Bidirectional Relations Between Dating Violence Victimization and Substance Use in a Diverse Sample of Early Adolescents

Abstract

Substance use and dating violence victimization are common in adolescence and represent significant public health concerns. Although theoretical accounts suggest a bidirectional association between substance use and victimization within dating relationships, this has not been tested during early adolescence. Thus, the current study examined bidirectional associations between physical and psychological dating violence victimization and substance use across 6 months among an ethnically diverse sample of early adolescents. Sex was also examined as a moderator. Participants included two cohorts of sixth graders from 37 schools who were in dating relationships in the last 3 months at Wave 1, in the fall of sixth grade, and 6 months later at Wave 2, in the spring of sixth grade (n = 2,022; 43% female; 55% Black, 17% Latino/a, 16% White, 9% as multiracial, and 3% as another race/ethnicity). Students reported on the frequency of dating violence in the past 3 months and substance use in the past 30 days. Multilevel models, with students at Level 1 and classes (i.e., clusters of students in the same cohort at the same school; n = 74) at Level 2, tested hypotheses that positive reciprocal relations between physical and psychological dating violence victimization and substance use would be found over time, and that relations would be stronger for girls than boys. Sex, race/ethnicity, and family structure variables were included as Level 1 covariates; intervention condition and neighborhood concentrated disadvantage were included as Level 2 covariates. Results showed that higher levels of physical dating violence victimization at Wave 1 predicted increased substance use at Wave 2. Higher levels of substance use at Wave 1 predicted increased physical and psychological dating violence victimization at Wave 2. Findings highlight the importance of prevention efforts for dating violence and substance use early in adolescence.

Keywords

dating violence domestic violence alcohol and drugs youth violence

Both substance use and dating violence victimization are common in adolescence and represent significant public health concerns. Theoretical accounts suggest a bidirectional association between substance use and victimization within dating relationships (Temple & Freeman, 2011). Although empirical studies provide evidence for these longitudinal bidirectional associations starting in early adolescence with follow-up time points in late adolescence and early adulthood (Ackard, Eisenberg, & Neumark-Sztainer, 2007; Exner-Cortens, Eckenrode, & Rothman, 2013; Foshee, Benefield, Ennett, Bauman, & Suchindran, 2004; Foshee, Reyes, Gottfredson, Chang, & Ennett, 2013; O’Donnell et al., 2006; Roberts, Klein, & Fisher, 2003), they have not examined these paths within the time frame of early adolescence (i.e., ages 10-14) and have largely focused on physical forms of dating violence. The current study addressed these limitations by examining bidirectional associations between physical and psychological dating violence victimization and substance use during early adolescence.

Better understanding these longitudinal associations has important implications for prevention efforts targeting early adolescent problem behaviors. For example, Grant and Dawson (1998) found that youth who begin using substances before the age of 14 are at greater risk of developing substance use problems in adulthood. Thus, identifying risk factors for substance use, such as dating violence victimization, early in adolescence may be critical to preventing later problematic use. There are also few prevention programs focused on healthy dating relationships throughout middle school (see “Shifting Boundaries”: Taylor, Stein, Mumford, & Woods, 2013, and “The Fourth R”: Wolfe et al., 2009, for exceptions). Thus, the current study may underscore the need for such programs by identifying risk factors for and negative consequences of such victimization experiences during early adolescence.

Adolescent Dating Violence

Dating is an important part of adolescent development, with approximately 25% of 12-year-old youth reporting involvement in a dating relationship and 70% of 18-year-old youth reporting involvement (Connolly & McIsaac, 2009). Early adolescent dating relationships often differ from those in later adolescence based on their shorter duration (i.e., rarely lasting more than 4 months), greater fluidity, and the fact that they are highly entrenched in peer groups, which inform dating norms, attitudes, and behaviors (Connolly & McIsaac, 2009). Although dating can positively affect youth’s socioemotional competence, they may also place youth at risk for dating violence victimization. Two commonly studied forms of dating violence are physical and psychological. Physical dating violence includes physically aggressive behaviors, such as kicking, pushing, and scratching. In contrast, psychological dating violence includes nonphysical acts that may be verbal or nonverbal and are intended to cause psychological or emotional harm (e.g., saying things to hurt a dating partner’s feelings) and/or to be controlling in nature (e.g., damaging a dating partner’s belongings; Breiding, Basile, Smith, Black, & Mahendra, 2015). After compiling prevalence rates across studies, Goncy, Sullivan, Farrell, Mehari, and Garthe (2017) reported rates of dating violence victimization ranging from 20% to 53% for early adolescents (Arriaga & Foshee, 2004; Lormand et al., 2013; Niolon et al., 2015; Orpinas, Hsieh, Song, Holland, & Nahapetyan, 2013; Simon, Miller, Gorman-Smith, Orpinas, & Sullivan, 2010). These findings suggest that early adolescents experienced high frequencies of dating violence victimization, which may be due in part to transient nature of early adolescent dating relationships, their enmeshment within peer groups, and the novelty and challenges at this developmental stage in coping with the complexities of dating.

Association Between Dating Violence Victimization and Substance Use

Several theories suggest bidirectional associations between dating violence victimization and substance use. According to routine activities and lifestyles approaches, youth who engage in substance use may be at increased risk for dating violence victimization because their routine activities place them in risky contexts (Gover, 2004). Researchers also assert that the pharmacological effects of substance use may also place youth at risk for victimization (Temple & Freeman, 2011). In contrast, dating violence victimization experiences are stressful and may create negative emotions and trauma symptoms (Kaukinen, 2014). Stress and coping theory suggests that dating violence victimization may lead to substance use as a way to cope with the negative emotions generated by these victimization experiences (Temple & Freeman, 2011). Other researchers posit that substance use may be a form of self-medication used to alleviate negative emotions and trauma symptoms stemming from victimization experiences in dating contexts (Saunders, Kilpatrick, Hanson, Resnick, & Walker, 1999).

Although longitudinal paths between dating violence victimization and substance use have been assessed among early adolescents, several limitations exist in this literature. Primarily, no studies to our knowledge have examined bidirectional associations between these constructs within the time frame of early adolescence. A couple of studies showed that substance use in early adolescence predicted dating violence victimization in young adulthood (Foshee et al., 2004; O’Donnell et al., 2006). The reverse relation was found in that dating violence victimization among adolescents led to increased substance use 5 years later (Ackard et al., 2007; Exner-Cortens et al., 2013). Other studies that included combined samples of early- and mid-adolescents (Foshee et al., 2013) or participants ranging in age from early adolescence to emerging adulthood (Roberts et al., 2003) also demonstrated that dating violence victimization resulted in higher levels of substance use over time. A further limitation within this literature is the predominant focus on physical forms of dating violence victimization. Although a few studies examined psychological victimization (e.g., Exner-Cortens et al., 2013; Foshee et al., 2013), many addressed only physical victimization. Last, few studies controlled for the potential influence of dating violence perpetration in data analyses. Research showing that dating violence perpetration is positively related to both dating violence victimization (e.g., O’Keefe, 1997) and substance use (e.g., Swahn, Bossarte, & Sullivent, 2008) suggests the importance of controlling for perpetration when examining these paths.

It is important to consider the extent to which sex moderates the strength of associations between dating violence victimization and substance use. Research suggests that dating violence victimization may be a stronger predictor of substance use for girls as they are more likely than boys to develop depressive symptoms in response to stress, which may then lead to maladaptive coping strategies such as substance use (Agnew, 2006; Broidy & Agnew, 1997). In addition, sex differences in physical size and strength may increase girls’ vulnerability to dating victimization when using substances. Finally, most studies that address sex differences have compared separate samples of male and female adolescents and reported inconsistent findings (e.g., Ackard et al., 2007; Exner-Cortens et al., 2013; O’Donnell et al., 2006; Roberts et al., 2003). From an analytic perspective, the examination of sex as a moderator is necessary to make conclusions about differences in the strength of associations between the study variables for boys and girls. Thus, based on research supporting stronger bidirectional associations between dating victimization and substance use for girls and the need to more rigorously test these associations over time, we examined the moderating role of sex on these associations.

The Present Study

Our study extends the literature to date by examining bidirectional associations between physical and psychological forms of dating violence victimization and substance use, and also the moderating role of sex. Given research showing that the rates of dating violence victimization vary by sex, race/ethnicity, family structure, and neighborhood disadvantage (e.g., Spriggs, Halpern, & Martin, 2009), we controlled for these variables in all models. We also controlled for intervention condition. Based on theory and empirical research, we hypothesized that physical and psychological dating violence victimization at Wave 1, in the fall of sixth grade, would predict increased substance use at Wave 2, in the spring of sixth grade. We also anticipated stronger associations between both forms of victimization and substance use for girls than for boys. Second, we expected that substance use at Wave 1 would predict increased physical and psychological dating violence victimization at Wave 2, and that the path between substance use and physical victimization would be stronger for girls as compared with boys.

Method

Setting

The current study used data from the Multisite Violence Prevention Project (MVPP; 2004). MVPP evaluated the impact of three violence prevention programs for sixth graders (i.e., a universal intervention, a targeted intervention, and a combined universal and targeted intervention). The universal intervention consisted of the Guiding Responsibility and Expectations for Adolescents Today and Tomorrow (GREAT) student curriculum and the GREAT Teacher Program (Orpinas, Horne, & Multisite Violence Prevention Project, 2004). The 20-session student curriculum was designed to educate students on how to avoid dangerous situations, ignore provocation, ask for help, talk things through, defuse situations, and help their peers. The Teacher Program consisted of a 12-hr workshop and 10 consultation sessions designed to increase awareness of and ability to reduce aggressive behavior and improve classroom management. The targeted intervention was a 15-week family program for sixth graders who were identified by their teachers as demonstrating high levels of aggression and peer influence (Smith et al., 2004). None of these interventions specifically targeted dating violence victimization or substance use prevention.

Participating students were from 37 schools across four geographical sites (Durham, North Carolina; Richmond, Virginia; Chicago, Illinois; and Northeast Georgia). Between 42% and 96% of students at the participating schools qualified for the federally subsidized lunch program, and the average poverty rate and youth arrest rate in the selected school districts (28% and 63/100,000, respectively) were higher than the national average (16% and 43/100,000, respectively; Henry, Farrell, & MVPP, 2004).

Participants

The project included two cohorts of sixth graders. With the exception of Chicago schools, the sample consisted of a random sample of approximately 80 students in sixth grade from each school. At the Chicago schools, all sixth graders were recruited. Eligibility for participation was limited to students who were not in self-contained special education classrooms. Of the 7,343 eligible students, active parental permission and student assent were obtained from 5,811. After Wave 1, additional waves of data were only collected from students who remained in their original school. At least one wave of data were available for 5,465 students (97% of those consented and eligible), 49% were male; 52% were Black, 21% were Latino/a, 20% were White, and 7% were another race/ethnicity. Only youth who reported being involved in a dating relationship within the last 3 months at Wave 1 (in the fall of sixth grade) and Wave 2 (in the spring of sixth grade; N = 2,022; 43% female) were included in the final sample. For this final sample, more than half (55%) identified themselves as Black, 17% as Latino/a, 16% as White, 9% as multiracial, and 3% as another race/ethnicity. Approximately 43% reported living in a two-parent household, 28% lived in a single-parent household, 12% lived with a parent and a stepparent or parent’s significant other, 9% lived with a single parent and additional family members, 5% lived with an adult relative, and 3% reported another family structure.

Procedure

The Institutional Review Boards at the four participating universities and the Centers for Disease Control and Prevention (CDC) approved all study procedures. The MVPP used a cluster-randomized design, first recruiting eight to 12 schools per geographic site, then randomly assigning each school to one of four violence prevention program conditions (i.e., universal intervention, targeted intervention, combined universal and targeted intervention, or no-intervention control) so that there were equal numbers of schools in each condition. Participating schools were selected based on their willingness to participate. In Georgia, North Carolina, and Virginia, these included mostly large middle schools (i.e., with 200 to 400 sixth graders per school), two to three of which were assigned to each of the four conditions. Schools in Chicago were comparably smaller; thus, schools that met certain criteria with regard to grades served, enrollment, percentage of low-income students, residence of student, and distance from the university were approached to participate. In Chicago, three schools with 61 to 120 sixth graders per school were assigned to each condition. Additional details on the school recruitment can be found in Henry et al. (2004).

Two cohorts of youth were recruited, the first during the fall of 2001 and the second during the fall of 2002. Youth in Cohort 1 who repeated the sixth grade were not eligible for recruitment in the fall of 2002. Consent and assent forms were sent home with students and at three of the sites, students who returned their forms received a US$5 gift card, despite whether they agreed to participate or not. Survey administration took place at school with small groups of students. Students completed the survey using computer-assisted self-interviewing with audio (Audio-CASI), which allowed them to read the questions on the computer while listening to them through headphones, and then enter their responses using the keyboard. Students received a US$5 gift card for participation, when permitted by schools.

Measures

Demographic characteristics

Students reported their sex, race/ethnicity, and family structure. To assess race/ethnicity, students first indicated whether they were Hispanic or Latino; then selected as many of the following options that applied: White, Black or African American, American Indian or Alaska Native, Asian Indian, Other Asian, or some Other race. Based on students’ responses, a single race/ethnicity variable was created that represented the following race/ethnicities: White, Black or African American, American Indian or Alaska Native, Asian Indian, Other Asian, Hispanic, Other race, non-Hispanic, and multiracial. To assess family structure, students indicated who lived in their home and could choose as many of the options that applied (e.g., mother, father, stepmother, grandmother). A variable was created to reflect the following typical family structures: two-parent family, multigenerational single-parent family, parent with either stepparent or parent’s significant other, single parent with or without other adults, foster family, adult relative with neither parent, or other.

Dating violence

Dating violence was measured using an adapted version of the Dating Violence and Norms measure (Foshee et al., 1996). The adapted measure included two additional items (i.e., “Stomped out of the room or house during a disagreement”; “Punched or hit you/him or her with something that could hurt”), three items with minor wording changes, and a different response scale for the Physical Violence scales (changed to be consistent with the Psychological Violence scales and described below). If youth indicated that they had a boyfriend/girlfriend in the last 3 months, they were then asked whether or not they had “done any of the following things to a boyfriend/girlfriend” (perpetration) and whether “a boyfriend/girlfriend has done the following things to you” (victimization) in the last 3 months (not counting instances of self-defense). To measure victimization, the nine-item Physical Victimization (e.g., “How many times has a boyfriend or girlfriend punched or hit you with something that could hurt?”) and the four-item Psychological Victimization (e.g., “How many times has a boyfriend or girlfriend said things to hurt your feelings on purpose?”) scales were used. The Physical and Psychological Perpetration scales were also used and their items mirror the Victimization items. Youth indicated the frequency of these behaviors, using a 4-point response scale, where 0 = never, 1 = 1 to 3 times, 2 = 4 to 9 times, 3 = 10 or more times. Higher scores indicate higher levels of victimization and perpetration. At Wave 1, Cronbach’s alphas for the Physical and Psychological Victimization and Perpetration scales ranged from .72 to .92. At Wave 2, Cronbach’s alphas for the Physical and Psychological Victimization scales were .91 and .73, respectively.

Substance use

Substance use was measured using the Problem Behavior Frequency Scale (PBFS; Farrell, Kung, White, & Valois, 2000; Miller-Johnson, Sullivan, Simon, & MVPP, 2004), a self-report measure comprised of seven scales assessing the frequency of problem behaviors (i.e., Physical Aggression, Nonphysical Aggression, Relational Aggression, Delinquent Behavior, Substance Use, Overt Victimization, and Relational Victimization). The Substance Use scale assessed the frequency of alcohol, tobacco, and marijuana use (e.g., “How many times have you drunk beer [more than a sip or taste])?” Students reported the past 30-day frequency of these behaviors using a 6-point response scale, where 0 = never, 1 = 1 to 2 times, 2 = 3 to 5 times, 3 = 6 to 9 times, 4 = 10 to 19 times, and 5 = 20 or more times. Cronbach’s alphas for the scale at Wave 1 and Wave 2 were .85 and .88, respectively.

Concentrated disadvantage

A neighborhood concentrated disadvantage variable was created using an approach by Sampson, Raudenbush, and Earls (1997), whereby factor scores were calculated for census tracts within each school district from 2000 U.S. Census data. The variable represents a weighted composite of poverty rate, per capita income, percentage of female-headed households, percentage receiving public assistance, percentage renting, percentage with less than a high school diploma, and percentage unemployed.

Data Analysis

Data from MVPP were thoroughly inspected and cleaned to identify and address data entry errors, outliers, and suspect response patterns. Analyses were conducted in Mplus version 7 (Muthén & Muthén, 2012) and used a multilevel approach. Student-level variables were represented at Level 1 and class-level variables representing scores for students in the same cohort at the same school were represented at Level 2 (see Figure 1). There were a total of 74 clusters, with an average of 27 students per cluster. Full-information maximum likelihood (FIML) was used to handle missing data on the outcome variable (i.e., Wave 2 substance use) due to incomplete responses. This method estimates missing data based on the variables that are present (Muthén & Muthén, 2012). To account for the non-normality and nonindependence of study variables, models were calculated with the maximum likelihood estimation with robust standard errors (MLR) estimator.

Figure 1.

Multilevel model depicting regression of Wave 2 substance use on (a) Wave 1 dating violence victimization, sex, and (b) class-level covariates.

Preliminary analyses involved a content analysis to confirm the placement of items into scales. Item characteristic curves were examined for each item in the PBFS Substance Use scale at Waves 1 and 2 for youth in the current study. Scores on each item were treated as ordered categorical variables. Although youth rated each item on a 6-point scale, very few youth endorsed the last two response categories, and an examination of the item characteristic curves supported collapsing the response category 20 or more times (0.5%-2.7% of participants chose this response across items at Waves 1 and 2) with the option 10 to 19 times (0.1%-1.1% of participants chose this response across items at Waves 1 and 2).

Multilevel analyses tested hypotheses that substance use varied as a function of physical and psychological dating violence victimization and vice versa, and that these relations were moderated by sex. Sex, race/ethnicity, and family structure variables were included as Level 1 covariates. Family structure was dummy-coded into two variables: single-parent and other family structure, with two-parent family structure as the reference group. Race/ethnicity was dummy-coded into the following three variables: White, Latino/a, and Other race/ethnicity, with Black as the reference group. Level 2 covariates included intervention condition and concentrated disadvantage. Intervention condition was dummy-coded into three variables: universal intervention, targeted intervention, and combined intervention, with the control condition as the reference group. Interaction terms were defined in Mplus by creating variables that represented the product of each variable. Except for the dummy-coded variables, all predictor variables at Levels 1 and 2 were grand mean centered to facilitate interpretation.

A hierarchical approach was utilized, such that sets of variables were entered step-by-step. First, (a) we tested a baseline model with random intercepts, (b) next, Level 2 covariates (i.e., intervention condition and concentrated disadvantage) were added, followed by (c) Level 1 demographic variables (i.e., family structure, sex, and race/ethnicity), (d) Wave 1 levels of outcome(s) and the predictor(s), and last (e) dating violence perpetration. Interactions between predictors and sex were added to Model 4 one by one. Parameter estimates and standard errors were examined within each model to determine the significance of individual variables. To determine the proportion of Level 1 and Level 2 variance explained by each set of variables the difference in the amount of Level 1 ( $R_{1}^{2}$ ) and Level 2 ( $R_{2}^{2}$ ) residual variance was calculated as a proportion of the total error variance (Snijders & Bosker, 1994). This results in a statistic that is comparable with the multiple R².

Results

Descriptive Statistics

The percentage of youth who reported experiencing each type of dating violence one or more times in the past 3 months at Waves 1 and 2 (fall and spring of sixth grade) is shown in Table 1. Having a partner do something to evoke jealously was the most common type of psychological victimization endorsed at both waves across sex. Having a dating partner stomp out of the room or house during a disagreement was the most common physically violent act experienced by girls at Waves 1 and 2, whereas being scratched was the most common form of physical victimization reported by boys across waves. For the total sample, more severe types of physical victimization, such as being slammed or held against a wall, having something thrown at them that could hurt, and being punched or hit with something that could hurt were less commonly experienced. The percentage of youth who reported having used each substance one or more times in the past 30 days at Wave 1 and Wave 2 is reported in Table 2. For the total sample, the percentage of youth who reported engaging in each type of substance use, except cigarette use, increased from Wave 1 to 2. Drinking wine or wine coolers was the most common type of substance use reported across Waves 1 and 2 for boys and girls, and a large percentage of boys and girls also reported having drunk beer. For the total sample, the percentage of youth who had engaged in marijuana use at Wave 1 was relatively low, but more than doubled by Wave 2.

Table 1.

Percentage of Sixth Graders Who Reported Experiencing Specific Forms of Psychological and Physical Dating Violence Victimization One or More Times in the Past 3 Months at Wave 1, in the Fall of Sixth Grade, and at Wave 2, in the Spring of Sixth Grade.

Type of Dating Violence Victimization	Wave 1			Wave 2
Type of Dating Violence Victimization	Total Sample(%)	Girls(%)	Boys(%)	Total Sample(%)	Girls(%)	Boys(%)
Psychological victimization
Damaged something that belonged to you	19.7	13.3	24.6	16.5	13.7	18.6
Said things to hurt your feelings on purpose	27.5	29.0	26.3	26.1	29.5	23.6
Would not let you do things with other people	22.2	18.2	25.2	21.8	17.6	24.9
Did something just to make you jealous	36.0	38.7	34.0	34.2	40.1	29.8
Physical victimization
Threatened to hit or throw something at you	20.4	15.6	24.0	20.1	14.4	24.4
Scratched you	30.6	17.3	40.6	27.0	16.7	34.7
Slapped you	18.3	7.6	26.3	19.0	8.2	27.1
Slammed you or held you against a wall	11.1	7.9	13.6	9.9	8.4	11.1
Kicked you	20.2	10.0	27.8	16.9	7.2	24.1
Pushed or shoved you	21.0	14.2	26.1	20.8	15.1	25.1
Stomped out of the room or house during a disagreement	20.6	19.2	21.7	20.1	18.6	21.3
Threw something at you that could hurt	14.8	7.7	20.2	11.2	5.9	15.2
Punched or hit you with something that could hurt	12.6	6.5	17.2	10.5	6.7	13.3

Note. At Wave 1, ns ranged from 2,016 to 2,018. At Wave 2, ns ranged from 2,015 to 2,018.

Table 2.

Percentage of Sixth Graders Who Reported Using Specific Substances One or More Times in the Past 30 Days at Wave 1, in the Fall of Sixth Grade, and at Wave 2, in the Spring of Sixth Grade.

Type of Substance Use	Wave 1			Wave 2
Type of Substance Use	Total Sample(%)	Girls(%)	Boys(%)	Total Sample(%)	Girls(%)	Boys(%)
Drunk beer (more than a sip or taste)	18.4	16.7	19.6	25.5	25.2	25.9
Drunk wine or wine coolers (more than a sip or taste)	21.9	21.0	22.5	26.9	25.6	28.7
Been drunk	6.8	5.5	7.7	18.3	18.0	18.9
Drunk liquor (like whiskey or gin)	8.6	7.6	9.3	12.2	13.1	11.1
Used marijuana (pot, hash, reefer)	4.1	3.0	4.9	13.3	13.2	13.4
Smoked cigarettes	11.5	9.7	12.8	8.3	9.3	7.1

Note. At Wave 1, ns ranged from 2,008 to 2,014. At Wave 2, ns ranged from 2,014 to 2,019.

Means, standard deviations, and correlations among the uncentered Level 1 variables were calculated in Mplus (see Table 3). Wave 1 physical and psychological dating violence victimization were significantly correlated with each other (r = .76), both forms of perpetration at Wave 1 (rs = .49-.58), and Wave 2 substance use (rs = .19 and .14, respectively). Wave 1 substance use was significantly correlated with physical and psychological perpetration at Wave 1 (rs = .34) and Wave 2 physical and psychological victimization (rs = .22 and .19, respectively).

Table 3.

Means, Standard Deviations, and Intercorrelations Among Level 1 Study Variables (N = 2,022).

Variable	1	2	3	4	5	6	7	8	M	SD
1. W1 psychological victimization	—								.35	.51
2. W1 physical victimization	.76***	—							.27	.49
3. W2 psychological victimization	.34***	.31***	—						.33	.50
4. W2 physical victimization	.31***	.38***	.75***	—					.24	.48
5. W1 psychological perpetration	.58***	.49***	.29***	.26***	—				.24	.45
6. W1 physical perpetration	.50***	.52***	.27***	.27***	.81***	—			.18	.43
7. W1 substance use	.27***	.29***	.19***	.22***	.34***	.34***	—		.19	.47
8. W2 substance use	.14***	.19***	.27***	.26***	.20***	.20***	.53***	—	.32	.65

p < .05. **p < .01. ***p < .001.

Multilevel Analyses

Models predicting Wave 2 substance use

These models tested hypotheses that Wave 1 physical and psychological dating violence victimization would predict changes in substance use and that these associations would be moderated by sex (see Table 4). Model 1, the baseline random intercept model, indicated significant variance in Wave 2 substance use at Level 1 (σ² = .42, p ≤ .001), but not Level 2 (τ = .01, p = .053). The intraclass correlation (ICC) was .02. The addition of Level 2 covariates (i.e., intervention conditions and concentrated disadvantage) accounted for 0% of the modeled variance in Level 1 and 2 scores (see Model 2). Neither concentrated disadvantage, nor any of the intervention conditions were significant predictors of substance use. The addition of demographic characteristics at Level 1 (i.e., family structure, sex, and race/ethnicity) accounted for an additional 29% of the modeled variance in Level 1 scores and 44% in Level 2 scores (see Model 3). Youth who were Latino reported higher levels of Wave 2 substance use than youth who were Black (β = .10, p = .020).

Table 4.

Unstandardized Parameter Estimates for Multilevel Regression of Substance Use on Dating Violence Victimization (N = 2,022).

Variable	Model 1	Model 2	Model 3	Model 4	Model 5
Level 1 (student)
Intercept	.317*** (.018)	.304*** (.029)	.301*** (.032)	.314*** (.031)	.313*** (.031)
Single-parent family			.028 (.024)	.027 (.024)	.026 (.024)
Other family structure			−.004 (.027)	−.004 (.027)	−.007 (.026)
White			−.011 (.031)	−.007 (.031)	−.005 (.030)
Latino/a			.099* (.043)	.098* (.043)	.099* (.043)
Other race/ethnicity			−.034 (.040)	−.032 (.040)	−.032 (.040)
Male sex			−.028 (.025)	−.049* (.025)	−.047 (.025)
W1 substance use				.734*** (.068)	.730*** (.066)
W1 physical victimization				.126* (.059)	.128* (.062)
W1 psychological victimization				−.097 (.052)	−.109* (.053)
W1 physical perpetration					−.022 (.061)
W1 psychological perpetration					.043 (.062)
Level 2 (class)
Universal intervention condition		.016 (.053)	.018 (.041)	.020 (.042)	.020 (.042)
Targeted intervention condition		.035 (.041)	.004 (.028)	.001 (.027)	.001 (.027)
Combined intervention condition		−.005 (.045)	.028 (.030)	.027 (.030)	.027 (.030)
Concentrated disadvantage		.027 (.019)	.002 (.014)	.002 (.014)	.002 (.014)
Variance estimates
Within level residual variance	.421*** (.034)	.421*** (.034)	.304*** (.028)	.302*** (.027)	.302*** (.027)
Between level residual variance	.008 (.004)	.008 (.004)	.002 (.002)	.002 (.002)	.002 (.002)
$R_{1}^{2}$		.000	.287	.291	.291
$R_{2}^{2}$		.000	.439	.442	.442

Note. Standard errors are in parentheses. All models were fully saturated.

p < .05. **p < .01. ***p < .001.

In Model 4, the addition of Wave 1 substance use, physical dating violence victimization, and psychological dating violence victimization accounted for an additional 1% of the modeled variance in Level 1 scores and no additional variance in Level 2 scores. Wave 1 substance use was significantly associated with Wave 2 substance use (β = .73, p < .001). Physical dating violence victimization predicted changes in substance use from Wave 1 to Wave 2, such that higher levels of physical dating violence victimization were associated with greater change in substance use (β = .13, p = .031).

Interactions between sex and Wave 1 physical and psychological dating violence victimization were added one by one. The interactions between sex and physical dating violence victimization (β = −.06, p = .449) and between sex and psychological dating violence victimization (β = −.07, p = .272) did not predict changes in substance use. Thus, paths between Wave 1 dating violence victimization and Wave 2 substance use did not significantly vary for boys and girls.

In Model 5, the addition of Wave 1 physical and psychological dating violence perpetration did not account for any additional modeled variance in Level 1 or 2 scores. Neither form of perpetration significantly predicted substance use. When controlling for dating violence perpetration, physical dating violence victimization remained a significant predictor of change in substance use (β = .13, p = .039).

Models predicting Wave 2 physical dating violence victimization

These models tested hypotheses that Wave 1 substance use would predict changes in physical dating violence victimization, and that these relations would be moderated by sex (see Table 5). Model 1, the baseline random intercept model, indicated significant variance in Wave 2 physical dating violence victimization at Level 1 (σ² = .26, p ≤ .001) and Level 2 (τ = .00, p = .007). The ICC was .02. The addition of Level 2 covariates (i.e., intervention condition variables and concentrated disadvantage) accounted for 1% of the modeled variance in Level 1 scores and 25% in Level 2 scores (see Model 2). All intervention conditions were significantly associated with Wave 2 physical dating violence, such that clusters of youth who received the universal (β = .10, p = .001), targeted (β = .07, p = .036), or combined interventions (β = .07, p = .042) reported higher cluster averages of Wave 2 physical dating violence victimization as compared with the control group. Concentrated disadvantage was also significantly associated with Wave 2 physical victimization, such that higher levels of neighborhood concentrated disadvantage were associated with higher cluster averages of physical victimization (β = .04, p < .001). The addition of Level 1 demographic characteristics accounted for an additional 4% of the modeled variance in Level 1 scores and 11% in Level 2 scores (see Model 3). Youth who were White reported lower levels of Wave 2 physical victimization than youth who were Black (β = −.10, p = .002). In addition, boys reported higher levels of Wave 2 physical victimization as compared with girls (β = .19, p < .001).

Table 5.

Unstandardized Parameter Estimates for Multilevel Regression of Physical Victimization on Substance Use (N = 2,022).

Variable	Model 1	Model 2	Model 3	Model 4	Model 5
Level 1 (student)
Intercept	.242*** (.013)	.182*** (.021)	.105** (.031)	.168*** (.028)	.162*** (.026)
Single-parent family			.002 (.024)	−.015 (.023)	−.016 (.023)
Other family structure			−.006 (.028)	−.040 (.024)	−.044 (.023)
White			−.095** (.031)	−.054* (.027)	−.048 (.026)
Latino/a			−.045 (.028)	−.071** (.024)	−.066** (.024)
Other race/ethnicity			−.007 (.032)	−.014 (.031)	−.013 (.031)
Male sex			.193*** (.024)	.122*** (.023)	.133*** (.022)
W1 physical victimization				.257*** (.053)	.228*** (.058)
W1 psychological victimization				.061 (.043)	.038 (.046)
W1 substance use				.117*** (.032)	.097** (.032)
W1 physical perpetration					.077 (.065)
W1 psychological perpetration					.034 (.056)
Level 2 (class)
Universal intervention condition		.098** (.029)	.078** (.026)	.074** (.023)	.073** (.023)
Targeted intervention condition		.067* (.032)	.051 (.031)	.032 (.028)	.033 (.028)
Combined intervention condition		.069* (.034)	.061 (.031)	.047 (.027)	.047 (.027)
Concentrated disadvantage		.044*** (.011)	.031** (.011)	.027** (.011)	.026* (.011)
Variance estimates
Within level residual variance	.225*** (.017)	.225*** (.017)	.216*** (.016)	.187*** (.014)	.185*** (.014)
Between level residual variance	.004** (.001)	.001 (.001)	.000 (.001)	.000 (.001)	.000 (.001)
$R_{1}^{2}$		.013	.057	.183	.192
$R_{2}^{2}$		.245	.354	.441	.447

Note. Standard errors are in parentheses. All models were fully saturated.

p < .05. **p < .01. ***p < .001.

In Model 4, the addition of Wave 1 physical and psychological victimization and Wave 1 substance use accounted for an additional 13% of the modeled variance in Level 1 scores and 9% additional variance in Level 2 scores. Wave 1 physical victimization was significantly associated with Wave 2 physical victimization (β = .26, p < .001). Wave 1 substance use significantly predicted change in physical victimization from Wave 1 to Wave 2, such that higher levels of substance use were associated with greater change in physical victimization over time (β = .13, p = .031). This path was not moderated by sex (β = −.03, p = .653), indicating that the association was not significantly different for boys and girls.

In Model 5, the addition of Wave 1 physical and psychological dating violence perpetration accounted for 1% additional modeled variance in Level 1 and 2 scores. Neither form of perpetration significantly predicted physical dating violence victimization; however, when controlling for perpetration, substance use remained a significant predictor of change in physical dating violence victimization (β = .10, p = .003).

Models predicting Wave 2 psychological dating violence victimization

These models tested hypotheses that Wave 1 substance use would predict changes in psychological dating violence victimization and that these relations would be moderated by sex (see Table 6). Model 1, the baseline random intercept model, indicated significant variance in Wave 2 psychological victimization at Level 1 (σ² = .25, p ≤ .001), but not Level 2 (τ = .00, p = .308). The ICC was .01. The addition of intervention condition variables and concentrated disadvantage accounted for 1% of the modeled variance in Level 1 scores and 10% in Level 2 scores (see Model 2). Clusters of youth who received the universal (β = .07, p = .008) or combined interventions (β = .06, p = .046) reported higher cluster averages of Wave 2 psychological victimization as compared with the control group. In addition, higher levels of neighborhood concentrated disadvantage were associated with higher cluster averages of psychological victimization (β = .04, p < .001). The addition of Level 1 demographic characteristics in Model 3 accounted for an additional 1% of the modeled variance in Level 1 and 2 scores. Youth who were White reported lower levels of Wave 2 psychological victimization than youth who were Black (β = −.12, p < .001).

Table 6.

Unstandardized Parameter Estimates for Multilevel Regression of Psychological Victimization on Substance Use (N = 2,022).

Variable	Model 1	Model 2	Model 3	Model 4	Model 5
Level 1 (student)
Intercept	.334*** (.012)	.289*** (.017)	.288*** (.032)	.337*** (.029)	.329*** (.028)
Single-parent family			−.011 (.025)	−.027 (.025)	−.026 (.025)
Other family structure			.009 (.034)	−.040 (.028)	−.046 (.027)
White			−.124*** (.029)	−.081** (.025)	−.074** (.025)
Latino/a			.007 (.033)	−.016 (.028)	−.012 (.027)
Other race/ethnicity			.002 (.033)	−.006 (.031)	−.005 (.030)
Male sex			.051* (.025)	.006 (.024)	.020 (.024)
W1 physical victimization				.102* (.047)	.073 (.052)
W1 psychological victimization				.237*** (.044)	.198*** (.045)
W1 substance use				.099** (.032)	.074* (.035)
W1 physical perpetration					.056 (.064)
W1 psychological perpetration					.082 (.061)
Level 2 (class)
Universal intervention condition		.074** (.028)	.059* (.026)	.051* (.025)	.050* (.024)
Targeted intervention condition		.040 (.027)	.028 (.028)	.011 (.025)	.012 (.025)
Combined intervention condition		.062* (.031)	.058* (.029)	.042 (.028)	.042 (.028)
Concentrated disadvantage		.040*** (.011)	.024* (.011)	.021 (.012)	.018 (.012)
Variance estimates
Within level residual variance	.253*** (.016)	.252*** (.016)	.250*** (.015)	.219*** (.014)	.217*** (.014)
Between level residual variance	.001 (.001)	.000 (.002)	.000 (.002)	.000 (.002)	.000 (.002)
$R_{1}^{2}$		.008	.016	.138	.146
$R_{2}^{2}$		.101	.108	.219	.226

Note. Standard errors are in parentheses. All models were fully saturated.

p < .05. **p < .01. ***p < .001.

In Model 4, the addition of Wave 1 physical and psychological victimization and Wave 1 substance use accounted for an additional 12% of the modeled variance in Level 1 scores and 11% in Level 2 scores. Wave 1 physical (β = .10, p = .031) and psychological victimization (β = .24, p < .001) were significantly associated with Wave 2 psychological victimization. Wave 1 substance use significantly predicted change in psychological victimization from Wave 1 to Wave 2, such that higher levels of substance use were associated with greater change in psychological victimization over time (β = .10, p = .002). This association did not differ by sex (β = −.08, p = .154).

In Model 5, the addition of Wave 1 physical and psychological dating violence perpetration accounted for 1% additional modeled variance in Level 1 and 2 scores. Neither form of perpetration significantly predicted psychological victimization. When controlling for perpetration, substance use remained a significant predictor of change in psychological victimization (β = .07, p = .034).

Discussion

The current study assessed bidirectional associations between physical and psychological dating violence victimization and substance use in early adolescence, and the moderating effect of sex on these relations. We addressed several gaps in the literature by including physical and psychological forms of dating violence victimization, incorporating physical and psychological dating violence perpetration as covariates, and testing sex as a moderator of these associations. Key findings showed that substance use predicted increased physical and psychological dating violence, but only physical dating violence victimization led to subsequent increases in substance use across 6 months. These paths were invariant across sex.

As expected, higher frequencies of substance use predicted increased physical and psychological dating violence victimization over time. A potential explanation for this association may be the pharmacological effects of substance use, which impair a victim’s ability to interpret cues of aggressive behavior and limit their ability to defend themselves. Adolescent substance use also tends to occur in risky environments, such as those that may be more secluded without parental supervision and may place potential victims in the presence of aggressors. Prior research showed that substance use in early adolescence led to physical dating violence victimization later in development (e.g., Ackard et al., 2007; Exner-Cortens et al., 2013). However, the current study extended this research based on our finding that substance use predicted changes in both physical and psychological dating violence victimization during early adolescence.

Hypotheses regarding dating violence victimization as a predictor of substance use were partially supported. The finding that physical dating violence victimization predicted increased substance use was consistent with both stress and coping theory (Lazarus, 2006) and the self-medication model (Ostrowski, 2009). For adolescents who experience victimization in dating contexts, substance use may operate as a form of mental disengagement or affect regulation to reduce negative emotions and increase positive affect (Carver, Scheier, & Weintraub, 1989). The fact that physical, and not psychological, dating violence victimization predicted larger changes in substance use is inconsistent with some studies that have found psychological forms of victimization to be predictive of later substance use (Exner-Cortens et al., 2013; Foshee et al., 2013), but consistent with other evidence that exposure to psychological dating violence alone is less likely to result in negative mental health consequences (Choi, Weston, & Temple, 2017). Furthermore, this finding may be due in part to the greater severity of physical victimization in that it can result in physical injury and associated trauma symptoms (Kaukinen, 2014). In addition, physical dating violence victimization may be part of a syndrome of problem behaviors. More specifically, Orpinas, Nahapetyan, and Truszczynski (2017) highlighted problem behavior theory (Donovan, Jessor, & Costa, 1988; Jessor et al., 2003) as a model to understand the covariation of problem behaviors such as aggression (in peer and dating contexts), substance use, and risky sexual behavior during adolescence. More specifically, youth who are involved in one type of problem behavior are more likely to be involved in others, forming a syndrome of problem behaviors. Adolescents who experience higher frequencies of physical dating violence victimization spend time with partners who are physically aggressive and in turn may perpetrate this behavior themselves. Thus, they may be more likely to be exposed to situations and peer contexts where other problem behaviors such as substance use occur.

Contrary to expectations, bidirectional associations between both forms of dating violence victimization and substance use did not vary as a function of sex. The lack of sex differences in longitudinal relations between substance use and dating violence victimization is inconsistent with findings that substance use in early adolescence predicted victimization by intimate partners in young adulthood for girls only (O’Donnell et al., 2006). We proposed that bidirectional relations between physical and psychological dating victimization and substance use would be stronger for girls than boys. However, these hypotheses were not supported. Our findings indicated that during early adolescence, both boys and girls who experienced physical dating violence victimization were at risk for increased substance use, and conversely, that substance use was a risk factor for physical and psychological dating violence victimization. One reason may be related to the age of adolescents in the study and their stage of development in forming romantic relationships. As noted previously, boys reported higher frequencies of physical dating violence victimization than did girls in the current study, which may reflect boys and girls taking on less traditional roles in conflict resolution (Wincentak, Connolly, & Card, 2017). In this context, there may be no significant differences in the strength of relations between substance use and the vulnerability to dating violence victimization or the influence of physical dating victimization on subsequent substance use. However, replication is needed as our findings only partially aligned with our hypotheses and were inconsistent with some other studies in regard to the relation between psychological victimization and substance use and sex differences in relations between dating violence victimization and substance use.

Findings highlighted several differences in the prevalence of substance use and physical and psychological dating violence victimization. First, students who participated in the interventions (universal, targeted, and combined) showed higher average instances of physical and psychological dating victimization compared with students in the control group. Although none of the interventions specifically targeted dating violence, the interventions did address peer aggression. Therefore, students who received the intervention may have become more aware and sensitive to aggressive behavior and been more likely to report higher levels of dating victimization at Wave 2 as compared with students who did not receive any intervention. Similar findings have been demonstrated in evaluations of bullying prevention programs (e.g., Bowllan, 2011). Second, there were several differences in the frequencies of substance use and physical and psychological dating victimization by race/ethnicity. Youth who were White reported lower levels of both physical and psychological dating victimization as compared with youth who were Black. Some research suggests that the stressors faced by youth who live in urban communities with high rates of poverty and crime place them at increased risk for dating violence exposure (Foshee et al., 2008). In the current study, youth who were Black were more likely than youth who were White to reside in such urban settings (see Simon et al., 2009). However, differences in the characteristics of the sites including the rural, suburban, and urban composition and distinct geographic locations prohibit direct comparisons. Another finding was that Latino youth reported higher levels of substance use than did youth who were Black. This finding is consistent with trends in data across time that showed Latino youth reported higher rates of alcohol and cigarette use than Black youth (Johnston, O’Malley, Miech, Bachman, & Schulenberg, 2017).

In addition to differences in prevalence rates for dating violence victimization found by race/ethnicity, findings also showed that boys reported higher rates of physical dating victimization as compared with girls. In a recent review, Wincentak et al. (2017) highlighted three points that help to explain this finding. First, this finding is consistent with other studies that showed girls are more likely to perpetrate dating aggression in the initial stages of forming romantic relationships in early- to mid-adolescence (e.g., O’Leary, Slep, Avery-Leaf, & Cascardi, 2008). Second, in transitioning from same-sex friendships into early romantic relationships, adolescents may adopt conflict resolution strategies that are more commonly used by the opposite sex, with girls being more likely to use physical aggression than boys (Rose & Rudolph, 2006). Last, these behavior patterns may reflect findings that early adolescents view girls’ aggression against boys as more socially acceptable than boys’ aggression against girls (e.g., Simon et al., 2010).

Limitations

Although the present study had several strengths, it is important to note study limitations. First, our sample was not a nationally representative sample (e.g., the current sample had a higher than average poverty rate); therefore, findings cannot be generalized to all adolescents in the United States. Second, the current study relied solely on adolescents’ reports of dating violence victimization and substance use. Thus, it is important to acknowledge the potential influence of method variance in relations among the study variables given the exclusive use of self-report data. However, it is also important to note that adolescents may be the best reporters of these types of behavior, and research indicates that youth are typically truthful in disclosing their involvement in problem behaviors like substance use (Oetting & Beauvais, 1990). In addition, although our dating violence measures assessed important forms of dating violence (i.e., physical and psychological), only 13 items were used to measure these constructs and we were not able to assess sexual dating violence based on concerns of district administrators at some schools from which our sample was drawn. Our measure of substance use also focused on cigarette, alcohol, and marijuana use and did not look at other forms of licit or illicit drug use. The current study is limited by its short-term longitudinal design assessing changes across a 6-month time period early in middle school. The nature of associations between dating victimization and substance use may evolve across adolescence due to changing dating relationships dynamics (e.g., increased capacities for managing relationship conflict and greater intimacy and commitment within relationships in later adolescence; McNaughton Reyes, Foshee, Bauer, & Ennett, 2012). In addition, because relationship status was based on the presence of a relationship in the past 3 months and additional questions about the relationship (e.g., duration) were not asked, it could not be determined whether youth in the current sample were in a relationship with the same person at both waves. Examining the influence of relationship factors such as longevity, age of dating partner, and level of conflict on relations between dating violence victimization and negative outcomes may be an important direction for future research. Thus, future studies should examine these associations within the developmental periods of early and later adolescence and should also examine the influence of relationship factors such as longevity, age of dating partner, and level of conflict on these associations.

Implications and Future Research

Current study findings have important implications for programs aimed at promoting positive youth development. Prevalence rates of specific forms of substance use and dating violence victimization indicate that a substantial number of early adolescents reported engaging in or experiencing these behaviors, highlighting the need to address substance use and dating violence prevention early in middle school. We found that substance use was both a risk factor and consequence of dating violence victimization. Thus, prevention efforts should target these two problem behaviors simultaneously during early adolescence to prevent long-term negative outcomes, such as problematic substance use (e.g., Grant & Dawson, 1998) and intimate partner violence (e.g., O’Donnell et al., 2006). Prevention programs should also address the unique dynamics of early adolescent dating relationships. From a developmental perspective, these partnerships tend to be short-lived, fluid and less intimate, and heavily influenced by peer groups (Connolly & McIsaac, 2009). Thus, prevention efforts may need to focus on appropriate ways to approach and initiate dating activities, and the adaptive and maladaptive roles that peers play and how they may affect dating partners’ behaviors.

Current findings suggest several directions for future research. First, there is a need for continued examination of how specific forms of dating violence victimization, such as cyber and relational dating violence victimization, are related to other types of problem behaviors. For instance, Zweig, Dank, Yahner, and Lachman (2013) found that among a sample of 3,745 youth, 26% of adolescents experienced cyber abuse by a dating partner. Additional research is also needed to consider bidirectional associations between dating violence victimization and specific types of substance use in early adolescence. Previous research indicated that particular types of substances (e.g., alcohol and marijuana) may be more strongly related to dating violence victimization and that these paths may differ for boys and girls (e.g., Exner-Cortens et al., 2013; Foshee et al., 2013). Another important direction for future research is to identify processes underlying paths between substance use and dating violence victimization, such as emotion and coping responses, pharmacological side effects of substances, and patterns of risk-taking behaviors.

Our null findings related to sex differences suggest that longitudinal paths between dating violence victimization and substance use may be less affected by sex and coping tendencies and more a function of other factors, such as delinquent behavior and features of the peer context. Previous research has also found that peer deviancy is associated with dating violence victimization (Vézina et al., 2011) and substance use (Deković, 1999). Furthermore, deviant peer affiliations may place youth at risk for substance use as they provide opportunities, peer pressure, and reinforcement for engaging in substance use. In turn, they may also increase the likelihood of victimization in dating contexts as they place youth in risky environments with peers who may share attitudes and norms supportive of aggression. As peer group membership transitions from same-sex to mixed-sex groups in early adolescence, further research is needed to understand processes of peer influence and selection occurring in mixed-sex peer groups that may place youth at risk for these outcomes.

Last, future research should examine these relations among more diverse groups of adolescents. For instance, recent research indicated that compared with youth who are heterosexual, youth who identify as lesbian, gay, or bisexual are at higher risk for multiple forms of dating violence, including physical, psychological, cyber, and sexual violence (Dank, Lachman, Zweig, & Yahner, 2014). In addition, given that variations in the prevalence of substance use and dating violence victimization were found by race/ethnicity, another direction for future studies would be to assess whether bidirectional relations between these variables differ based on race/ethnicity. Unfortunately, the current study could not test racial/ethnic differences due to sample size limitations; however, this is an important area for future research.

Footnotes

Acknowledgements

The authors are grateful to the Multi-Site Violence Prevention Project for permission to use the data for this study. Investigators from each site are as follows (changes in affiliations in parentheses): Centers for Disease Control and Prevention, Atlanta, Georgia: Thomas R. Simon, Robin M. Ikeda, Emilie Smith (Penn State University); Le’Roy E. Reese (Morehouse University); Duke University, Durham, North Carolina: David L. Rabiner, Shari Miller (Research Triangle Institute), Donna-Marie Winn (University of North Carolina–Chapel Hill), Kenneth A. Dodge, Steven R. Asher; University of Georgia, Athens, Georgia: Arthur M. Horne, Pamela Orpinas, Roy Martin, William H. Quinn (Clemson University); University of Illinois at Chicago, Chicago, Illinois: Patrick H. Tolan (University of Virginia), Deborah Gorman-Smith (University of Chicago), David B. Henry, Franklin N. Gay (University of Chicago), Michael Schoeny (University of Chicago); Virginia Commonwealth University, Richmond, Virginia: Albert D. Farrell, Aleta L. Meyer (Administration for Children and Families, Washington, D.C.); Terri N. Sullivan, Kevin W. Allison.

Authors’ Note

This article is the result of the author’s independent research and the views expressed in the article do not represent the views of the U.S. Department of Education of the United States. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded, in part, by the National Center for Injury Prevention and Control, Centers for Disease Control and Prevention, CDC Cooperative Agreements U81/CCU317633 and 1U01CE001956.

Author Biographies

Katherine A. Taylor is an education research analyst at the Institute of Education Sciences, U.S. Department of Education. She received her PhD in developmental psychology from Virginia Commonwealth University. Her research interests include risk factors and consequences of youth violence exposure, including peer victimization and dating violence. She is also interested in school-based prevention programs that target youth violence and promote positive development.

Terri N. Sullivan is a professor at Virginia Commonwealth University. She received her PhD in clinical psychology from Virginia Commonwealth University. Her work focuses on the impact of violence exposure (e.g., exposure to community violence and peer victimization) on children’s psychosocial and emotional development. She is especially interested in the development and evaluation of the effectiveness of school-based youth violence prevention programs for children and adolescents, with a focus on youth with disabilities.

References

Ackard

D. M.

Eisenberg

M. E.

Neumark-Sztainer

(2007). The long-term impact of adolescent dating violence on the behavioral and psychological health of male and female youth. Journal of Pediatrics, 151, 476-481. doi:10.1016/j.jpeds.2007.04.034

Agnew

(2006). Pressured into crime: An overview of general strain theory. Los Angeles, CA: Roxbury Press.

Arriaga

X. B.

Foshee

V. A.

(2004). Adolescent dating violence: Do adolescents follow in their friends’, or their parents’, footsteps? Journal of Interpersonal Violence, 19, 162-184. doi:10.1177/0886260503260247

Bowllan

N. M.

(2011). Implementation and evaluation of a comprehensive, school-wide bullying prevention program in an urban/suburban middle school. Journal of School Health, 81, 167-173. doi:10.1111/j.1746-1561.2010.00576.x

Breiding

M. J.

Basile

K. C.

Smith

S. G.

Black

M. C.

Mahendra

R. R.

(2015). Intimate partner violence surveillance: Uniform definitions and recommended data elements (Version 2.0). Atlanta, GA: National Center for Injury Prevention and Control, Centers for Disease Control and Prevention.

Broidy

Agnew

(1997). Gender and crime: A general strain theory perspective. Journal of Research in Crime & Delinquency, 34, 275-306.

Carver

C. S.

Scheier

M. F.

Weintraub

J. K.

(1989). Assessing coping strategies: A theoretically based approach. Journal of Personality and Social Psychology, 56, 267-283.

Choi

H. J.

Weston

Temple

J. R.

(2017). A three-step latent class analysis to identify how different patterns of teen dating violence and psychosocial factors influence mental health. Journal of Youth and Adolescence, 46, 854-866. doi:10.1007/s10964-016-0570-7

Connolly

McIsaac

(2009). Adolescents’ explanations for romantic dissolutions: A developmental perspective. Journal of Adolescence, 32, 1209-1223. doi:10.1016/j.adolescence.2009.01.006

10.

Dank

Lachman

Zweig

J. M.

Yahner

(2014). Dating violence experiences of lesbian, gay, bisexual, and transgender youth. Journal of Youth and Adolescence, 43, 846-857. doi:10.1007/s10964-013-9975-8

11.

Deković

(1999). Risk and protective factors in the development of problem behavior during adolescence. Journal of Youth and Adolescence, 28, 667-685.

12.

Donovan

J. E.

Jessor

Costa

F. M.

(1988). Syndrome of problem behavior in adolescence: A replication. Journal of Consulting and Clinical Psychology, 56, 762-765.

13.

Exner-Cortens

Eckenrode

Rothman

(2013). Longitudinal associations between teen dating violence victimization and adverse health outcomes. Pediatrics, 131, 71-78. doi:10.1542/peds.2012-1029

14.

Farrell

A. D.

Kung

E. M.

White

K. S.

Valois

R. F.

(2000). The structure of self-reported aggression, drug use, and delinquent behaviors during early adolescence. Journal of Clinical Child & Adolescent Psychology, 29, 282-292. doi:10.1207/S15374424jccp2902_13

15.

Foshee

V. A.

Benefield

T. S.

Ennett

S. T.

Bauman

K. E.

Suchindran

(2004). Longitudinal predictors of serious physical and sexual dating violence victimization during adolescence. Preventive Medicine, 39, 1007-1016. doi:10.1016/j.ypmed.2004.04.014

16.

Foshee

V. A.

Karriker-Jaffe

K. J.

Reyes

Ennett

S. T.

Suchindran

Bauman

K. E.

Benefield

T. S.

(2008). What accounts for demographic differences in trajectories of adolescent dating violence? An examination of intrapersonal and contextual mediators. Journal of Adolescent Health, 42, 596-604. doi:10.1016/j.jadohealth.2007.11.005

17.

Foshee

V. A.

Linder

G. F.

Bauman

K. E.

Langwick

S. A.

Arriaga

X. B.

Heath

J. L.

. . . Bangdiwala

(1996). The Safe Dates Project: Theoretical basis, evaluation design, and selected baseline findings. American Journal of Preventive Medicine, 12, 39-47.

18.

Foshee

V. A.

Reyes

H. L. M.

Gottfredson

N. C.

Chang

L. Y.

Ennett

S. T.

(2013). A longitudinal examination of psychological, behavioral, academic, and relationship consequences of dating abuse victimization among a primarily rural sample of adolescents. Journal of Adolescent Health, 53, 723-729. doi:10.1016/j.jadohealth.2013.06.016

19.

Gover

A. R.

(2004). Risky lifestyles and dating violence: A theoretical test of violent victimization. Journal of Criminal Justice, 32, 171-180. doi:10.1016/j.jcrimjus.2003.12.007

20.

Goncy

E. A.

Sullivan

T. N.

Farrell

A. D.

Mehari

K. R.

Garthe

R. C.

(2017). Identification of patterns of dating aggression and victimization among urban early adolescents and their relations to mental health symptoms. Psychology of Violence, 7, 58-68. doi:10.1037/vio0000039

21.

Grant

B. F.

Dawson

D. A.

(1998). Age of onset of drug use and its association with DSM-IV drug abuse and dependence: Results from the National Longitudinal Alcohol Epidemiologic Survey. Journal of Substance Abuse, 10, 163-173. doi:10.1016/S0899-3289(99)80131-X

22.

Henry

D. B.

Farrell

A. D.

, & Multisite Violence Prevention Project. (2004). The study designed by a committee: Design of the Multisite Violence Prevention Project. American Journal of Preventive Medicine, 26, 12-19. doi:10.1016/j.amepre.2003.09.027

23.

Jessor

Turbin

M. S.

Costa

F. M.

Dong

Zhang

Wang

(2003). Adolescent problem behavior in China and the United States: A cross-national study of psychosocial protective factors. Journal of Research on Adolescence, 13, 329-360. doi:10.1111/1532-7795.1303004

24.

Johnston

L. D.

O’Malley

P. M.

Miech

R. A.

Bachman

J. G.

Schulenberg

J. E.

(2017). Demographic subgroup trends among adolescents in the use of various licit and illicit drugs, 1975–2016 (Monitoring the Future Occasional Paper No. 88). Ann Arbor: Institute for Social Research, The University of Michigan. Retrieved from http://www.monitoringthefuture.org/pubs/occpapers/mtf-occ88.pdf

25.

Kaukinen

(2014). Dating violence and college students: The risk and protective factors. Trauma, Violence, & Abuse, 15, 283-296.

26.

Lazarus

R. S.

(2006). Stress and emotion: A new synthesis. New York, NY: Springer.

27.

Lormand

D. K.

Markham

C. M.

Peskin

M. F.

Byrd

T. L.

Addy

R. C.

Baumler

Tortolero

S. R.

(2013). Dating violence among urban, minority, middle school youth and associated sexual risk behaviors and substance use. Journal of School Health, 83, 415-421.

28.

McNaughton Reyes

H. L.

Foshee

V. A.

Bauer

D. J.

Ennett

S. T

. (2012). Developmental associations between adolescent alcohol use and dating aggression. Journal of Research on Adolescence, 22, 526-541. doi:10.1111/j.1532-7795.2012.00799.x

29.

Miller-Johnson

Sullivan

T. N.

Simon

T. R.

, & Multisite Violence Prevention Project. (2004). Evaluating the impact of interventions in the Multisite Violence Prevention Study: Samples, procedures, and measures. American Journal of Preventive Medicine, 26, 48-61. doi:10.1016/j.amepre.2003.09.015

30.

Multisite Violence Prevention Project. (2004). The Multisite Violence Prevention Project: Background and overview. American Journal of Preventive Medicine, 26, 3-11.

31.

Muthén

L. K.

Muthén

B. O.

(2012). Mplus user’s guide. Los Angeles, CA: Author.

32.

Niolon

P. H.

Vivolo-Kantor

A. M.

Latzman

N. E.

Valle

L. A.

Kuoh

Burton

. . . Tharp

A. T.

(2015). Prevalence of teen dating violence and co-occurring risk factors among middle school youth in high-risk urban communities. Journal of Adolescent Health, 56, S5-S13. doi:10.1016/j.jadohealth.2015.07.019

33.

O’Donnell

Stueve

Myint-U

Duran

Agronick

Wilson-Simmons

(2006). Middle school aggression and subsequent intimate partner physical violence. Journal of Youth and Adolescence, 35, 693-703. doi:10.1007/s10964-006-9086-x

34.

Oetting

E. R.

Beauvais

(1990). Adolescent drug use: Findings of national and local surveys. Journal of Consulting and Clinical Psychology, 58, 385-394. doi:10.1037/0022-006X.58.4.385

35.

O’Keefe

(1997). Predictors of dating violence among high school students. Journal of Interpersonal Violence, 12, 546-568. doi:10.1177/088626097012004005

36.

O’Leary

K. D.

Slep

A. M. S.

Avery-Leaf

Cascardi

(2008). Gender differences in dating aggression among multiethnic high school students. Journal of Adolescent Health, 42, 473-479. doi:10.1016/j.jadohealth.2007.09.012

37.

Orpinas

Hsieh

H. L.

Song

Holland

Nahapetyan

(2013). Trajectories of physical dating violence from middle to high school: Association with relationship quality and acceptability of aggression. Journal of Youth and Adolescence, 42, 551-565. doi:10.1007/s10964-012-9881-5

38.

Orpinas

Horne

A. M.

, & Multisite Violence Prevention Project. (2004). A teacher-focused approach to prevent and reduce students’ aggressive behavior: The GREAT teacher program. American Journal of Preventive Medicine, 26(Suppl1), 29-38. doi: 10.1016/j.amepre.2003.09.01

39.

Orpinas

Nahapetyan

Truszczynski

(2017). Low and increasing trajectories of perpetration of physical dating violence: 7-year associations with suicidal ideation, weapons, and substance use. Journal of Youth and Adolescence, 46, 970-981. doi:10.1007/s10964-017-0630-7

40.

Ostrowski

M. K.

(2009). Self-medication and violent behavior. New York, NY: LFB Scholarly Publishing.

41.

Roberts

T. A.

Klein

J. D.

Fisher

(2003). Longitudinal effect of intimate partner abuse on high-risk behavior among adolescents. Archives of Pediatrics & Adolescent Medicine, 157, 875-881. doi:10.1001/archpedi.157.9.875

42.

Rose

A. J.

Rudolph

K. D.

(2006). A review of sex differences in peer relationship processes: Potential trade-offs for the emotional and behavioral development of girls and boys. Psychological Bulletin, 132, 98-131. doi:10.1037/0033-2909.132.98

43.

Sampson

R. J.

Raudenbush

S. W.

Earls

(1997). Neighborhoods and violent crime: A multilevel study of collective efficacy. Science, 277, 918-924. doi:10.1126/science.277.5328.918

44.

Saunders

Kilpatrick

Hanson

Resnick

Walker

(1999). Prevalence, case characteristics, and long-term psychological correlates of child rape among women: A national survey. Child Maltreatment, 4, 187-200.

45.

Simon

T. R.

Ikeda

R. M.

Smith

E. P.

Reese

L. R. E.

Rabiner

D. L.

Miller

. . . Orpinas

(2009). The ecological effects of universal and selective violence prevention programs for middle school students: A randomized trial. Journal of Consulting and Clinical Psychology, 77, 526-542. doi:10.1037/a0014395

46.

Simon

T. R.

Miller

Gorman-Smith

Orpinas

Sullivan

(2010). Physical dating violence norms and behavior among sixth-grade students from four U.S. sites. The Journal of Early Adolescence, 30, 395-409. doi:10.1177/0272431609333301

47.

Smith

E. P.

Gorman-Smith

Quinn

W. H.

Rabiner

D. L.

Tolan

P. H.

Winn

, (2004). Multisite violence prevention project. American Journal of Preventive Medicine, 26(Suppl1), 39-47. doi: 10.1016/j.amepre.2003.09.01

48.

Snijders

T. A. B.

Bosker

R. J.

(1994). Modeled variance in two-level models. Sociological Methods & Research, 22, 342-363. doi:10.1177/0049124194022003004

49.

Spriggs

A. L.

Halpern

C. T.

Martin

S. L.

(2009). Continuity of adolescent and early adult partner violence victimisation: Associaiton with witnessing violent crime in adolescence. Journal of Epidemiology and Community Health, 63, 741-748. doi: 10.1136/jech.2008.078592

50.

Swahn

M. H.

Bossarte

R. M.

Sullivent

E. E.

(2008). Age of alcohol initiation, suicidal behavior, and peer and dating violence vicitmization and perpetration among high-risk, seventh grade adolescents. Pediatrics, 121, 297-305. doi: 10.1542/peds.2006-2348

51.

Taylor

B. G.

Stein

N. D.

Mumford

E. A.

Woods

(2013). Shifting boundaries: An experimental evaluation of a dating violence prevention program in middle schools. Prevention Science, 14, 64-76. doi:10.1007/s11121-012-0293-2

52.

Temple

J. R.

Freeman

D. H.

(2011). Dating violence and substance use among ethnically diverse adolescents. Journal of Interpersonal Violence, 26, 701-718. doi:10.1177/0886260510365858

53.

Vézina

Hébert

Poulin

Lavoie

Vitaro

Tremblay

R. E.

(2011). Risky lifestyle as a mediator of the relationship between deviant peer affiliation and dating violence victimization among adolescent girls. Journal of Youth and Adolescence, 40, 814-824. doi:10.1007/s10964-010-9602-x

54.

Wincentak

Connolly

Card

(2017). Teen dating violence: A meta-analytic review of prevalence rates. Psychology of Violence, 7, 224-241. doi: 10.1037/a0040194

55.

Wolfe

D. A.

Crooks

C. V.

Jaffe

P. G.

Chiodo

Hughes

Ellis

. . . Donner

(2009). A school-based program to prevent adolescent dating violence: A cluster randomized trial. Archives of Pediatric & Adolescent Medicine, 163, 693-699.

56.

Zweig

J. M.

Dank

Yahner

Lachman

(2013). The rate of cyber dating abuse among teens and how it relates to other forms of teen dating violence. Journal of Youth and Adolescence, 47, 1063-1077. doi:10.1007/s10964-013-9922-8