (1=freshman; 2=sophomore; 3=junior; 4=senior) average grade of last academic year (1=A; 2=B; 3=C; 4=D; 5=F) sexual orientation (1=heterosexual; 2=gay; 3=lesbian; 4=bisexual); currently in an intimate relationship (1=yes; 0=no); relationship status (0=never been in an intimate relationship in the past; 1=I have been or currently am in an intimate relationship in the past that lasted at least 1 month); partner's sex (0=female; 1=male; 2=both). Revised Controlling Behaviors Scale Controlling behaviors were measured using the Revised Controlling Behaviors Scale (CBS-R: Graham-Kevan & Archer, 2003), Modifications were made to the original instrument for use with high school to help with survey flow and format conformity. The 41 original CBS-R has has five subscale behavioral categories that involve five types of control tactic: using economic abuse, using coercion and threats, using intimidation, using emotional abuse, and using isolation. However, a modification was made to the scale including only two subscales (using coercion and threats and using intimidation). The adjustment made to the survey adapted the scale to 10 questions. The CBS-R is a 48- item behavioral scale based on the Domestic Abuse Intervention Project and the Duluth Model (DAIP; Pence & Paymar, 1993, 1986). Using a 4-point scale (1-4), respondents are asked to indicate how often during the past year they had used any of the 24 behaviors listed to influence their partners. Additionally, the respondents are asked if their partners had tried to influence them using any of the 24 behaviors listed (for a total of 48 items). The anchor scores for both the overall controlling behaviors and the subscales ranged between 1 (strongly disagree) to 4 (strongly agree). Similarly, adding the appropriate items for the subscale and dividing by the respective number of items creates the mean score for each subscale. Therefore, all respondents had mean scores for total control, economic control, emotional control, threat control, intimidation control, and isolation control that ranged between 0 and 4 (both perpetration and victimization for each subscale). The CBS-R scale has shown discriminant ability (Graham-Kevan & Archer, 2003) and the internal consistency revealed the following Cronbach's alphas for partner (P) and self (S) reports: economic, P: a = .58; S: a = .45; coercion and threats, P: a = .72; S: a = .70; intimidation, P: a = .74; S: a = .62; emotional abuse, P: a = .81; S: a = .75; isolation, P: a = .88; S: a 42 = .84. The reliability alpha coefficient for the entire instrument in the present study was a =.94. Justification of Dating Violence Scales These scales were developed by Shen (2008) based on previous studies by O'Keefe (1998), Pflieger and Vazsonyi (2006), and Yick and Agbayani-Siewert (2000). Each scale assessed agreement upon the use of violence within dating partners in specific situations (e.g., infidelity, separation, or disobedience) for female-to-male perpetrated IPV and male-to-female perpetrated IPV. Respondents were instructed to rate their agreement on each item for each scale, "It is ok for a girl to hit her boyfriend if" and "It is ok for a boy to hit his girlfriend if," on a 4-point scale (1 = Strongly disagree, 2 = Disagree, 3 = Agree, 4 = Strongly agree), respectively. Each scale consisted of 9 items in which participants' possible scores ranged from 9-36. Higher scores for each scale indicated higher levels of agreement for the use of violence in IPV situational contexts. Alpha coefficients for the present study were .92 and .97 for female to-male IPV and male-to-female IPV, respectively. Dating Violence Scale - Victimization This scale was developed by Shen (2008) to assess experiences of dating violence victimization based on previous studies by O'Keefe (1998), Straus (1979), and Wolfe, Scott, Reitzel-Jaffe, Wekerle, Grasley, and Straatman (2001). The scale consisted of 17 items to assess psychological, physical, and sexual aggression, as well as coercive control and stalking (Shen, 2008). Respondents were instructed to rate this scale if they were in a 43 current intimate relationship on a 7-point scale (0 = Never, 6 = Always) and were prompted by "When we have conflicts…" for self-reported ratings. Higher scores indicated higher levels of IPV victimization experience. The alpha coefficient for the current study was .97 Statistical Analysis Methods Descriptive statistics were used to depict the composition of the study's sample, the percentage of participants classified into the IPV categories (intimate terrorism, situational couple violence, & violent resistance), and the prevalence of IPV perpetration and victimization (psychological, sexual, and physical). Analysis of Covariance (ANCOVA) was conducted to investigate treatment effect of the RWV prevention program to see if violent attitudes and behaviors were reduced as measured by the controlling behavior scale (modified), justification for dating violence scale, and dating violence scale (Research Question 1). K-means cluster analysis was conducted to determine if three of Johnson's four typologies are manifested among an adolescent population, which are an intimate terrorist, situational couple violence, and violent resister, and to test the theory to see if there is a nonviolent controller (Research Question 2). Structural equation modeling (SEM) was conducted to test differences between the entire group of participants (both males and females together), and compare whether differences between male and female participants in the violence prevention program independent variables, violent attitudes and controlling behaviors, have a direct predictive effect of the dependent variable, IPV perpetration (physical, emotional/psychological, 44 and sexual) and IPV victimization (physical, emotional/psychological, and sexual) (Research Question 3). CHAPTER 4 FINDINGS Descriptive Statistics The present study consisted of 837 high school students (613 treatment group and 233 control group). The mean age of the participants was 15.7 years, and 77% reported their education level as being sophomores. Fifty-one percent of the sample was female, and 92.8% reported heterosexual as their sexual preference with 66% of participants being in a current relationship. The racial/ethnic make-up for the youth was 4% African- American, 7% Asian, 36% Hispanic, 35% White, 3% Pacific Islander, and 11% classified themselves as "mixed race." Sixty-four percent of the participants reported their parents' marital status as married living together, 15% divorced, 4% married living separately, and 4% remarried. Eighteen percent of the participants were classified as intimate terrorists, 21% as situational couple violence/violent resisters, 20.3% as nonviolent controllers, and 40% as nonviolent and noncontrolling (see Table 2). The descriptive statistics also revealed that participants at pretest experienced physical intimate partner violence (33.7%), emotional/psychological violence (46.2%), followed by sexual coercion at 16.5%. Posttest descriptive statistics revealed that participants experienced physical intimate partner violence (22.4%), emotional/psychological violence (41.9%), followed by sexual coercion (15.8%). 46 Table 2 Characteristics of Respondents Variable N % Total N 846 100% Treatment 613 72.5% Control 233 27.5% Age of Participants 14 7 1% 15 399 48% 16 298 36% 17 104 12% 18 30 4% (M=15.7; SD=.83) Participant's Gender Male 426 51% Female 411 49% Participant's Ethnicity African-American 35 4.2% Asian 60 7.1% Hispanic 302 35.8% White 291 34.5% Pacific Islander 29 3.4% Mixed race 94 11.2% Native American 4 .5% Other 28 3.3% Parents' Marital Status Married living together 535 64.2% Married living separately 35 4.2% Divorced 127 15.2% Widowed 27 3.2% Remarried 31 3.7% Other 78 9.4% Education Level of Participant Freshman 5 .6% Sophomore 641 76.7% Junior 113 13.5% Senior 77 9.2% Sexual Preference Heterosexual 775 92.8% Gay 6 .7% Lesbian 6 .7% Bisexual 48 5.7% Relationship Status in Past 6 Months No Relationship 278 34% Current Relationship 540 66% 47 Table 2 Continued Characteristics of Respondents Variable N % Types of IPV Intimate terrorism 155 18.3% Nonphysically Violent Controller 172 20.3% Situational couple violence/VR 178 21% Nonviolent & Noncontroller 341 40.3% IPV Perpetration Total 607 86.9% (M=.37; SD=.42) Psyc Viol Pretest 801 46.2% Posttest 702 41.9% (M=.76; SD=.69) Phys Viol Pretest 798 33.7% Posttest 697 22.4% (M=.21; SD=.42) Sex Viol Pretest 799 16.5% Posttest 698 15.8% (M=.16; SD=.39) 48 The mean scores for all of types of IPV perpetration were less than 1. For means, standard deviations, and percentages for total IPV perpetration and psychological, physical, and sexual IPV perpetration, please refer to Table 2. K-means Cluster Analysis The K-means creates means for the two clusters to form two categories, one high and one low on a particular variable. A K-means cluster analysis was conducted on the variable controlling behaviors in order to separate participants who scored high on coercion from those who scored moderate and low on coercion. There were two clusters that were categorized as 1=low and 2=high. Additionally, participants who reported any violent perpetration of an intimate partner were coded as 1 and participants who did not report any violent perpetration of an intimate partner were coded as 0. Two new variables (controlling behaviors and IPV perpetration) were created based on the respondents' scores. Since the controlling behavior variables only had respondents' scores and not their partners, exploration was limited to respondents. This process is different from Johnson's dyadic context of measuring variables in IPV categories (2001), which includes the controlling behaviors and violent tendencies of the individual and his or her intimate partner. However, these data do provide information on individual controlling behavior and IPV of the partners. The two new variables were combined to create the five IPV categories (see Table 3): Intimate terrorists (IT) are categorized as participants who reported violent perpetration of an intimate partner and high coercive control. 49 Table 3 Creating Control Typology Variables and Categories IPV Categories Violent Control SCV/VR Yes No IT Yes Yes NPVCC No Yes NV/NC No No Violent resisters (VR) are categorized as participants who reported violent perpetration or nonviolence and low coercive control. Situational couple violence/violent resisters (SCV/VR) were categorized as participants who reported violent perpetration of an intimate partner and low coercive control. Nonphysically Violent Coercive Controllers (NPVCC) were categorized as participants who reported nonviolence and high coercive control. Nonviolent and Noncontrollers (NVNC) were categorized as participants who reported no violent perpetration of an intimate partner and no coercive control. ANCOVA Analysis of covariance was used to investigate treatment effect in two different groups, treatment and control, of the RWV prevention program with three different types of violence (physical, sexual, and emotional/psychological). The ANCOVA found that 50 none of the three types of violence was significant for the two groups (treatment and control). Therefore, there was no need to conduct further analysis of treatment effect. Structure Equation Modeling Structural equation modeling was conducted to investigate three different models among the RWV participants: 1) full mediation model, which explored whether the independent variables "violent attitudes" and "controlling behaviors" predicted the dependent variable IPV perpetration (physical, emotional/psychological, and sexual violence; 2) partial mediation model, which explored whether the independent variables "violent attitudes" and "controlling behaviors" predicted the dependent variable IPV perpetration (physical, emotional/psychological, and sexual violence); and 3) stacked model, partial mediation, which explored whether there were differences between male and female participants regarding the predictive power of the independent variables "violent attitudes" and "controlling behaviors." The results of the full mediation model (Figure 1) were good: chi-square (32, N = 1133, p<.05, RMSEA = .19. The model fit is outside of the ideal range (RMSEA is .19, where values less than .08 are considered good); however, this is likely due to the fact that the data are Poisson and contain some floor effects, which makes good model fit less likely. This analysis revealed that the direct effect of violent attitudes on controlling behaviors for the full mediation model was high (regression weight estimate = .68, 51 p<.05). Additionally, the direct effect of controlling behaviors on IPV perpetration (regression weight estimate = .49, p<.05) and IPV victimization (regression weight 52 estimate = .60, p<.05) for the full mediation model was high. Violent attitudes significantly predict controlling behaviors, and controlling behaviors significantly predict both IPV victimization and IPV perpetration. Therefore, controlling behaviors act as a mediator variable in the predictive relationship between violent attitudes and IPV perpetration and IPV victimization. However, the full mediation model did not show the direct effect of violent attitudes on IPV perpetration and IPV victimization when controlling for controlling behaviors. Therefore, the partial mediation model was conducted to reveal this effect. The partial mediation model (Figure 2) is a significant improvement on the full mediation model and was accepted as the better model, based on the chi-squared difference test ( 2(2)= 51, p<.001), which adds a significant improvement to the model. All factor loadings were significant (p<.05), indicating that violent attitudes, controlling behaviors, and IPV victimization and IPV perpetration variables were well represented by the indicators. Additionally, the partial mediation model freed two parameters, revealing the direct effect between violent attitudes and IPV perpetration (standardized regression weights = -.38) and IPV victimization (standardized regression weights = -.88). The anlaysis found that the direct effect of violent attitudes on controlling behaviors for the partial mediation model was high (regression weight estimate = .75, p<.05). Additionally, the direct effect of controlling behaviors on IPV perpetration (regression weight estimate = .90, p<.05) and IPV victimization (regression weight estimate = 1.50, p<.05) for the partial mediation model was large. 53 54 The direct effect of controlling behaviors on the three different types of IPV perpetration was also large: (physical violence, standardized regression weight = .86, p<.05); emotional/psychological violence (standardized regression weight = .90, p<.05); and sexual violence (standardized regression weight = .84, p<.05). Additionally, the direct effect of controlling behaviors on the three different types of IPV victimization was large: (physical violence, standardized regression weight = .74, p<.05); emotional/psychological violence (standardized regression weight = .85, p<.05); and sexual violence (standardized regression weight = .63, p<.05). In sum, the results suggest that controlling behaviors partially mediated the relationship between violent attitudes and IPV perpetration and IPV victimization. However, when controlling for controlling behaviors, violent attitudes had a negative effect on IPV perpetration and IPV victimization. The results of the stacked partial mediation model (Figure 3) were good: chi-square (64, N = 1204, p<.05, RMSEA = .14). Again, as with the full and partial mediation model, the model fit is outside of the ideal range (RMSEA is .14, where values less than .08 are considered good). In this model, males and females were allowed to have different relationships between the factors in the model (i.e., the hypothesis being tested is that the partial mediation model works differently for males and females).The factor weights, that is the loadings from the factor to the observed variables, are set to be fixed between males and females. The stacked, partial mediation models for males and females were significant at p<.05. 55 56 Male Model The direct effect of violent attitudes on controlling behaviors for the stacked partial mediation model was high (regression weight estimate = .65, p<.05). Additionally, the direct effect of controlling behaviors on IPV perpetration (regression weight estimate = .61, p<.05) and IPV victimization (regression weight estimate = 1.33, p<.05) was high. Violent attitudes significantly predict controlling behaviors and controlling behaviors significantly predict both IPV victimization and IPV perpetration. Therefore, for male study participants, controlling behaviors act as a mediator variable in the predictive relationship between violent attitudes and IPV perpetration and IPV victimization. Female Model The direct effect of violent attitudes on controlling behaviors for the female stacked, partial mediation model was also high (regression weight estimate = .79, p<.05). Additionally, the direct effect of controlling behaviors on IPV perpetration (regression weight estimate = .1.22, p<.05) and IPV victimization (regression weight estimate = 1.86, p<.05) was high. The female stacked, partial mediation model revealed the direct effect between violent attitudes and IPV perpetration (standardized regression weights = -.68) and IPV victimization (standardized regression weights = -1.30). The analysis revealed a unique contribution to the model: When controlling for controlling behaviors, females' violent attitudes (with no controlling behaviors present) had a negative effect on IPV perpetration and IPV victimization. Violent attitudes significantly predict controlling behaviors and controlling behaviors significantly predict both IPV victimization and IPV 57 perpetration. Therefore, controlling behaviors act as a mediator variable in the predictive relationship between violent attitudes and IPV perpetration and IPV victimization. Thus, the male and female stacked, partial mediation models answered the third research question regarding differences between male and female RWV participants. In the male stacked, partial mediation model, controlling behaviors fully mediated the relationship between violent attitudes and IPV victimization and IPV perpetration. However, in the female stacked, partial mediation model, after controlling for controlling behaviors, violent attitudes had a negative effect on IPV victimization and IPV perpetration. This is probably due to the fact that the indirect effect through controlling behaviors is so strong. CHAPTER 5 DISCUSSION The present dissertation investigated three areas of a youth sample who participated in the Relationships Without Violence (RWV) program: (1) This study evaluates a four-session RWV program in order to answer the following research questions: (1) Are the violent attitudes and behaviors of RWV participants reduced as measured by the Controlling Behavior Scale (modified), Justification for Dating Violence Scale, and Dating Violence Scale?; (2) Which of Johnson's four typologies are manifested among an adolescent population?; and (3) Are there differences in RWV program participants, particularly between males and females, in regard to violent attitudes and controlling behaviors having a direct predictive effect on IPV perpetration (psychological/emotional, physical, and sexual) and IPV victimization (psychological/emotional, physical, and sexual)? Summary of Findings The present dissertation found that there was no treatment effect from control group to treatment group when testing the four-session RWV prevention program (Research Question 1). These findings are reflected in the dating violence prevention program literature when addressing primary prevention (Avery-Leaf,Cascardi, O'Leary, 58 & Cano, 1997). One reason there may not have been a significant treatment effect when comparing control and treatment groups is because the RWV program targets all students (primary prevention) and does not target only those students at different risk levels. Another reason for the lack of a treatment effect is the need for longer follow-up to provide enough time to show behavior change (Cornelius & Resseguie, 2006). However, there was a reduction in IPV violent attitudes and IPV violent behaviors from pretest to posttest, but there was no significant change when the treatment group was compared to the control group. In regard to testing Johnson's typology among an adolescent population, the dissertation revealed that all three of Johnson's four coercive controlling types of behaviors (intimate terrorist, situational couple violence, and violent resister) are present among adolescents (Research Question 2). Additionally, this dissertation found a new typology, which is a nonphysical violence coercive controller. This typology is a youth who uses only coercive control against his or her partner with no physical violence. The nonphysical violence coercive controller may be unique to a youth sample because cognitive development is just starting to occur and therefore might not yet lead the controlling behavior to IPV perpetration. This exploration had two parts: a) looking at the entire sample's violent attitudes leading to controlling behaviors, which then leads to violent IPV perpetration and IPV victimization; and b) exploring different gender models'(i.e., male and female) violent attitudes leading to controlling behaviors, which then leads to violent IPV perpetration and IPV victimization. These models found that controlling behaviors did mediate violent 59 IPV perpetration and IPV victimization and that there were gender differences (Research Question 3). Findings from the model exploring the entire sample (boys and girls combined) revealed the powerful effects of controlling behaviors on intimate partners, and the difference between violent attitudes only and violent behaviors (IPV perpetration and IPV victimization) towards an intimate partner. The present dissertation found the coercive effects of control and IPV perpetration on intimate partners. Findings from the study revealed that coercive control did mediate the predictive relationship between controlling behaviors and all three types of IPV perpetration (physical, sexual, and emotional/psychological). However, if participants did not have controlling behaviors and only had violent attitudes, their level of IPV perpetration was reduced. In addition, the findings reveal the coercive effects of control on IPV victimization of an intimate partner. These findings suggest that coercive control not only leads to violent perpetration but also leads to violent victimization. Although it was expected that an individual's coercive control would have a direct-effect on perpetration, it was not expected that it would have a direct-effect on IPV victimization. This leads to a unique view of the complex nature of the consequences of IPV and controlling behaviors. An exploration of gender differences in IPV perpetration and IPV victimization revealed no gender differences. This finding was expected, because the instrument used to find IPV perpetration and IPV victimization looked at both perpetration and victimization of the participants, which is similar to studies using the Conflict Tactic Scale and similar to the family violence perspective studies (e.g., Straus & Gelles, 1990). However, the present study did find gender differences in regard to females' IPV 60 perpetration and IPV victimization being reduced when controlling behaviors are not present. This may be explained through the large difference between having violent attitudes and actually acting out the violent behaviors. The leap from violent attitudes to violent behaviors is large, and one that most people do not take. However, once you take that leap into violent behaviors (controlling or physical), you open yourself up to being more likely to perpetrate and also become a victim. For example, many people get upset at someone for a variety of reasons and may say to themselves, "I am going to hurt that person" (violent attitude), but they never do (violent behavior). Additionally, society's acceptance of males' use of violence as a means to resolve conflict is in contrast to societal norms that do not as readily accept females' use of violence. This may explain why women with violent attitudes and no controlling behaviors are less likely to perpetrate violence than their male counterparts. The data also show sex symmetry when analyzing violent behaviors. Findings reveal that, when controlling behaviors are present, both males and females are more likely to perpetrate all three different types of IPV (physical, sexual, and emotional/psychological). Therefore, regardless of whether you are female or male, if you have controlling behaviors, you are more likely to perpetrate violence. Researchers should be aware of the possible differences in the different types of IPV measured. Sex symmetry also was found when analyzing IPV victimization, as both males and females reported IPV victimization when violent attitudes and controlling behaviors were present. The findings support earlier research that found males and females use violence equally to control their intimate partner. It is possible that when you put yourself in a risky position with violent behaviors, you may be just as likely to be hurt (victim). 61 This could begin to explain the complex nature of violence and how it is likely that one can be both perpetrator and victim at the same time. For example, when moving into the arena of violent behaviors, you may actually be hurt more than you hurt others. Therefore, one should be cautious about using any type of violence, even if it is out of self-defense, because it could also lead to victimization. According to the dissertation's findings, coercive controlling behaviors are a direct link to IPV victimization and IPV perpetration. Coercive controlling behaviors, such as intimidating an intimate partner to do what they want, and threatening the partner that they will leave the relationship, had significant negative effects on intimate partners. Coercive control was so powerful that if it was absent in an intimate relationship, there was a reduction in IPV. These findings lead me to conclude that coercive control is a form of psychological violence that leads to both partner violence victimization and perpetration. These findings are supported through feminist literature, which states that verbal threats and intimidation are forms of violence (Kelly, 1998). In the current study sample, the prevalence of IPV was considerable, with 46% of respondents reporting emotional/psychological violence, 34% reporting physical violence, and 17% reporting sexual violence. These results mirror findings from other youth samples (Cornelius & Resseguie, 2006). Further, 66% of the current youth sample is in an intimate relationship and mutually violent couples are common, which is also reflected in the adult literature on IPV (Straus & Gelles, 1990). These findings point to a critical need to address IPV among youth through dating violence prevention programs. Results of the dissertation reveal that controlling typologies do exist among youth who are in intimate relationships. About 60% of the sample fit into one type of coercive 62 control typology (intimate terrorist, situational couple violence, and nonphysical violence coercive controller). Youth samples have rarely if ever explored specific types of controlling behaviors, which makes this dissertation unique. Limitations of the Study There are several limitations to the present dissertation. The first limitation is the threat to internal validity of the study due to problems with instrumentation. Structural equation modeling revealed that there was a floor effect, which means that most of the IPV perpetration and IPV victimization was not reported by the participants, as the scales used mainly explored antisocial behaviors; since the sample is not identified and targeted risk behaviors, most individuals do not have these antisocial behaviors. These results show a Poisson regression that is often used for rare events, which reveals variables with non-negative integer values and can be useful for variables that often include zero. Another study limitation is that there was no treatment effect when testing the four-session RWV program from pretest to posttest. Future research on the RWV should include longer term follow-up on the participants, which may reveal significant behavior change from treatment to control group. An additional limitation is the selection of the study sample, which is also a threat to internal validity. The sample for this dissertation was high school students in health classes, which is commonly found in the family violence perspective literature. Therefore, the sample itself may have contributed to the effects or lack of effects found. For instance, the findings of mutual violence between male and female participants might be due to the fact that it is a general high school sample. However, high school students 63 may have unique characteristics that lead to gender symmetry in intimate partner violence. Therefore, findings from this sample have concerns with generalizability, which becomes a concern to external validity. The external validity of the current study is limited, as high school students may have certain characteristics that do not generalize to the larger population. Although the sample was diverse in race/ethnicity, the results may not generalize from Salt Lake City, Utah to other cities or regions of the country. Nonetheless, study results can be used for further development of primary and secondary prevention programs addressing partner violence. A limitation that is present in research using surveys is validity of the participants' responses, which may be influenced by social desirability or memory. The participants from the present study may have underreported the frequency of violent perpetration for fear of social consequences. Although the surveys were anonymous and the participants were free to fill them out when and where it was most convenient, it is likely that other persons (e.g., students, friends, intimate partners) were in the area during their completion of the survey. Memory may also be a threat to the validity of the study, as participants were requested to recall incidents that may have occurred in their relationships within the past 6 months. Underreporting may not be due to social pressure but rather due to problems recalling an incident that occurred many months ago. 64 Strengths of the Study There are strengths to the present dissertation. The first strength is the testing of Johnson's violent control typology in intimate partner violence, which permitted exploring the presence of a coercive control typology among adolescents. Notably, a new typology was revealed, nonphysically violent coercive controller, which consists of coercive behaviors and attitudes towards violence without physical violence itself. Individuals fitting this typology likely will not be detected by law enforcement because the abuse is psychological, with an absence of physical violence. Future research needs to explore the effects of a "nonphysically violent coercive controller" on his or her intimate partner and how we can develop policy to help victims affected by such typology. Adolescence is an optimum time to address a "nonphysically violent coercive controller" because youth are starting to develop cognitively at this age and prevention programs with skill-building components targeting coercive control may prevent future physical abuse. Additionally, dating violence prevention programs, such as RWV, need to incorporate programming addressing coercive controlling behaviors and target intimate terrorists, situational couple violence, violent resisters, and nonphysically violent controllers. The final strength of the present dissertation is the model of controlling behaviors mediating IPV perpetration and IPV victimization. Notably, contributing to the model of controlling behaviors mediating IPV is the "female effect," by which females with no controlling behaviors reduced their IPV victimization and perpetration. Therefore, addressing males and females differently in regard to controlling behaviors in the RWV program might be a good starting point to process controlling behaviors specifically to 65 each gender. Additionally, the current dissertation revealed the predictive power of controlling behaviors on IPV perpetration and IPV victimization. Therefore, future RWV programming should reflect preventative skill building, e.g., helping participants learn to recognize controlling behaviors and how to minimize them. Since the "female effect" reveals a difference between men and women in regard to controlling behaviors, it is important to design skill-building programs that take this difference into account. This could include separating males and females to discuss society's expectations regarding gender roles and violent behavior, what coercion is (specific to their gender), and what that might look like in the real world (what they have seen or experienced). For example, societal norms reflect that it is okay for boys to be protective and hyperaggressive in their relationships (Fawson, 2011; Fawson & Prospero, 2009), which may influence a boy to use coercive tactics to control his partner. Implications for Social Work Practice The present study combined several areas of IPV theory (feminist, family violence, and Johnson's violent control typology) to answer three research questions: (1) This study evaluates a four-session RWV program in order to answer the following research questions: (1) Are the violent attitudes and behaviors of RWV participants reduced as measured by the Controlling Behavior Scale (modified), Justification for Dating Violence Scale, and Dating Violence Scale?; (2) Which of Johnson's four typologies are manifested among an adolescent population?; and (3) Are there differences in RWV program participants, particularly between males and females, in regard to violent attitudes and controlling behaviors having a direct predictive effect on IPV 66 perpetration (psychological/emotional, physical, and sexual) and IPV victimization (psychological/emotional, physical, and sexual)? The study revealed that couples with both members or one member with controlling behaviors are significantly more likely than couples without controlling behaviors to experience both IPV victimization (physical, sexual, and psychological/emotional) and IPV perpetration (physical, sexual, and psychological/emotional). Addressing the impact of coercive controlling behaviors on the different types (physical, sexual, psychological/emotional) of IPV perpetration and IPV victimization among youth who are dating is paramount in reducing violence from occurring. Practitioners and dating violence prevention programs need to provide different ways to identify and reduce these behaviors in individuals, keeping in mind the differences between males and females revealed in this study. The current dissertation also revealed different types of controlling behaviors present in youth (situational couple violence, intimate terrorists, violent resisters, and nonphysically violent controllers). These different types of controlling behaviors should individually be addressed in future dating violence prevention programs to potentially reduce violence in relationships. Understanding that controlling behaviors are a powerful predictor of partner violence experienced by victims and perpetrators provides significance guidance for practitioners who work with youth suffering from dating IPV. For example, if a practitioner has a client (youth) who is involved in reciprocal violence, the practitioner should assess whether the couple is experiencing situational couple violence and what to what degree control is a factor in their relationship. However, since situational couple 67 violence is low on control, the practitioner could provide a service plan involving goals and objectives with strategies that would improve the couple's and possible family's social and stress management skills. If a practitioner has a client who is involved in controlling violence, the practitioner should assess whether the couple is experiencing intimate terrorism, violent resistance, or a nonphysically violent controller and to what degree control influences their relationship. If the practitioner determines the couple is involved in violence, with one partner being an intimate terrorist or nonphysically violent controller, the treatment plan would then focus the services on reducing coercive attitudes, beliefs, and behaviors of the individuals, such as addressing the harmful effects of "owning" their partner that allows for the use of various controlling behaviors (intimidation, threats, isolation). However, there may be difficulties for social workers in assessing what type of controlling behavior is going on because these types may not be mutually exclusive in the "real world." Dating violence prevention programs that work with youth who experience IPV should focus services on different types of controlling tactics and improving conflict resolution skills. For example, dating violence prevention programs should identify controlling behaviors (e.g., threats, intimidation, isolation, and different skills to deal with these behaviors, including both healthy and unhealthy approaches), which will help the youth develop healthy conflict resolution skills. Therefore, when dating violence prevention programs address different types of controlling behaviors, they can address control and violence together (intimate terrorists/violent resister), mutual violence (situational couple violence), and control with no physical violence (nonphysical violent coercive controller). For instance, both boys and girls can be socialized to believe that an 68 intimate partner is "property" that belongs to them and that coercion is an appropriate method of ensuring that they keep their "property." Additionally, different types of control tactics are often a social norm in dating relationships. This can be seen, on one hand, when boys in an intimate relationship push their female partner to say "yes" when their partner says "no," which is forced consent. On the other hand, females in an intimate relationship are socialized to say "no" until their partner deserves or is rewarded with a "yes," and this can be done through the male proving that he deserves the "yes." These types of coercive tactics, which society teaches both boys and girls as normal behaviors when dating, can dangerously lead to some type of violence. Dating violence prevention programs should resocialize youth to understand that controlling behaviors and violence are not acceptable methods of addressing conflict and teach them that power and control are not the fundamental elements of an intimate relationship. Implications for Social Work Policy The knowledge of different types of IPV violence and IPV victimization with controlling behaviors among youth calls attention to potential policy changes in high schools regarding awareness of IPV. Raising awareness in high schools of IPV and the effects that these types of violence have on their students can lead to more appropriate measures to address these issues, such as more involvement from teachers, parents, and students in dating violence prevention programs. If principals, teachers, parents, and students understand that coercive controlling behaviors are a strong predictor of future intimate partner violence perpetration and victimization, appropriate measures can be 69 taken by the parties involved to support the best interventions for both victims and perpetrators. The present dissertation has implications for families and society in that sexism needs to be addressed in the family, school, and the larger community. For example, male dominance in the family contributes significantly to violence within the family (Straus, Gelles, & Steinmetz, 1980). Patriarchal social structures that promote the use of violence to achieve power and control over others socialize girls and boys to believe that these types of dominance are normal and culturally appropriate. Additionally, if teachers in the high school reflect these patriarchal social structures of power and control, students learn that these forms of dominance are culturally accepted. Therefore, addressing male dominance, patriarchal social structures, and the influence of coercive attitudes on youth is essential at the family and school level. Implications for Social Work Research Future research of the RWV and other dating violence prevention programs should address longer follow-up time (1 to 2 years). Researchers who are able to provide longer follow-up and collect longitudinal data will be able to more accurately assess a program's effectiveness in reducing violent attitudes and behaviors. This type of research model will require high schools and researchers to work together to provide best strategies to access and follow students over a longer period of time. Additionally, future research should explore noncontrolling men and women who refuse to be controlled. Future research on dating violence prevention programs should also test Johnson's control typology. These findings may continue to help address types of 70 controlling behaviors among youth in intimate relationships and different paths between each typology. For example, the present study found that about 60% of the youth sample fit into different typologies of control (18% intimate terrorist, 21% situational couple violence, and 20% nonphysically violent controller) and explored how these typologies influence IPV perpetration and IPV victimization among youth. Additionally, the study revealed a new coercive controlling typology, the "nonphysically violent controller," and future research should explore this typology to see if the finding can be replicated. Finally, future research should focus on investigating the etiology of coercion (what causes some individuals to use coercion and not others?) and coercion's relationship to IPV (are the two factors subtypes of the same construct?). Future research should employ a mixed method design, i.e., both quantitative and qualitative. Findings from mixed method studies would provide a deeper and richer explanation of the lived experiences of youth who are dating and their experience with IPV and controlling behaviors. Ultimately, addressing violence as a human problem and understanding that all citizens play a role in ending violence might help develop new social structures, where coercion and violence are not accepted or valued in society and partner violence can be significantly reduced. APPENDIX PRE- AND POSTTEST INSTRUMENTS 72 Part 1: Background Information ID# 1. What is your sex: (1) Male (2) Female 2. Age______ 3. Race/Ethnicity (check ALL that apply): Asian Native American Other: African American White Hispanic ______ Pacific Islander 4. Your parents' marital status: (1) Married and living together (4) Widowed (2) Married and living separately (5) Remarried (3) Divorced (6) Others, please describe: ____________ 5. What grade are you in? 6. Your average grades of last academic year: (1) A (4) D (2) B (5) F (3) C 7. Sexual orientation: Heterosexual Lesbian Gay Bisexual 8. Which following description suits your intimate relationship status? (1) I have been or currently am in an intimate relationship in the past that lasted at least 1 month. (2) I have never been in an intimate relationship in the past. 9. Your partner's sex: (1) Male (2) Female 73 Part 2: The following statements are about you. Please read each statement and decide how much you agree with it. THERE ARE NO RIGHT OR WRONG ANSWERS. (1) Strongly disagree (2) Disagree (3) Agree (4) Strongly agree 1 I get aggressive if my partner is trying to make me look stupid. (1) (2) (3) (4) 2 I believe that physical aggression is necessary to get through to my partner. (1) (2) (3) (4) 3 During an intense argument, I get physically violent. (1) (2) (3) (4) 4 I get aggressive if my partner makes me look bad in public. (1) (2) (3) (4) 5 I believe that my aggression makes my partner act like I want them to. (1) (2) (3) (4) 6 During a physical fight with my partner, I know exactly what I'm doing. (1) (2) (3) (4) 7 I feel that my aggression comes from being pushed too far by my partner. (1) (2) (3) (4) 8 I will look for help if I am getting too violent with my partner. (1) (2) (3) (4) 9 I know where to ask for help if my partner is getting violent with me. (1) (2) (3) (4) 10 Violence is part of any relationship so I do not need help if we are violent. (1) (2) (3) (4) Part 4: Here is a list of things you and your partner may have done during your relationship. Indicate the frequency of the behavior that describes your actions toward your partner and your partner's actions towards you. (0) Never (6) Always Part 3: Here is a list of things that might happen between dating partners. Please read each statement and decide how much you agree with it. THERE ARE NO RIGHT OR WRONG ANSWERS. (1) Strongly disagree (2) Disagree (3) Agree (4) Strongly agree It is ok for a girl to hit her boyfriend if ： 1 He threatens her that he would leave the relationship. (1) (2) (3) (4) 2 He hits her first during an argument. (1) (2) (3) (4) 3 He is caught having an affair. (1) (2) (3) (4) 4 He calls her unpleasant names. (1) (2) (3) (4) 5 He is flirting with other girls. (1) (2) (3) (4) 6 He insults her in front of friends. (1) (2) (3) (4) 7 He does not allow her to go out with friends at night. (1) (2) (3) (4) 8 He is unwilling to have sex. (1) (2) (3) (4) 9 He disobeys her. (1) (2) (3) (4) It is ok for a boy to hit his girlfriend if ： 1 She threatens him that she would leave the relationship. (1) (2) (3) (4) 2 She hits him first during an argument. (1) (2) (3) (4) 3 She is caught having an affair. (1) (2) (3) (4) 4 She calls him unpleasant names. (1) (2) (3) (4) 5 She is flirting with other boys. (1) (2) (3) (4) 6 She insults him in front of friends. (1) (2) (3) (4) 7 She does not allow him to go out with friends at night. (1) (2) (3) (4) 8 She is unwilling to have sex. (1) (2) (3) (4) 9 She disobeys him. (1) (2) (3) (4) 74 When we have conflicts…… 1 My partner accused me of being a lousy lover. (0) (1) (2) (3) (4) (5) 6 2 My partner threw something at me. (0) (1) (2) (3) (4) (5) 6 3 My partner pushed, grabbed, slapped, or shoved me. (0) (1) (2) (3) (4) (5) 6 4 My partner kicked, bit, or hit me with a fist. (0) (1) (2) (3) (4) (5) 6 5 My partner choked me. (0) (1) (2) (3) (4) (5) 6 6 My partner used a knife or a weapon to hurt me. (0) (1) (2) (3) (4) (5) 6 7 My partner threatens me that he/she will leave the relationship. (0) (1) (2) (3) (4) (5) 6 8 My partner threatens me that he/she would harm themselves. (0) (1) (2) (3) (4) (5) 6 9 My partner threatened to hurt my family and friends. (0) (1) (2) (3) (4) (5) 6 10 My partner prohibited me from contacting friends. (0) (1) (2) (3) (4) (5) 6 11 My partner harassed me by phone calls, text, and etc. (0) (1) (2) (3) (4) (5) 6 12 My partner stalked me. (0) (1) (2) (3) (4) (5) 6 13 My partner is aggressive towards my pets. (0) (1) (2) (3) (4) (5) 6 14 My partner touched my body in an unwanted way. (0) (1) (2) (3) (4) (5) 6 15 My partner used force to make me have sex. (0) (1) (2) (3) (4) (5) 6 16 My partner made me have sex without using a condom. (0) (1) (2) (3) (4) (5) 6 17 My partner threatened to show naked pictures of me. (0) (1) (2) (3) (4) (5) 6 (0) Never (6) Always When we have conflicts, …… 18 I accused my partner of being a lousy lover. (0) (1) (2) (3) (4) (5) (6) 19 I threw something at him/her. (0) (1)