Prospective associations among relationship abuse, sexual harassment and bullying in a community sample of sexual minority and exclusively heterosexual youth

Abstract

Introduction:

Sexual minority youth (SMY) are at significantly greater risk for experiencing adolescent relationship abuse (ARA) than exclusively heterosexual youth, yet little is known about the factors that elevate their risks for such abuse. Peer victimization (i.e., bullying, sexual harassment) has been associated with ARA among heterosexual youth. SMY experience higher rates of peer victimization than heterosexual youth, suggesting that it may be a risk factor for ARA among these youth. Using longitudinal data from a community sample of adolescents recruited from the northeastern US, we examined whether sexual identity was associated prospectively with ARA, and whether that relationship would be indirect, and mediated via bullying and sexual harassment at a 12-month follow-up. We expected to find higher rates of peer victimization and ARA among SMY than heterosexual youth, and we expected that sexual harassment and bullying would predict subsequent relationship abuse.

Methods:

Adolescents (N = 800; 58% female; 81% European-American; 19% SMY) between 13 to 15 years (M = 14.45, SD = 0.85) completed a web survey at baseline, 6-months and 12-months.

Results:

Consistent with prior studies, SMY reported higher rates of bullying, sexual harassment, and relationship abuse than heterosexual youth. SMY who reported sexual harassment at baseline were more likely to report relationship abuse over time. Contrary to our expectations, however, bully victimization did not predict subsequent relationship abuse.

Conclusions:

Sexual harassment can be particularly harmful for SMY because it targets gender and sexual identity at a formative developmental period and puts youth at risk for relationship abuse.

Adolescent relationship abuse (ARA), also known as teen dating violence, is a common and serious problem experienced by many adolescents. However, sexual minority youth (SMY; i.e., lesbian, gay, bisexual, questioning, and other non-heteronormative youth, or LGBQ) are at significantly greater risk than exclusively heterosexual youth for experiencing abuse within their romantic relationships (Dank, Lachman, Zweig, & Yahner, 2014; Espelage, Merrin, & Hatchel, 2018; Martin-Storey, 2015). Data from national and regional samples consistently indicate that SMY experience all types of ARA more frequently than heterosexual youth. For example, Zaza, Kann, and Barrios (2016) reported that SMY experienced significantly higher rates of physical (17.5% vs 8.3%) and sexual (22.7% vs 9.1%) dating violence than heterosexual youth in national-level data. Similar patterns have been observed in other national and regional samples (Luo, Stone, & Tharp, 2014; Martin-Storey, 2015; Reuter, Sharp, & Temple, 2015). Furthermore, youth who are unsure or questioning about their sexual identity or who have had same-sex experiences are at least as likely as those who identified as LGB to report relationship abuse (Edwards, 2015; Whitton, Newcomb, Messinger, Byck, & Mustanski, 2016). The factors that place SMY at increased risk of ARA are not well understood, and it is unknown whether SMY and heterosexual youth share similar risk factors for relationship abuse. Studies are critically needed to identify the risk factors and mechanisms associated with ARA among SMY to effectively target prevention and intervention efforts targeting these vulnerable youth.

What Constitutes Adolescent Relationship Abuse (ARA)?

ARA refers to the use of physical, psychological, emotional, verbal or sexual aggression perpetrated by one adolescent against another within the context of a dating, romantic or sexual relationship (Bandyopadhyay, Doekar, & Omar, 2010; Centers for Disease Control, 2014; Taylor & Mumford, 2014). Although the term is often used interchangeably with teen dating violence, ARA better reflects the commonly experienced dimensions of verbal, psychological and emotional abuse, in addition to physical and sexual violence. It also allows for a broader conceptualization of the relationships in which abuse can occur, which is important for SMY, who may not be in traditional “dating” relationships, yet may nonetheless be subject to abuse from romantic or sexual partners. The adverse mental and physical consequences of ARA are well-documented among heterosexual youth, resulting in outcomes ranging from acute harm (i.e., injury and death) to chronic consequences, such as depression, substance use and re-victimization, including intimate partner violence in adult relationships (Chiodo et al., 2012; Exner-Cortens, Eckenrode, & Rothman, 2012). Boys and girls are at risk for experiencing ARA as victims and perpetrators (Coker et al., 2000; O'Leary, Smith Slep, Avery-Leaf, & Cascardi, 2008). Much less is known about ARA among SMY, but research suggests that this group may face unique risks for ARA that are not found among heterosexual youth.

What May Explain the Heightened Vulnerability to Relationship Abuse among SMY?

Victimization across the lifespan.

ARA is one of many different types of violence experienced at higher rates among SMY than heterosexual youth. For example, SMY are at greater risk than heterosexual youth for experiencing childhood maltreatment, sexual abuse, and peer victimization (Anderson & Blosnich, 2013; Institute of Medicine, 2011; Kosciw, Greytak, Bartkiewicz, Boesen, & Palmer, 2012; Mitchell, Ybarra, & Korchmaros, 2014; Schneeberger, Dietl, Muenzenmaier, Huber, & Lang, 2014). Unfortunately, youth who experience one type of victimization are at risk of experiencing other types as well, a phenomenon known as polyvictimization (Finkelhor, Ormrod, & Turner, 2007). Although the mechanisms underlying polyvictimization are not well-understood, it is believed that early victimization experiences can have emotional, behavioral and social consequences that leave youth vulnerable to victimization in new contexts (Chiodo, Wolfe, Crooks, Hughes, & Jaffe, 2009; Finkelhor, Ormrod, Turner, & Holt, 2009; Finkelhor et al., 2007; Turner, Finkelhor, & Ormrod, 2010; Wolfe, Crooks, Chiodo, & Jaffe, 2009). Sterzing, Ratliff, Gartner, McGeough, and Johnson (2017) examined polyvictimization and related adverse outcomes among SMY. Although their study did not include a heterosexual comparison group, they found higher rates of polyvictimization among sexual minority boys (33%) than those found in national studies of heterosexual boys (20%). This pattern of revictimization also has been identified among adult sexual minorities (Balsam, Lehavot, & Beadnell, 2011; Hughes et al., 2010; Morris & Balsam, 2003), including studies showing the role of childhood sexual abuse in elevated risk for adult sexual victimization among adult sexual minority men and women (Hequembourg, Livingston, & Parks, 2013; Hequembourg, Parks, Collins, & Hughes, 2015). SMY, who are already disadvantaged due to sexual minority stress, appear to be highly vulnerable given their propensity to also be victimized in different life domains. Espelage et al. (2018), for example, found higher rates of both peer victimization and dating violence among SMY compared to heterosexual youth, with higher rates of both types of victimization associated with worse mental health outcomes. Similarly, using retrospective survey data, Walls et al. (2019) found that bully victimization increased the odds of experiencing partner violence for LGB and transgendered youth. Given higher rates of victimization among SMY, polyvictimization serves as a useful lens through which we can better understand how two forms of peer victimization—bully victimization and sexual harassment—may increase SMY vulnerability to relationship abuse.

Bully victimization.

Alarmingly high numbers of LGBT youth report experiences of peer victimization in the form of bully victimization and sexual harassment. Bullying is repeated aggressive behavior perpetrated by a more powerful peer or group against a less powerful one, with the intention to harm or disturb (Olweus, 1992). Bullying includes verbal (e.g., name-calling, threats), physical (e.g., hitting, kicking), relational (e.g., spreading rumors, exclusion), and cyber-based (i.e., via the internet or other electronic devices Agatston, Kowalski, & Limber, 2007; Ybarra, Boyd, Korchmaros, & Oppenheim, 2012) aggression. Data from the CDC’s 2017 High School Youth Risk Behavior Survey show that LGB students across the U.S. report significantly higher rates of bully victimization at school than heterosexual youth (34.2% vs 18.8%), including more experiences of electronic bullying (28.0% vs 14.2%; Zaza et al., 2016).

Sexual harassment.

Sexual harassment is “unwanted or unwelcome sexual behavior,” that interferes with one’s life (Hill & Silva, 2005 p. 2), ranging from sexual comments, gestures or requests to taking or sharing of sexual pictures. Like bully victimization, sexual harassment may manifest as verbal, physical, relational or cyber-aggression; but importantly, sexual harassment often targets the victim’s sexuality, gender, or sexual orientation (Hill & Silva, 2005). Sexual harassment during adolescence can be particularly damaging because it attacks an individual’s sexual identity at a developmentally sensitive period. Peer acceptance and approval are very important during mid-adolescence (Steinberg, 2008). Many teens begin to discover and explore their sexuality during this time, with many adolescents beginning to self-identify as LGBQ between the ages of 14-16 years (D'Augelli, 2003; Human Rights Campaign, 2013; PewResearch, 2013; Rosario et al., 1996). Heterosexism dominates this developmental period, with adolescents—particularly SMY—experiencing high rates of victimization from their peers for deviating from prescribed heterosexual norms (Birkett, Espelage, & Koenig, 2009; Collier, van Beusekom, Bos, & Sandfort, 2013; Hill & Silva, 2005). National survey data found that nearly three-quarters (73%) of college students who identify as LGBTQ experience sexual harassment (Hill & Silva, 2005). Questioning and LGB teens also are more likely than heterosexual youth to experience sexual harassment in the form of homophobic teasing (Espelage, Aragon, Birkett, & Koenig, 2008; Williams, Connolly, Pepler, & Craig, 2005).

Peer victimization also may have implications for the quality of functioning in romantic relationships later in adolescence. Indeed, research with predominantly heterosexual samples of adolescents show associations between bully victimization, sexual harassment victimization and dating violence in cross-sectional studies (Boivin, Lavoie, Hébert, & Gagné, 2012; Espelage & Holt, 2007) and at least one study found a prospective association between sexual harassment victimization and later involvement in ARA (Chiodo et al., 2009). However, there is a dearth of longitudinal studies to document these relationships prospectively with SMY. Further, it is unknown whether bullying and sexual harassment have differential influences on ARA. Peer victimization in the form of bullying can be devastating to all adolescents, but sexual harassment may be particularly harmful to adolescents who identify as LGBQ. Because sexual harassment may be especially damaging to the formation of a sexual identity among youth who are non-heteronormative, it may have an even greater impact on their relationship functioning than bullying compared to that of their heteronormative peers. This possibility has yet to be examined prospectively among SMY. Longitudinal studies are needed to examine whether bullying and sexual harassment are unique predictors of relationship abuse among SMY.

Purpose & Hypotheses

The purpose of the current study was to examine whether bullying and sexual harassment contribute uniquely and prospectively to ARA among SMY, as compared with heterosexual youth. Using longitudinal data collected from a community sample, we examined the direct and indirect prospective effects of sexual identity on ARA. We hypothesized that (1) SMY would experience higher rates of bullying, sexual harassment, and relationship abuse than heterosexual youth, (2) there would be direct effects of both bullying and sexual harassment on subsequent relationship abuse for all youth, and (3) the relationship between sexual identity and relationship abuse at 12-month assessment would be indirect and mediated via baseline and 6-month bullying and sexual harassment. We further expected that that indirect effect of sexual minority status on ARA would be stronger via sexual harassment than via bullying.

Method

Participants

Participants (N = 800, 58% female) ranged in age from 13 to 15 years (M = 14.45, SD = 0.85) at baseline recruitment. Consistent with the local demographics, approximately 81% of participants self-identified as White, 11% as Black, 4% as multiracial, 1% Asian American and < 1% as Native American. About 7% of the participants indicated that they were Hispanic/Latinx. The majority of participants were in 8^th (32%), 9^th (35%) or 10^th (23%) grades and attended a public (84%), private (12%) or charter school (4%). Median household income as reported by mothers of adolescents in this study was $80,000 and over.

Procedure

Recruitment.

Participants were recruited using address-based sampling, as part of a separate study designed to understand the mechanisms associated with peer victimization among adolescents. The mailing list was purchased from Click2Mail, a service that uses publicly available data to create targeted mailing lists. Mailing lists were comprised of household addresses from neighborhoods in Erie County, NY, that contained a high concentration of families with children in the targeted age range. An additional list targeting households in urban communities and with a high concentration of ethnic and racial minorities also was purchased, and additional mailing attempts were made to these addresses to increase sample diversity. At least two mailings were sent to each household on the list (up to four mailings for addresses in ethnically diverse and low-income neighborhoods). The project was described as a study of adolescent social relationships and health.

Individuals who responded to mailings were screened for eligibility over the phone. Eligible adolescents were between the ages of 13 and 15 years; attended a public, charter, or private school; and lived with a mother or legal female guardian who was also willing to participate in the study. Participants had to speak and read English at a 6^th grade level. Out of 1,152 responses to the mailing, 916 adolescents met eligibility criteria and were enrolled in the study. Of these, 29 screened eligible but declined to participate and 86 failed to complete the baseline survey after receiving the link. A total of 801 adolescents completed the baseline surveys. One case was excluded from the analysis due to veracity concerns, resulting in a sample of 800. Mothers provided informed consent for their children to participate, and adolescents provided assent prior to completing the surveys. The consent/assent forms and all procedures were approved by the University Institutional Review Board.

Survey.

The web-based surveys were administered using a secure server. Participants completed a baseline survey (Wave 1) and follow-up surveys at six (Wave 2) and 12 (Wave 3) months after completion of the baseline survey. Measures used in the follow-up were nearly identical to those completed at baseline, with the exception that some stable demographic information (e.g., race and ethnicity) were not reassessed and involvement in teen dating violence was added at the Wave 3 assessment. Retention rates were excellent, with 93.9% reporting at Wave 2 and 90.9% at Wave 3. Each survey took about 1 ½ hours to complete and participants received a $25 check for participation in each wave. Using available data, we compared those who successfully completed baseline surveys (n = 800) and those who started but did not complete (n = 77). Compared with those who completed baseline surveys, those with incomplete data were more likely to have experienced bully victimization in the past six months (M = 1.07, SD = 1.20 vs. M = 0.53, SD = 0.76), t(808) = 2.674, p = .008, and more likely to be males (4.5% vs. 0.7%), χ²(1) = 12.990, p < .001, and non-white (4.9% vs. 1.7%), χ²(1) = 5.998, p = .014. there were no differences in sexual identity, sexual harassment, age and the type of school.

Measures

Measures of bully victimization and sexual harassment victimization were administered at Waves 1 and 2. We began assessing ARA at Wave 3 based on research showing that the typical age of onset for relationship abuse is in mid- to–late adolescence (i.e., 15-16; Shorey et al., 2017). Demographics were assessed at Wave 1.

Demographics.

Participants reported their date of birth (extrapolated for age), race (recoded as White or non-White), sex (male or female), year in school, and type of school (i.e., public, private, charter) at baseline. Household income data were collected from participants’ mothers at baseline.

Bully victimization.

The California Bully Victimization Scale (CBVS; Felix, Sharkey, Green, Furlong & Tanigawa, 2011) was used to assess bullying perpetrated by and against peers in the past six months. The measure consists of six items that assess intentionally mean or hurtful acts of teasing, spreading rumors, ignoring/excluding others, physical harm (i.e., hit, push), making threats, and stealing or damaging property (e.g., In the past 6 months, how often have you been… ”). Reponses were, 0, ‘Never happened,’ 1, ‘less than once a month,’ 2, ‘about once a month,’ 3, ‘2 or 3 times per month,’ 4, ‘about once a week,’ 5, ‘several times per week,’ and 6, ‘every day or almost every day.’ The bully victimization scale had good internal consistency (α = 0.80 at wave 1; α = 0.83 at wave 2). We created a bully victimization score for each adolescent by taking the average of Wave 1 (baseline) and Wave 2 scores (r = 0.49), with higher scores indicating higher levels of bully victimization.

Sexual harassment.

Sexual harassment was assessed using a seven-item modified version of the American Association of University Women survey (American Association of University Women, 2001; Espelage, Basile, & Hamburger, 2012). Participants were asked to indicate how many times in the past six months they had experienced each of the following when they were not wanted (7 items): someone spreading sexual rumors, unwanted sexual touch, homophobic teasing, was sent or asked to send sexual pictures, received sexual threats, and someone made sexual jokes, gestures, or comments, sexual name calling (e.g., “slut,” “ho”). Reponses were, 0 ‘Never happened,’ 1, ‘less than once a month,’ 2, ‘about once a month,’ 3, ‘2 or 3 times per month,’ 4, ‘about once a week,’ 5, ‘several times per week,’ and 6, ‘every day or almost every day.’ The sexual harassment scale had good internal consistency (α = 0.78 at wave 1; α = 0.86 at wave 2). We averaged Wave 1 (baseline) and Wave 2 scores to create a sexual harassment score for each adolescent (r = 0.36), with higher scores indicating higher levels of sexual harassment.

As a follow up, those reporting bullying and/or sexual harassment were asked to identify the main person who did this. Responses included, classmate/student, friend, ex-friend, boyfriend/girlfriend, ex-boyfriend/girlfriend, sibling, friend of brother/sister, someone from a different school, unknown, or someone else. This question was asked separately for bullying and sexual harassment.

Relationship abuse.

The Conflict in Adolescent Dating Relationships Inventory - Short Form (CADRI-SF; Fernández-González, Wekerle, & Goldstein, 2012; Wolfe et al., 2001), a self-report measure of physical, sexual, threatening, relational and emotional/verbal abuse in adolescent dating relationships, was administered at Wave 3. Participants indicated how often in the past six months they and their partner engaged in each of the behaviors during a conflict in their current or most recent romantic relationship (e.g., I insulted my partner with put downs; my partner insulted me with put-downs). Items are rated on a four-point scale (0 “never” to 3 “often”). Self-to-partner dating violence and partner-to-self dating violence were highly correlated (r = 0.90). We summed all 22 items (including 11 self-to-partner and 11 partner-to-self) to create a single relationship abuse score for each teen. The relationship abuse scale had excellent internal consistency (α = 0.92).

Sexual identity.

Participants were asked to self-identify as “Completely heterosexual (attracted to persons of the opposite sex),” “mostly heterosexual,” “Bisexual (attracted to both men and women),” “Mostly homosexual,” “Completely homosexual (gay/lesbian attracted to persons of the same sex)” or “not sure.” Responses other than “Completely heterosexual” were coded as sexual minority. This approach was chosen because adolescent sexuality is fluid, and research suggests that adolescents who are not heteronormative, regardless of their sexuality, are at greater risk for peer victimization (DePalma & Atkinson, 2010). We used Wave 1 (baseline) assessment to differentiate heterosexual (0) and SMY (1).

Data Analytic Plan

Our data analytic plan was guided by our hypothesis that sexual identity would predict greater bully victimization and sexual harassment, which would in turn predict greater ARA. We tested the direct and indirect effects of sexual identity on ARA using a parallel mediator path analysis model in Mplus Version 8.2 (B. O. Muthen, Muthen, & Asparouhov, 2016; L. K. Muthen & Muthen, 2017). We used maximum likelihood estimation with robust standard errors. Missing data were addressed using full information maximum likelihood method and, hence, all available data were used. The statistical significance of the indirect effect was tested using 50,000 bootstrap draws to estimate with greater precision the 95% confidence intervals (CI) determined from the lower and upper 2.5 percentiles (Hayes, 2013; MacKinnon, Lockwood, & Williams, 2004). Sex (0 = male; 1 = female) was included as a covariate to control for its influence on the outcome (ARA) and on the two mediators (i.e., bully victimization and sexual harassment). Age and race (0 = White; 1 = non-White) also were included as covariates.

Results

Preliminary Analyses

Of 800 adolescents, 80.6% were self-identified as exclusively heterosexual and 19.4% as SMY. Eighty-eight percent (n = 707/800) of adolescents completed all three waves of data. Completers and non-completers (i.e., who did not complete survey at wave 2 and/or at wave 3) did not differ on baseline sex, age, sexual identity, type of school, race, bully victimization, and sexual harassment. However, compared with completers, non-completers were more likely to be males (54.8% vs. 40.9%), χ²(1) = 6.556, p = .010. Table 1 displays the descriptive results and the bivariate correlations among key variables. As expected, sexual identity was positively associated with the three variables of interest (i.e., relationship abuse, bully victimization, sexual harassment). ARA also was positively correlated with bully victimization and sexual harassment. SMY (n = 154) and heterosexual youth (n = 639) differed on baseline sex (27.9% male and 72.1% female vs. 46.0% male and 54.0% female), χ²(1) = 16.613, p < .001. Table 2 shows the results of group comparisons indicating that SMY report significantly higher relationship abuse, bully victimization, and sexual harassment scores than heterosexual youth. In addition, compared with male adolescents, females were more likely than males to have experienced bullying (M = 0.55, SD = 0.70 vs. M = 0.45, SD = 0.65), t(797) = −2.166, p = .031, and sexual harassment (M = 0.17, SD = 0.42 vs. M = 0.09, SD = 0.28), t(798) = −3.177, p = .002. Females and males did not differ in ARA (p = .570).

Table 1.

Descriptivea and Correlations for Bully Victimization, Sexual Harassment, and Adolescent Relationship Abuse

Variable	Descriptive Statistics				Correlations
	Min	Max	Mean	S.D.	1	2	3	4
1. Adolescent Relationship Abuse	0.00	44.00	2.40	5.49	--
2. Bully victimization	0.00	5.00	0.51	0.68	0.11*	--
3. Sexual harassment	0.00	3.40	0.14	0.37	0.20**	0.55**	--
4 Sexual identity	0.00	1.00	--	--	0 11*	0.20**	0.24**	--

^*Correlation is significant at the 0.05 level (2-tailed).

^**Correlation is significant at the 0.01 level (2-tailed).

Note: Bully victimization = averaged baseline (Wave 1) and 6-month victimization (Wave 2); Sexual harassment = averaged baseline (Wave 1) and 6-month sexual harassment (Wave 2); Relationship abuse measured at Wave 3; Sexual identity (baseline/Wave 1; 0 = exclusively heterosexual; 1 = sexual minority).

Table 2.

Comparison of Heterosexual and Sexual Minority Youth (SMY) on Primary Variables of Interest

Variable	Heterosexual				SMY				T-Test
	Min	Max	Mean	S.D.	Min	Max	Mean	S.D.	t(df)
Adolescent relationship abuse	0.00	44.00	2.10	4.92	0.00	44.00	3.59	7.31	−2.056 (351)*
Bully victimization	0.00	3.33	0.44	0.57	0.00	5.00	0.78	0.94	−5.818 (791)***
Sexual harassment	0.00	2.70	0.09	0.26	0.00	3.40	0.30	0.58	−6.796 (791)***

^***p < .001

^**p < .01

^*p < .05

Direct and Indirect Effects Model

Primary analyses involved testing the proposed parallel mediator model of the effects of sexual identity on ARA as mediated via bully victimization and sexual harassment. We tested the model controlling for effects of sex, age, and race on the outcome (ARA) and sex on the two mediators (bully victimization, sexual harassment). Data fit the model well, χ²(4) = 6.052, p = .195, CFI/TLI = 0.992/0.972, RMSEA = 0.025, SRMR = 0.016, BIC = 4148.674. As indicated in Figure 1, SMY (vs. heterosexual youth) predicted significantly higher bully victimization, b = 0.336 (0.079), p < .001, and sexual harassment, b = 0.190 (0.046), p < .001 (hypothesis 1). Sexual harassment positively predicted subsequent ARA, b = 2.918 (1.429), p = .041, but bully victimization did not, b = −0.041 (0.559), p = .942 (hypothesis 2). The indirect effect of sexual identity (SMY vs. heterosexual youth) on ARA via sexual harassment was significant, b = 0.555 (0.265), p = .036. The statistical significance of the indirect effect was confirmed using 50,000 bootstrap draws (95% CI = 0.004, 1.105; hypothesis 3). The direct effect from sexual identity to ARA was non-significant (p = .643). In addition, female sex positively predicted sexual harassment, b = 0.075 (0.021), p < .001, but did not predict bully victimization (p = .080) or ARA (p = .587). Age was positively associated with ARA, b = 0.981 (0.348), p = .005. Non-White adolescents (vs. White) were more likely to be sexually harassed, b = 2.472 (1.196), p = .039. We ran an additional model that excluded sexual harassment as a mediator. Bully victimization did not mediate the relationship between sexual identity and ARA, even when sexual harassment was removed from the model (p = .085).

Figure 1. Indirect Effects of Sexual Identity on Adolescent Relationship Abuse (ARA) via Bully Victimization and Sexual Harassment.

Path coefficients are unstandardized. The parallel mediators were allowed to be correlated.

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

Sexual identity (baseline/Wave 1; 0 = heterosexual; 1 = SMY). Covariates included sex, age and race. Sex was also allowed to control for its influence on the two mediators (i.e., bully victimization and sexual harassment).

We then tested whether sexual identity interacted with sex. Sexual identity did not interact with sex (sexual identity X sex; p = .589), and inclusion of the interaction term did not alter other pathways depicted in Figure 1.

At baseline, 163/176 (92.6%) adolescents who reported sexual harassment also reported their relationship to the main person who sexually harassed them. Of these 163 adolescents, 9 (5.5%) reported that they were sexually harassed by a boyfriend/girlfriend or ex-boyfriend/girlfriend. At wave 2, 88/108 (81.5%) adolescents who reported sexual harassment also reported their relationship to their main perpetrator. Of these 88 adolescents, 4 (4.5%) were sexually harassed by boyfriend/girlfriend or ex-boyfriend/girlfriend. To examine potential overlap of items across the sexual harassment and relationship abuse scales (e.g., questions on unwanted sexual touching appearing in both scales) and ensure that results were robust, we performed sensitivity analyses excluding these adolescents (that is, 9 participants at baseline and 4 participants at Wave 2) by coding them as zero for the sexual harassment scale and re-ran models (as displayed in Figure 1); the pattern of results was unchanged.

Discussion

The current study sought to determine whether peer victimization is a mechanism that may help to account for the relationship abuse disparities among SMY. We expected to find higher rates of peer victimization and ARA among SMY than heterosexual youth (hypothesis 1) and we expected that sexual harassment and bullying would predict subsequent ARA (hypothesis 2). Finally, in conjunction with conceptualizations of polyvictimization, we hypothesized that sexual minority status would be indirectly associated with 12-month ARA via 6-month bullying and sexual harassment victimization, with stronger effects for sexual harassment than for bullying (hypothesis 3). Our hypotheses were partially supported. Our results for Hypothesis 1 are consistent with prior studies indicating that SMY are more vulnerable to peer victimization, including bullying and sexual harassment, and ARA than heterosexual youth (e.g., Espelage et al., 2008; Hill & Silva, 2005; Williams, Connolly, Pepler, & Craig, 2003; Zaza et al., 2016). We also found that although both bullying and sexual harassment were correlated with ARA, contrary to expectations, only sexual harassment directly predicted ARA. Findings also supported an indirect effect of sexual identity on relationship abuse mediated via sexual harassment. That is, in comparison to exclusively heterosexual youth, SMY were more likely to experience sexual harassment, which in turn predicted ARA over time. This suggests that type of victimization is an important factor to consider in predicting subsequent victimization.

Results from the current study indicate that bully victimization did not predict ARA. We measured bully victimization and sexual harassment separately because we believe that they are distinct experiences with differential impact on victims (Espelage et al., 2008). Research about peer victimization among SMY, with some exceptions (Williams et al., 2003, 2005), has measured and included sexual harassment as a domain of bullying and peer victimization (Birkett et al., 2009). However, these measurement approaches of combining bullying and sexual harassment, we believe, obscure the critical and unique impact of sexual harassment on SMY. The results from the current study, indeed, support this conclusion: sexual harassment—but not bullying—predicted ARA among SMY. Thus, the novelty and strength of the current study is that our disentanglement of these related concepts fully highlights the importance of focusing on sexual harassment and its various dimensions to better understand unique risks for relationship abuse and potential points of intervention to reduce ARA among SMY (see Future directions below).

Perhaps one of the reasons that peer sexual harassment is so devastating to SMY is that it occurs against a larger social backdrop dominated by heterosexual norms and homophobic attitudes. SMY may already be stressed as they forge their sexual identity amid homophobic messages that are pervasive in society. When SMY are attacked directly and personally through peer sexual harassment, particularly in the form of homophobic teasing, it can adversely affect their self-concept, self-worth, confidence, and their beliefs about their potential to attract and retain romantic partners (American Association of University Women, 2001; Espelage et al., 2008; Martin-Storey & Crosnoe, 2012). As SMY grapple to reconcile their sexual identity and attraction with socially-prescribed heterosexist norms it can create dissonance, resulting in depression, internalized heterosexism and externalizing behaviors such as substance abuse and aggression (Espelage et al., 2008; Williams et al., 2005). These are all documented risk factors for relationship abuse (Maas, Fleming, Herrenkohl, & Catalano, 2010).

A novel aspect of the current study was our inclusion of youth across a spectrum of sexual identifications. Rather than focus narrowly on adolescents who self-identified as gay, lesbian or bisexual, we included all those participants who identified as other than exclusively heterosexual. Adolescence is a critical formative period when youth are exploring their sexuality. Although some youth are certain of their identities as LGB, others are questioning and still exploring their feelings. Furthermore, labels utilized by adolescents to describe their own sexual identities are rapidly expanding beyond the traditional gay, lesbian or bisexual categories (Flanders, LeBreton, Robinson, Bian, & Caravaca-Morera, 2017; Morandini, Blaszczynski, & Dar-Nimrod, 2017). Research consistently finds that LGB, as well as questioning and other non-heteronormative identified youth, are at heightened risk for experiencing peer victimization, including homophobic teasing, which can have detrimental consequences for the quality of their romantic relationships throughout adolescence and into adulthood. Knowledge about peer victimization and ARA among youth who report identities other than exclusively heterosexual is critical to prepare health care providers with the skills to recognize warning signs of victimization, intervene, and provide holistic care to adolescent patients and their families. Our results support focusing on sexual harassment, which can be particularly harmful for adolescents because it targets gender and sexual identity at a formative developmental period.

Study limitations.

The current study was part of a larger study of peer victimization whose focus was not specifically on SMY and thus, sexual identity status was not a criterion for eligibility. Maternal consent was required for adolescents to participate in the parent study. Therefore, it is possible that this requirement resulted in lower participation by adolescents who were not close to their mothers. Furthermore, unique factors associated with sexual minority stress were not assessed in the current study. Future studies are needed to include larger samples focusing on SMY to better understand how these stressors contribute to ARA. For example, sexual minority stressors such as experiences of discrimination and internalized heterosexism have been positively associated with identity concealment and intimate partner violence among adult sexual minorities (Carvalho, Lewis, Derlega, Winstead, & Viggiano, 2011; Edwards & Sylaska, 2013). Other factors that may elevate SMY vulnerability to ARA include: challenges of negotiating socially prescribed gender norms in relationships (Gillum & DiFulvio, 2012); having fewer options for dating partners, leading to over-reliance on unhealthy romantic relationships (Gillum & DiFulvio, 2012); and less disclosure and help-seeking for fear of prejudice. Furthermore, uncertainty about one’s sexual identity or not being out to others may grant unique power to aggressors by isolating and threatening to ‘out’ SMY, thereby preventing them from leaving violent relationships.

Eighty-one percent of the sample was White. Given the lack of racial diversity in the sample, we used a simplistic dichotomy (i.e., White/Other) for the current analyses. However, we recognize that patterns of dating violence differ by race (Fish, Livingston, VanZile-Tamsen, & Patterson Silver Wolf, 2017). The racial and ethnic composition of the sample are reflective of the geographic region from which the sample was drawn, but underscore that need for future studies that include more racially and ethnically diverse SMY participants to better understand the role of intersectionality in ARA outcomes among this vulnerable group. Likewise, as noted on page 7, the sample was very homogenous in terms of (high) reported household income at baseline. Samples that include a greater socioeconomic variability would allow for more nuanced analyses assessing the moderating role of these factors in ARA among SMY.

Sexual orientation is a challenging concept to measure for research purposes (Ridolfo, Miller, & Maitland, 2012), but particularly for adolescents who are still in the formative stages of life (Friedman et al., 2004; Matthews, Blosnich, Farmer, & Adams, 2014; Saewyc, 2011). In this manuscript, we use the term “sexual minority youth” to refer to adolescents who identified as other than exclusively heterosexual at Wave 1 of our study. Consequently, one limitation is that our measure of sexual identity does not account for changes in sexual identity from baseline to Wave 3. Furthermore, our measurement of sexual identity was limited, as it failed to differentiate between attraction and identity, and also did not account for behavior. Future studies need to consider the multidimensional and fluid nature of sexual orientation among adolescents, as it changes over time. These studies should include not only those adolescents who self-identify as lesbian, gay or bisexual, but also those who are questioning, identify as something other than exclusively heterosexual (e.g., pansexual, queer, asexual) or are engaged in same-sex behaviors. The rudimentary nature of the current study assessment of sexual identity as attraction and behavior likely resulted in an under-identification of SMY in the sample.

Improved measurement of sexual minority status in early adolescence is a work-in-progress. The challenge for future researchers—as pointed out in a recent article by Schrager, Steiner, Bouris, Macapagal, and Brown (2019)—is that there is virtually no precedence for these procedures nor current guidelines for addressing the complexity of this fluid identity process over time. Determining how best to include a spectrum of non-heteronormative identities, while also accounting for changes in youth’s sexual identifications over time, is imperative because research indicates that adolescents with greater variability in their sexual orientations over time experience more victimization and may be at greater risk for poor outcomes, such as substance use (Espelage et al., 2008; Reuter et al., 2015). Given the small number of adolescents in each of these categories for the current study, we were unable to examine whether there were differences between the self-identified LGB and questioning youth. Future research about the relationship between sexual harassment and ARA among SMY should include sufficient subsamples to examine these differences.

We tested the interaction between sex and sexual identity but did not find any differences. Our assessment of sex was rudimentary, allowing adolescents at baseline to select from only two options: male or female. Our sample may have included a small number of undisclosed gender-nonconforming or transgender youth because we did not assess sex assigned at birth and current gender identity. Transgender and gender non-conforming youth appear to be at similar or elevated risk compared to SMY for ARA and peer victimization (Toomey, Ryan, Diaz, Card, & Russell, 2010; Walls et al., 2019). Consideration of their unique experiences is an important goal that should be addressed in future research to enhance intervention and prevention efforts to reduce poor relationship outcomes among diverse youth.

We assessed ARA at Wave 3 given that onset and escalation of ARA typically occurs at or after ages 15-16 years (Shorey et al., 2017). Participants were 14-16 years old at Wave 3 and our assessment captured relationship abuse occurring in the six months prior to the assessment, thus, the incidence of ARA occurring prior to this assessment point is likely to be low. Nonetheless, it is entirely possible that some of the participants could have been involved in relationship abuse prior to this point, which may confound measurement of peer victimization. We acknowledge this as a limitation and recommend earlier assessment of ARA in future studies.

Future directions: Implications for intervention and prevention efforts.

The solutions for reducing relationship violence are complex and require targeting of risk and protective factors simultaneously from multiple vantage points across the lifespan. Unfortunately, research identifying protective factors and buffering effects against ARA is sparse (Vagi et al., 2013) in general, and almost non-existent for SMY (Coulter et al., 2019). However, Niolon et al. (2017) outline a variety of promising points of intervention to reduce relationship violence, including those aimed at adolescents and emerging adults to promote healthy and safe relationships, engage influential parents and peers to condemn violence and unhealthy relationship dynamics, and improve school climates and communities to create protective environments that promote healthy relationships. When considering adolescents as a group, promising dating violence interventions have been developed to target these factors among middle and high school students (e.g., E. Miller et al., 2012; S. Miller et al., 2015; Munoz-Fernandez, Ortega-Rivera, Nocentini, Menesini, & Sanchez-Jimenez, 2019; Taylor, Stein, Mumford, & Woods, 2013). For example, Shifting Boundaries (Taylor, Mumford, & Stein, 2015; Taylor et al., 2013) targets reductions in both sexual harassment and dating violence by promoting interpersonal skills, establishing personal boundaries, and educating students about the legal aspects of sexual harassment and ARA. This program has been associated with a reduction in both forms of victimization. Coaching Boys into Men prepares athletic coaches to deliver social norms messages surrounding gender-equality, relationship abuse, and bystander intervention. This program has shown some success in enabling male high school athletes to recognize relationship abuse and increase positive attitudes towards intervening; however, it had no effect on gender-equitable attitudes or perpetration of relationship abuse (E. Miller et al., 2012).

Prevention programs targeting peer-to-peer sexual harassment, as well as related efforts to target normative and detrimental gender norms (i.e., hypermasculinity) also have reduced sexual violence among adolescents. For example, the Second Step Middle School Program (Committee for Children, 2008) aims to prevent violence, aggression and substance use through the promotion of social skills, empathy, and school connectedness. The program was successful in reducing bullying, sexual harassment, and homophobic name-calling through decreased involvement in delinquent activities (Espelage, Low, Polanin, & Brown, 2015). Although the impact on ARA was not evaluated, delinquency was a focus of these prevention efforts and—given its association with ARA—reductions in delinquency may ultimately contribute to overall decreases in ARA. Findings from the current study suggest that preventing sexual harassment, in particular, may have the additional benefit of reducing vulnerability to ARA among SMY and, therefore, should be a priority. Some efforts to target sexual harassment show promising results for improving conditions for SMY. Espelage et al. (2008), for example, found that positive school climate and parental support can protect SMY youth against poor health outcomes related to homophobic teasing. Mitchell et al. (2014) advocated for socio-economic programs that would strengthen individual self-esteem after finding lower levels of reported sexual harassment among gender and sexual minority youth who reported higher levels of social support and self-esteem.

Conclusion

SMY experience disproportionately higher rates of ARA compared with their heterosexual peers, and this vulnerability can be at least partially attributed to earlier sexual harassment victimization. Findings point to the need for prevention efforts to reduce adolescent sexual harassment, particularly that which targets sexual minority youth. Interventions to help SMY cope with the harmful effects of sexual harassment and polyvictimization also are warranted. Additional research is needed to identify protective factors that can reduce SMY risk of ARA.