Child Maltreatment

Child Maltreatment http://cmx.sagepub.com/

Pathways From Childhood Maltreatment to Emerging Adulthood: Investigating Trauma-Mediated Substance Use and Dating Violence Outcomes Among Child Protective Services−Involved Youth Breanne Faulkner, Abby L. Goldstein and Christine Wekerle Child Maltreat published online 6 October 2014 DOI: 10.1177/1077559514551944

The online version of this article can be found at: http://cmx.sagepub.com/content/early/2014/10/01/1077559514551944

Published by: http://www.sagepublications.com

On behalf of:

American Professional Society on the Abuse of Children

Additional services and information for Child Maltreatment can be found at: Email Alerts: http://cmx.sagepub.com/cgi/alerts Subscriptions: http://cmx.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://cmx.sagepub.com/content/early/2014/10/01/1077559514551944.refs.html

>> OnlineFirst Version of Record - Oct 6, 2014 What is This?

Downloaded from cmx.sagepub.com by guest on October 9, 2014

Original Article

Pathways From Childhood Maltreatment to Emerging Adulthood: Investigating Trauma-Mediated Substance Use and Dating Violence Outcomes Among Child Protective Services–Involved Youth

Child Maltreatment 1-14 ª The Author(s) 2014 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/1077559514551944 cmx.sagepub.com

Breanne Faulkner1, Abby L. Goldstein1, and Christine Wekerle2

Abstract Longitudinal survey data were used to examine the relationship between two types of childhood maltreatment, abuse/neglect and exposure to intimate partner violence (IPV), and two outcomes, substance use and dating violence, within the past year. Participants were youth (N ¼ 158, aged 16–19 at Time 3) involved with child protective services (CPS). A parallel multiple mediator model was used to test the hypothesis that trauma symptoms would mediate the relationship between both types of maltreatment and dating violence, marijuana, and alcohol use outcomes. Although both types of maltreatment were not directly associated with dating violence and substance use outcomes, the indirect effects of anxiety, anger, and dissociation on the relationship between maltreatment and substance use/dating violence were significant. Direct effects of both types of maltreatment on past year use of dating violence þ alcohol use and dating violence þ marijuana use were not significant, but results demonstrated a significant indirect effect for anger on the relationship between exposure to IPV and past year dating violence þ marijuana use. No other indirect effects were significant. Findings highlight the negative effects of exposure to IPV and have implications for the development of prevention programming for youth transitioning out of CPS. Keywords child maltreatment, domestic/intimate partner violence, substance abuse, adolescents, youth

Emerging adulthood is a stage of unique developmental challenge marked by greater independence and exploration as well as feelings of instability and uncertainty (Arnett, 2004). In addition, the transition into emerging adulthood represents a period of vulnerability for the development of a number of maladaptive behaviors, including increased risk for substance use and substance-related problems (Arnett, 2005; Tucker, Ellickson, Orlando, Martino, & Klein, 2005) as well as for becoming either a perpetrator or a victim of dating violence (Centres for Disease Control [CDC], 2010a; Statistics Canada, 2011); rates of dating violence as well as heavy alcohol use are highest among adolescents and emerging adults. Arnett (2005) suggested that substance use may constitute an aspect of identity development as young adults ‘‘try on’’ diverse experiences before settling into adulthood and that emerging adults may use substances to self-medicate when struggling with emotions related to instability. Youth who come in contact with personal, social, and environmental obstacles to identity development during or prior to emerging adulthood may also struggle with mature intimacy and may engage in behaviors disruptive to the development of positive relationships (Montgomery, 2005).

Childhood maltreatment may be one such obstacle. There is a large body of research establishing childhood maltreatment as a risk factor for multiple negative outcomes in both adolescence and adulthood, including interpersonal difficulties (Jaffee & Gallop, 2007; Wolfe, Scott, Wekerle, & Pittman, 2001); aggression, violence, and criminal behaviors (Casiano, Mota, Afifi, Enns, & Sareen, 2009; Gilbert et al., 2009; Ou & Reynolds, 2010); physical and mental health problems (MacMillan et al., 2001; Scott, Smith, & Ellis, 2010); and the development of substance abuse and dependence (Simpson & Miller, 2002; Tonmyr, Thornton, Draca, & Wekerle, 2010). Young people involved with the child protective services (CPS) system may be a particularly vulnerable group, given that, in addition to significant histories of

1 2

University of Toronto, Toronto, Ontario, Canada McMaster University, Hamilton, Ontario, Canada

Corresponding Author: Breanne Faulkner, OISE, University of Toronto, 252 Bloor St. W, Toronto, Ontario, Canada, M5S1V6. Email: [email protected]


2

Child Maltreatment

child maltreatment, they often experience multiple transitions, interrupted relationships, and a lack of continuity or consistency in activities, learning environments, friendships, and caregiver relationships (Fong, Schwab, & Armour, 2006; Healy & Fischer, 2011). Furthermore, the transition out of CPS, typically occurring during emerging adulthood, is likely to be fraught with multiple challenges for youth (Courtney & Dworsky, 2006; Keller, Cusick, & Courtney, 2007), further heightening their vulnerability for involvement in a variety of risk behaviors that are common among emerging adults within the general population. In light of numerous overlapping risk factors experienced by emerging adults with maltreatment histories and CPS involvement, research examining the mechanisms of risk-related outcomes among these youth is needed. The purpose of this study was to address gaps in the existing literature by examining outcomes associated with two dynamically related risk behaviors—alcohol/marijuana use and the perpetration of dating violence—among CPS-involved adolescents and emerging adults. It was hypothesized that experiences of abuse and neglect as well as exposure to domestic violence among caregivers would predict later likelihood of engaging in both alcohol and marijuana use and perpetration of dating violence in late adolescence and emerging adulthood; furthermore, it was expected that trauma symptoms would mediate these relationships.

Intimate Partner and Dating Violence In the United States, one in four women and one in seven men have experienced severe physical violence by an intimate partner (CDC, 2010a). In a recent survey of 19,000 Canadians, 6.4% of women reported that they were victims of physical violence at the hands of a partner or ex-partner within the previous 5 years (Statistics Canada, 2011). Victims of intimate partner violence (IPV; acts of physical aggression, psychological abuse, sexual coercion, or controlling or isolating behaviors toward one’s romantic partner or expartner) often suffer from serious and long-term health consequences, disruptions in employment, impaired psychosocial functioning, depression, self-harm, and even death (Afifi et al., 2009; Garcia, Soria, & Hurwitz, 2007; Swanberg, Logan, & Macke, 2005). Although rates of IPV are high among adults, prevalence is highest during adolescence and emerging adulthood. Individuals between the ages of 15 and 24 are both the most likely victims and perpetrators of IPV (CDC, 2010a; Statistics Canada, 2011)—or, more aptly, dating violence— with rates ranging from 10% using conservative definitions (i.e., physical or sexual violence; CDC, 2012) to up to 50% using less restrictive definitions (i.e., psychological and verbal abuse; Malik, Sorenson, & Aneshensel, 1997). Youth (i.e., adolescent and emerging adult) profiles of dating violence may differ in important ways from adult IPV. In contrast to the broad consensus that men perpetrate more severe violence and cause greater injury toward their partners than women (Archer, 2002; Statistics Canada, 2011), research

supports a more equal gender distribution for both victimization and perpetration of mild, moderate, and severe dating violence within youth couples (Foshee, 1996; Schnurr, Lohman, & Kaura, 2010; Windle & Mrug, 2009). Dating relationships also differ from adult relationships in terms of gendered power dynamics, experience and social skill development, and peer influence (Shorey, Cornelius, & Bell, 2008). During early experiences with romantic relationships, youth are largely unfamiliar with normative expectations and boundaries and may not immediately regard dating violence as problematic (Wekerle & Wolfe, 1999). Importantly, dating violence in adolescence and emerging adulthood is a risk factor for continued perpetration in later adulthood, particularly among youth already at risk (Capaldi, Shortt, & Crosby, 2003; Gomez, 2011). Wolfe and his colleagues (Wolfe et al., 2003) suggest that these early patterns of abusive behavior may be more experimental or less rigidly established and thus argue that adolescence represents an important opportunity to intervene with potential IPV perpetrators and to teach more adaptive, nonviolent relationship skills. Establishing more clearly the risk factors and underlying mechanisms of teen dating violence perpetration is essential to improving preventative and intervention efforts.

Substance Use and Dating Violence Alcohol and drug use is a risk factor for the perpetration of IPV that is particularly relevant to young populations. As with the typical age of initiation of dating relationships, initiation of alcohol use usually takes place during adolescence (Adlaf & Paglia-Boak, 2009; Health Canada, 2010) and escalates into emerging adulthood, with the highest rates of alcohol use and heavy drinking among individuals aged 15 to 25 (Health Canada, 2010; Substance Abuse and Mental Health Services Administration [SAMHSA], 2012). Although the relationship between alcohol use and perpetration of IPV is well documented in adult and college student samples (see Stuart et al., 2009), less is known about this relationship among adolescents and other samples of emerging adults. Furthermore, although marijuana is the most widely used illicit drug in both Canada and the United States, particularly among youth (Health Canada, 2010; SAMHSA, 2012), and has received attention as a potential risk factor for perpetration of IPV (e.g., Moore & Stuart, 2005), very little research has been directed at examining the relationship between marijuana and perpetration of dating violence among youth. Much of the research examining the relationship between substance use and IPV perpetration has assumed a causal link as a result of intoxication-related aggression; however, these behaviors may also be related in more distal ways. Especially within the context of the experimentation and identity confusion of emerging adulthood, in which relationship patterns begin to emerge, substance misuse and dating violence may together create dynamic risk for increased engagement in each as well as for the potential that dating violence behaviors begin to occur within the specific context of


Faulkner et al.

3

intoxication. As research suggests that violence occurring while intoxicated is more severe than violence occurring while sober (Testa, Quigley, & Leonard, 2003), understanding the precursors to a more direct link is an important endeavor.

Childhood Maltreatment, Substance Use, and Dating Violence A history of childhood maltreatment has also been linked to later perpetration of dating violence (Wekerle & Wolfe, 2003; Wekerle et al., 2009b; Wolfe et al., 2003) as well as to increased adolescent alcohol use and alcohol problems (see Tonmyr et al., 2010 for review), and marijuana use and dependence (Rogosch, Oshri, & Cicchetti, 2010; Wekerle, Leung, Goldstein, Thornton, & Tonmyr, 2009a). Although the link between childhood maltreatment and a variety of negative outcomes in adolescence and young adulthood is well established, the existing literature on this topic is primarily focused on the impact of physical and sexual abuse and, with the exception of a few studies, has not adequately addressed the effects of exposure to IPV among caregivers; this is true for research examining both substance use and IPV outcomes (e.g., Gomez, 2011; Moran, Vuchinich, & Hall, 2004; Widom, Czaja, & Dutton, 2013; Wekerle et al., 2009b; see Tonmyr et al., 2010). Recent research has identified exposure to IPV as having a significant role in the etiology of problem behaviors including substance misuse (Hamburger, Leeb, & Swahn, 2008; Kilpatrick et al., 2000; Tonmyr et al., 2010) and perpetration of dating violence (Gover, Park, Tomsich, & Jennings, 2011; Roberts, Gilman, Fitzmaurice, Decker, & Koenen, 2010). Older youth who are involved with CPS may be at particular risk due to the overlapping transition into emerging adulthood and out of CPS (Fong et al., 2006; Healy & Fischer, 2011). During this transition, critical support is lost, particularly for youth whose guardianship is provided by the state (i.e., Crown Wards in Canada). Little research has examined the dynamic role of maltreatment in creating risk for the interrelated problem behaviors of both substance use and dating violence. In one of the few studies examining the relationship between these variables cross-sectionally, Barnett and Fagan (1993) found that male perpetrators of IPV used more alcohol than their nonabusive counterparts and had also experienced more maltreatment and observed significantly more IPV between caregivers during their childhoods than nonabusive men. In addition, researchers have found evidence to support a causal pathway from a history of physical maltreatment in childhood to an increased likelihood of drinking problems and, in turn, to higher rates of IPV perpetration in adulthood (Schafer, Caetano, & Cunradi, 2004; White & Widom, 2003). Research efforts examining the association between marijuana use and perpetration of IPV are in much earlier stages, and we are not aware of any other investigations of the contribution of maltreatment history to dating violence perpetration that have also assessed the role of marijuana use.

Trauma symptoms as a shared mechanism. The trauma literature has much to contribute to the investigation of these complex relationships. Research has supported trauma symptomology as one pathway through which childhood maltreatment may lead to the development of both substance use problems (Goldstein et al., 2011; White & Widom, 2008) and dating violence (Wekerle et al., 2009b; Wolfe, Wekerle, Straatman, & Grasley, 2004). DeBellis’ (2001) developmental traumatology theory argues that childhood maltreatment impacts a number of psychobiological systems, resulting in, among other things, compromised arousal control, behavioral regulation, and executive functioning which might underlie harm to self or others and contribute to negative emotions, which may increase risk for ‘‘self-medicating’’ with alcohol or other drugs (DeBellis, 2001, 2002). In one study, researchers found that trauma symptoms fully mediated the effects of trauma exposure on both physical and psychological dating violence in college students and that higher levels of posttraumatic stress disorder (PTSD) were related to greater use of dating aggression (Taft, Schumm, Orazem, Meis, & Pinto, 2010). Using a sample of CPSrecruited youth from the Maltreatment and Adolescent Pathways (MAP) study, Wekerle and colleagues (Wekerle et al., 2009b) found that emotional abuse predicted trauma symptoms even after controlling for covariates and other types of childhood maltreatment; moreover, for males, the number of trauma symptoms mediated the relationship between emotional maltreatment and perpetration of dating violence. Researchers have also demonstrated that PTSD plays a mediating role between the experience of maltreatment and illicit drug use in adulthood (Plotzker, Metzger, & Holmes, 2007; White & Widom, 2008). Using another subset from the MAP, Goldstein and colleagues (2011) found that trauma symptoms (specifically, anger and dissociation) were a significant predictor of alcohol and illicit drug use and associated problems, even after controlling for age, sex, CPS status, and childhood maltreatment. Similarly, in an earlier study with the MAP sample, researchers found that trauma symptoms predicted past year marijuana use for female adolescents, over and above childhood maltreatment (Wekerle et al., 2009a). In sum, little is yet known about the nature of the relationships between childhood maltreatment, substance use, and dating violence among youth, particularly those who are at risk for substance use and dating violence behaviors based on their maltreatment histories and age (e.g., older youth making the transition out of child welfare and into emerging adulthood). Furthermore, a consideration of the role of exposure to IPV in predicting dating violence and substance use is lacking from the existing literature. Understanding the complex dynamics of these vulnerabilities and risk behaviors within both adolescent and emerging adult populations has important implications for the development of more adequate prevention, intervention, and education efforts for youth involved with CPS. Using data from a longitudinal survey, we examined the temporal nature of these relationships among CPS-involved youth in adolescence and early emerging adulthood, and investigated the


4

Child Maltreatment

mediating role of trauma symptoms on the relationship between childhood maltreatment, including exposure to IPV, and the occurrence of both alcohol and/or marijuana use and perpetration of dating violence over the past year.

Method Participants and Procedure Participants were adolescents and emerging adults involved in the MAP longitudinal study. The MAP examined multiple outcomes among youth (N ¼ 561, aged 14–17 at initial assessment), randomly selected from CPS caseloads (i.e., case management, resource linkages, out-of-home care, etc.) within a large urban catchment area in Ontario. Rolling recruitment occurred over 6 years, with all initial data collection occurring between 2003 and 2009. Data for each participant were then collected at 6-month intervals over 3 years, following developmental trajectories from adolescence into emerging adulthood. Ethics approval for the MAP was provided by both CPS and respective universities. Additional recruitment, consent, and procedural information may be found in Wekerle et al. (2009b). This study involved examining the relationship between childhood maltreatment assessed at the initial session (Time 1), trauma symptoms assessed at the 1.5-year follow-up (Time 2), and alcohol and marijuana use as well as perpetration of dating violence assessed 6 months later at the 2-year follow-up (Time 3). These particular time points were selected to capture the chronological relationship between child maltreatment, trauma symptoms, and the past year occurrence of both substance use and dating violence and to follow these trajectories from adolescence through emerging adulthood. Due to the transitional circumstances of many of the youth initially recruited for this study, in particular as a result of their aging out of the CPS system, the initial MAP sample of 561 participants was subject to significant attrition over its longitudinal course (for a more detailed discussion of these issues, see Wekerle et al., 2011). As a result, at the 1.5-year follow-up time point (Time 2), a total of 269 participants (59% female; 64% Crown Ward) were tested, a retention rate of 47% from baseline. At the 2-year follow-up time point (Time 3), a total of 218 youth (61% female; 63% Crown Ward) were tested, a retention rate of 81% from the previous time point. The current results are based on data from the 158 participants (62.7% female; 58.9% Crown Ward) remaining in the study at Time 3 who indicated that they had initiated dating. Participants ranged in age from 16 to 19 (M ¼ 17.89, SD ¼ 0.98) at Time 3. Half (47.5%) of the participants were living with foster parents at the time of initial assessment, 19.6% were living in group homes, 8.9% with a single parent, and the remainder were living with one biological and one stepparent (5.1%), on their own or with roommates (5.1%), with two biological parents (3.2%), with other relatives (1.9%), or in some alternative living arrangement (3.8%). The majority (58.9%) of participants were Crown wards; in Ontario, a Crown

ward is a child or youth for whom the government has assumed parental responsibility, given that parental rights have been terminated. Participants indicated that they had been involved with CPS for between 6 months and 17 years (M ¼ 5.6 years, SD ¼ 4.2). The majority of youth self-identified as Caucasian (25.8%), African or Caribbean Canadian (26.5%), or mixed race (30.3%), with the remainder identifying as Hispanic, Chinese, Aboriginal, South Asian, and West or South East Indian. Approximately three quarters (79.7%) of participants identified as heterosexual, 4.4% identified as bisexual, and 2.5% as Gay or Lesbian. The remaining 13.3% (n ¼ 21) of participants indicated that they did not know their sexual orientation or declined to answer. On average, participants had begun dating at the age of 14 (SD ¼ 2.48). Analyses were conducted to examine the possibility of systematic loss of participants between initial assessment (N ¼ 561) and Time 3. Chi-square analyses revealed that, compared to male youth, a greater proportion of female youth remained in the study from the initial assessment to Time 3, w2(1, N ¼ 556) ¼ 6.19, p < .05. In addition, a disproportionate number of Crown ward youth remained in the study compared to youth classified under other CPS statuses, w2(1, N ¼ 556) ¼ 9.37, p < .05. Independent sample t-tests revealed that youth who remained in the study did not differ in age, total childhood maltreatment, or age of initiation of binge drinking, alcohol use, or marijuana use from those who left the study. Because many youth had not initiated dating until Time 3, comparisons for dating violence were not possible.

Measures Childhood maltreatment. History of childhood abuse and neglect were assessed with the 28-item Childhood Trauma Questionnaire–Short Form (CTQ-SF; Bernstein et al., 2003). The CTQ-SF includes 25 clinical items measuring the frequency of physical, sexual, and emotional abuse and physical and emotional neglect using a 5-point Likert-type scale (1 ¼ Never true, 2 ¼ Rarely true, 3 ¼ Sometimes true, 4 ¼ Often true, and 5 ¼ Very often true) as well as 3 items testing validity. A subscale score was generated for each type of maltreatment and cutoff scores were calculated for establishing the prevalence of moderate to severe maltreatment experiences (Bernstein & Fink, 1998). Cronbach’s as for the CTQ subscales in this study ranged from a ¼ .86 to a ¼ .94, with the exception of the physical neglect scale (a ¼ .69). A total score was calculated based on the sum of all CTQ abuse and neglect items (Total abuse/neglect). Two items from the Childhood Experiences of Victimization Questionnaire (CEVQ; Walsh, MacMillan, Trocme, Jamieson, & Boyle, 2008) were used to assess witnessing IPV. Participants indicated on a 5-point Likert-type scale how often they had been exposed to aggression between caregivers in the home (0 ¼ Never, 1 ¼ 1 to 2 times, 2 ¼ 3 to 5 times, 3 ¼ 6 to 10 times, 4 ¼ More than 10 times). The CEVQ items were summed for a total score (total witnessing IPV).


Faulkner et al.

5

Trauma symptoms. The Trauma Symptom Checklist for Children (TSCC; Briere, 1996) was used to assess six areas of trauma-related functioning, namely anxiety, depression, posttraumatic stress (PTS), dissociation, anger, and sexual concerns. Participants rated how frequently they experienced specific events, ‘‘things that youth sometimes think, feel, or do,’’ on a 4-point Likert-type scale, ranging from 0 (Never) to 3 (Almost all of the time). For example, ‘‘going away in my mind, trying not to think’’ or ‘‘wanting to yell or break things.’’ Cronbach’s as for TSCC scales in the current sample ranged from a ¼ .84 to a ¼ .91. Alcohol and marijuana use and problems. Past 30-day frequency (0 ¼ 0 times, 1 ¼ 1–2 times, 2 ¼ 3–7 times, 3 ¼ 7 or more times) of binge drinking (i.e., consuming five or more alcoholic drinks on a single occasion) and past 30-day marijuana use were assessed using items from the 2009 National Youth Risk Behavior Survey (CDC, 2010b). The 23-item Rutgers Alcohol Problem Index (RAPI; White & Labouvie, 1989) was used to assess the frequency of alcohol problems in the past year. Response options range from 0 (Never) to 4 (More than 10 times); items were summed to create a total RAPI score. The RAPI possesses good internal consistency, adequate test–retest reliability, and correlates moderately well with other problem drinking markers (White & Labouvie, 1989). Cronbach’s a for the current sample was a ¼ .92. No measure of problem marijuana use was included in this study. Dating violence. Perpetration of dating violence was assessed with a 35-item subset of the Conflict in Adolescent Dating Relationships Inventory (CADRI; Wolfe et al., 2001), which assesses the frequency with which various behaviors are used during a conflict or argument between dating partners over the past year. The CADRI is specifically geared toward the dynamics of adolescent relationships. Items are rated on a 4-point Likert-type scale, 0 ¼ Never; 1 ¼ Seldom (1–2 times); 2 ¼ Sometimes (3–5 times); and 3 ¼ Often (6 or more times). For example, ‘‘I kicked, hit, or punched [him/her]’’; ‘‘I kissed [him/her] when [he/she] didn’t want me to’’; and ‘‘I said things just to make [him/her] angry.’’ Five types of dating violence are assessed including physical abuse, sexual abuse, verbal/emotional abuse, relational abuse, controlling behaviors, and threatening behaviors. In this study, analyses focused on physical, sexual, threatening, and verbal/emotional abuse only. The CADRI has demonstrated strong internal consistency and 2-week test–retest reliability (Wolfe et al., 2001) as well as acceptable partner agreement (Wolfe et al., 2003). Cronbach’s as in the current sample were a ¼ .84 for physical dating violence, a ¼ .63 for sexual dating violence, a ¼ .88 for verbal/emotional dating violence, and a ¼ .76 for threatening dating violence. CADRI subscale scores were summed to create a total dating violence variable. Dating violence þ substance use. To capture past year temporal overlap of dating violence and substance use, three dichotomous variables were first created, (a) problem drinking, (b)

current marijuana use, and (c) moderate to severe dating violence. An individual received a score of 1 on the Problem Drinking variable if they reported binge drinking at least once in the past 30 days (38.6%, n ¼ 61) or obtained a score of 15 or higher on the RAPI (10.1%, n ¼ 16; see Thombs & Beck, 1994). Current Marijuana Use was defined as marijuana use one or more times in the past 30 days (45.6%, n ¼ 72). Cutoff scores to capture perpetration of at least Moderate Dating Violence were established based on previous research (Roudsari, Leahy, & Walters, 2009; Wolfe et al., 2001). For physical, sexual, and threatening dating violence, perpetration of any act was considered moderate dating violence. For verbal/emotional abuse, perpetration 3–5 times or more in the past year of any subscale item was considered moderate dating violence. A positive score on one or more subtype constituted Moderate Dating Violence (80.4%, n ¼ 127), capturing moderate to severe dating violence perpetration. Past year occurrence of both dating violence and substance use was operationalized as having perpetrated any Moderate Dating Violence as well as having engaged in (a) problem alcohol use (Dating Violence þ Problem Drinking) or (b) marijuana use (Dating Violence þ Marijuana Use).

Statistical Analysis Prior to conducting primary analyses, the distributions of all variables were examined for deviations from normality. Due to positive skewness, drinking problems (RAPI) and dating violence (CADRI) scores were log transformed. Bivariate correlations were examined to identify significant relationships between the independent and the dependent variables as well as potential mediators. Tests of mediation were conducted using the PROCESS macro (Hayes, 2013), which integrates multiple mediation functions in SPSS Regression Models. We hypothesized that all trauma symptom subtypes could mediate the relationship between child maltreatment and the substance use/dating violence outcomes. As a result, we used a parallel multiple mediator model (Hayes, 2013) wherein all trauma symptom subscales were included as mediators. Analyses were structured in such a way that the temporal nature of the hypothesized relationships could be tested with predictors assessed at Time 1, mediators assessed at Time 2, and outcomes assessed at Time 3. Specifically, we tested whether trauma symptoms mediated the relationships between two predictor variables (child abuse/neglect [total CTQ scores] and witnessing IPV [total CEVQ scores]) and five outcome variables (dating violence, problem drinking, marijuana use, Dating Violence þ Problem Drinking, and Dating Violence þ Marijuana Use). This resulted in 10 separate multiple mediation models. A general outline of the mediation model is illustrated in Figure 1. For each model, we tested: (1) the direct effects of the predictor on the outcome variable; (2) the total effect of the predictor on the outcome; (3) the direct effects of the predictor on the mediators; (4) the direct effect of the mediators on the outcome; and (5) the indirect effect of the predictor on the outcome via each mediator.


6

Child Maltreatment

T2 PTS

b1

T2 Anxiety

a1

b2

a2 a3

T2 Depression

b3 T3 Substance Use Dating Violence Substance Use + Dating Violence

c'

T1 Child Maltreatment a4 a5 a6

b4 T2 Anger b5 T2 Dissociation

b6

T2 Sexual Concerns

Figure 1. Parallel multiple mediator model depicting trauma symptoms as mediators of the relationship between child maltreatment and substance use/dating violence outcomes. Note. T1 ¼ Time 1; T2 ¼ Time 2; T3 ¼ Time 3. PTS ¼ posttraumatic stress. All models include gender as a covariate.

To test for mediation, bootstrap confidence intervals were estimated using bias-corrected bootstrap confidence intervals for all indirect effects (95% confidence intervals, z ¼ 1,000 bootstrap samples); when zero is outside of the 95% confidence interval, this indicates a significant indirect effect. Finally, gender was included as a covariate in all models due to significant gender differences in some dating violence variables (physical and verbal/emotional, with female youth reporting greater perpetration; p < .01) and binge drinking (with male youth reporting more binge drinking; p < .05). Although we examined correlations with age, we did not include age as a covariate as it was not significantly related to any of the dependent variables.

the most commonly reported at 77.2% (n ¼ 122); 36.7% threatened violence (n ¼ 58), 32.9% (n ¼ 52) used physical violence, and 11.4% (n ¼ 18) used sexual violence. Over one third of the sample (38.6%, n ¼ 61) reported at least one incident of binge drinking in the last 30 days and almost half (45.6%, n ¼ 72) had used marijuana at least once in the last 30 days. On average, scores on the RAPI were 4.93 (SD ¼ 8.83), and 10.1% (n ¼ 16) of individuals met the cutoff for problem drinking. In terms of past year occurrence of both dating violence and substance use, 37.3% (n ¼ 59) of participants fell into the Dating Violence þ Problem Drinking category and 38.6% (n ¼ 61) into the Dating Violence þ Marijuana Use category.

Results With respect to maltreatment histories, when responses were limited to at least moderate experiences of abuse (as defined by Bernstein & Fink, 1998), 48.1% (n ¼ 76) of participants reported emotional neglect, 44.9% (n ¼ 71) physical abuse, 43.7% (n ¼ 69) physical neglect, 41.8% (n ¼ 66) emotional abuse, and 21.5% (n ¼ 34) sexual abuse. In addition, 60.8% (n ¼ 96) of participants witnessed verbal IPV among caregivers and 38.6% (n ¼ 61) witnessed physical IPV. Multiple traumatization was common; whereas 16.5% (n ¼ 26) of participants reported experiencing moderate to severe maltreatment of only one type, 64.6% (n ¼ 102) reported experiencing two or more types. The remaining 19% (n ¼ 30) reported no experiences of moderate to severe maltreatment. Among subtypes of dating violence perpetration (dichotomized as defined earlier), verbal/emotional dating violence was

Bivariate Relationships Table 1 shows correlations between childhood maltreatment, trauma symptom, substance use, and dating violence variables. Both maltreatment variables were correlated with all trauma symptoms. All trauma variables were correlated with problem drinking, while only anger was correlated with binge drinking; no trauma symptoms were correlated with marijuana use. Only PTS, depression, and anger were correlated with total dating violence. Among substance use variables, only problem drinking was correlated with total witnessing IPV. Similarly, total dating violence was correlated only with total witnessing IPV. Finally, total witnessing IPV was significantly correlated with Dating Violence þ Marijuana Use. Dating Violence þ Problem Drinking was not significantly correlated with any maltreatment variable.


Faulkner et al.

7

Table 1. Bivariate Correlations Between Maltreatment, Trauma Symptoms as well as Substance Use, Dating Violence, and their Combination. Range

M (SD)

%

Gender

1

2

3

.33 .30 .24 .32 .30 .29

.86 .84 .65 .84 .57

4

5

6

7

8

9

10

11

12

13

Maltreatment 1 Total abuse/neglect 2 Total witnessing IPV Trauma symptoms

25–117 53.90 (20.69) 0–8 2.31 (2.51)

.24 .07

.40 .41 .43 .32 .25 .31 .19

3 PTS 4 Anxiety 5 Depression 6 Anger 7 Dissociation 8 Sexual concerns Substance use

0–30 0–27 0–27 0–27 0–30 0–30

6.20 (6.48) 4.13 (5.15) 4.92 (5.23) 5.49 (5.75) 6.27 (6.04) 4.41 (4.86)

.20 .29 .28 .22 .18 .06

9 Past 30-day binge 10 Problem drinking 11 Past 30-day marijuana use Dating violence perpetration 12 Total dating violence Dating violence þ substance use

0–3 0–92 0–3

0.61 (0.85) 4.93 (8.83) 0.97 (1.26)

.15 .11 .14

13 Dating violence þ problem drinking 14 Dating violence þ marijuana use

.02 .08 .11 .08 .09 .24 .16 .13 .07 .30 .32 .26 .31 .37 .35 .29 .59 .05 .13 .13 .05 .07 .12 .16 .14 .47 .45

.25

0–52 11.10 (10.36)

.81 .63 .69 .84 .79 .62 .55 .62 .55 .60

.07 .28 .20 .13 .21 .32 .15 .17 .14 .26 .26

0–1

37.3

.11

.04 .06 .21 .11 .25 .26 .22 .25 .79 .64 .45 .19

0–1

38.6

.02

.02 .19 .21 .12 .18 .27 .22 .25 .54 .51 .76 .34 .61

Note. Boldface values represent significant correlation coefficients (p < .05). IPV ¼ intimate partner violence; RAPI ¼ Rutgers Alcohol Problem Index; PTS ¼ Posttraumatic Stress; Gender: female ¼ 0, male ¼ 1.

Table 2. Total and Direct Effects of Child Abuse/Neglect and Witnessing IPV on Dating Violence, Problem Drinking, Marijuana Use, Dating Violence þ Problem Drinking, and Dating Violence þ Marijuana Use. Outcome (Y)

Total Effect (c) Coefficient

SE

Direct Effect (c’) p

Coefficient

SE

p

.0025 .0024 .0111 .0016 .0017

.692 .983 .099 .441 .120

.0185 .0182 .0813 .0874 .0825

.059 .155 .886 .616 .842

Child Abuse/Neglect (X) Dating violence Problem drinking Marijuana use Dating violence þ problem drinking Dating violence þ marijuana use

.0003 .0027 .0091 .0014 .0073

.0023 .0024 .0093 .0096 .0095

.908 .254 .329 .881 .443

.0010 .0001 .0183 .0090 .0181

Witnessing IPV (X) Dating violence Problem drinking Marijuana use Dating violence þ problem drinking Dating violence þ marijuana use

.0471 .0539 .0967 .0471 .1077

.0172 .0176 .0718 .0729 .0721

.007 .003 .178 .518 .134

.0353 .0261 .0117 .0439 .0164

Note. IPV ¼ intimate partner violence; SE ¼ standard error. Boldface values represent significant effects (p < .01). All models include gender as a covariate.

Multiple Mediation Models Total and direct effects. Total and direct effects of the two maltreatment variables on all five outcomes are listed in Table 2. Controlling for gender, there were only two significant total effects for maltreatment on the dating violence/substance use outcomes. Specifically, the total effects of witnessing IPV on

both dating violence and problem drinking were significant. All other total effects were nonsignificant. In addition, there was no significant direct effect (c’) of child abuse/neglect on perpetration of dating violence, alcohol problems, or marijuana use. The direct effects of witnessing IPV on dating violence, alcohol problems, and marijuana use were also nonsignificant.


8

Child Maltreatment

Table 3. Direct Effects of Child Abuse/Neglect and Witnessing IPV on Trauma Symptoms. Direct Effect (ai) Independent Variable Mediator Child abuse/neglect

Witnessing IPV

PTS Anxiety Depression Anger Dissociation Sexual concerns PTS Anxiety Depression Anger Dissociation Sexual concerns

Coefficient

SE

p

.122 .096 .071 .057 .083 .039 .830 .579 .481 .727 .722 .595

.028 .021 .023 .026 .027 .022 .216 .170 .177 .191 .204 .166