Intimate Partner Violence Victimization in LGBT Young Adults: Demographic Differences and Associations with Health Behaviors

Abstract

Objective

Intimate partner violence (IPV) is an important public health problem with high prevalence and serious costs. Although literature has largely focused on IPV among heterosexuals, studies have recently begun examining IPV in LGBT samples, with mounting evidence suggesting IPV may be more common among LGBT individuals than heterosexuals. Less research has examined the specific health consequences of IPV in this population, particularly across time and among young people, and it remains unclear whether experiences of IPV differ between subgroups within the LGBT population (e.g. race, gender identity, and sexual orientation).

Method

An ethnically diverse sample of 172 LGBT young adults completed self-report measures of IPV, sexual behavior, mental health, and substance abuse at two time points (4- and 5-year follow-up) of an ongoing longitudinal study of LGBT youth.

Results

IPV was experienced non-uniformly across demographic groups. Specifically, female, male-to-female transgender, and Black/African-American young adults were at higher risk compared to those who identified as male, female-to-male transgender, and other races. Being a victim of IPV was associated with concurrent sexual risk taking and prospective mental health outcomes but was not associated with substance abuse.

Conclusions

Demographic differences in IPV found in heterosexuals were replicated in this LGBT sample, though additional research is needed to clarify why traditional risk factors found in heterosexual young people may not translate to LGBT individuals. Studies examining the impact of IPV on negative outcomes and revictimization over time may guide our understanding of the immediate and delayed consequences of IPV for LGBT young people.

Intimate partner violence (IPV; commonly referred to as dating violence in adolescence) is a serious public health problem in youth and young adults (Center for Disease Control and Prevention, 2012). Decades of empirical research have clearly demonstrated that IPV is associated with substantial physical and emotional costs (Campbell, 2002; Exner-Cortens, Eckenrode, & Rothman, 2013; Malik, Sorenson, & Aneshensel, 1997; Silverman, Raj, Mucci, & Hathaway, 2001) and has an alarmingly high prevalence, with an estimated 10–20% of youth and young adults having experienced physical or sexual violence from a romantic partner (Eaton, Davis, Barrios, Brener, & Noonan, 2007; Shorey, Cornelius, & Bell, 2008). Mounting evidence has also demonstrated that LGBT (lesbian, gay, bisexual, transgender) individuals, particularly young people, are at increased risk for IPV relative to heterosexuals (Dank, Lachman, Zweig, & Yahner; 2014; Reuter, Sharp, & Temple, 2015). Given the strong associations between IPV and serious consequences in heterosexual samples, the purpose of the present study was to examine the relation between IPV and health behaviors in an ethnically diverse, community sample of LGBT young adults.

IPV typically refers to abusive behavior occurring within the context of romantic relationships. More conservative definitions of IPV are often confined to physical and sexual abuse (e.g., hitting, punching, forced sex), while more liberal definitions typically extend IPV to include non-physical behaviors (e.g., isolating, ridicule, verbal threats, provoking jealousy). Importantly, both forms of abuse (i.e., physical and non-physical) are linked with depression, conflict, aggression, and numerous physical health outcomes (Coker, Smith, Bethea, King, & McKeown, 2000; Pico-Alfonso et al., 2006), with some studies finding that victims report psychological abuse as equally or more damaging than physical abuse (Reeves & Orpinas, 2012).

Serious short- and long-term consequences of IPV, coupled with high prevalence, have driven researchers to formulate theoretical frameworks to explain why individuals engage in abusive behaviors towards their partners. While there is arguably no unifying theory of IPV (Shorey et al., 2008), a few explanatory models have received significant attention with some empirical support, such as social learning theory (Bandura, 1973) and feminist theory (Dobash & Dobash, 1977). However, because results are often mixed and inconsistent, more recent models of IPV examine a broader variety of factors across multiple ecological levels in attempts to better understand and explain the etiology of partner violent behaviors (Bell & Naugle, 2008; O’Leary, Smith Slep, & O’Leary, 2007). This integrative framework of IPV has established a host of risk factors, cutting across demographic characteristics, psychopathology, exposure to violent models, and substance use (Rothman, Reyes, Johnson, & LaValley, 2012; Sims, Dodd, & Tejeda, 2008). Significantly less research, however, has examined IPV in LGBT populations, particularly its associations with demographics and health behaviors. Note that throughout the article, the LGBT acronym is used to describe the “LGBT community” or specific study samples that included all LGBT subgroups. The LGB acronym is used for samples that included lesbian, gay, and bisexual participants but did not include or specifically identify transgender participants.

Correlates of IPV

Regarding demographic differences, research has demonstrated that although women are equally or more likely to perpetrate at least one act of physical aggression compared to men, they are more likely to be victimized by severe acts of violence, suffer physical injury, and experience more frequent and longer lasting victimization (Archer, 2000; Foshee, 1996; Malik et al., 1997; Tjaden & Thoennes, 2000). The relation between race and IPV is less clear. Some studies have found higher rates and more permissive norms of violence among racial/ethnic minorities compared to White individuals (Chase, Treboux, & O’Leary, 2002), while other studies have found no such difference (Clark, Beckett, Wells, & Dungee-Anderson, 1994). Little research has examined this from an intersectional perspective in terms of demographic differences within minority populations.

Though the relation between substance use and partner violence is likely bi-directional, numerous studies have repeatedly demonstrated IPV to be a robust predictor of substance use (Foshee et al., 2004; Rothman et al., 2012, Temple & Freeman, 2011). IPV has also been linked to mental health problems and risky sexual behavior (Callahan, Tolman, & Saunders, 2003; Silverman et al., 2001; Wingood, DiClemente, McCree, Harrington, & Davies, 2001). In addition to these more immediate concerns, victims and perpetrators of IPV are at increased risk for perpetuating violence in their future romantic relationships, even when controlling for child abuse and socioeconomic status (Gidycz, Warkentin, & Orchowski, 2007; Stith, Smith, Penn, Ward, & Tritt, 2004).

In summary, numerous studies have successfully identified a host of factors associated with IPV victimization in heterosexual samples. However, research involving LGBT young people lags behind, and findings derived from predominately heterosexual samples may not necessarily generalize to LGBT samples. That is, it is possible that the dating experiences of LGBT couples are characteristically different from heterosexual couples due to a number of unique factors (e.g., internalized stigma, lack of same-sex couple role models, “outness” of one or both individuals), which may place additional stress on the dyad and/or limit the translatability of the dominant heteronormative model to this population. Further, as previous authors have noted (e.g., Exner-Cortens et al., 2013), the majority of research conducted is cross-sectional and thus it remains unclear whether identified health outcomes are causes or consequences of IPV.

Research on IPV in Non-Heterosexual Samples

The few empirical studies that have investigated IPV in LGBT samples consistently find that LGBT youth and young adults are more likely to experience IPV compared to their heterosexual counterparts (Edwards, Sylaska, & Neal, 2015; Halpern et al., 2004; Martin-Storey, 2014; Mustanski, Andrews, Herrick, Stall, & Schnarrs, 2014). A guiding theory that may best explain these findings is a minority stress framework (Meyer, 2003), which states that the combination of minority status and majority values often leads to conflict. Specifically, LGBT young people may experience a host of distal and proximal stressors across interpersonal and institutional contexts, including internalized homophobia, expectations of rejection, and hostility from the social environment, which, in turn, are associated with increased stress and poor health outcomes (Meyer, 2003). These minority stressors are also associated with important social, cognitive, and emotional processes (Hatzenbuehler, 2009), such as emotion dysregulation and social isolation, which may explain higher rates of IPV specifically. Indeed, when controlling for experiences of discrimination, the relation between sexual minority status and poor health outcomes is weakened significantly (Mays & Cochran, 2001). Individuals with multiple gender, racial, and/or sexual identities (e.g., female-identified, youth of color) may also face additional minority stressors and thus experience worse outcomes compared to those with a single minority status (Crawford, Allison, Zamboni, & Soto, 2002; Meyer, Dietrich, & Schwartz, 2008).

In one of the first empirical studies examining IPV among LGB youth, Freedner, Freed, Yang, and Austin (2002) administered self-report surveys measuring five types of IPV victimization (control, emotional, scared for safety, physical, sexual) to a convenience sample of 521 adolescents and found that, compared to heterosexuals, bisexual male adolescents were over three times more likely to report any form of IPV victimization and lesbians were twice as likely to report fearing for their safety. Significantly less research, however, has utilized within-group designs allowing for better understanding of specific factors associated with IPV among LGBT youth. For example, Dank and colleagues (2014) showed that LGBT youth, particularly transgender and female adolescents, were at increased risk for IPV victimization and perpetration across almost all forms of abuse (e.g., physical, psychological, cyber, sexual coercion), and this was associated with elevated depression scores, poorer academic achievement, and previous sexual activity. In addition to mental health correlates, research has also examined the relation between demographic factors and IPV. Luo et al. (2014) analyzed ten years of geographically diverse data from the Youth Risk Behavior Surveillance System (YRBS) and showed that LGB adolescents were twice as likely to report physical IPV victimization (odds ratio [OR] = 2.46). Interestingly, this finding was largely consistent across gender, age, and ethnicity.

Although the previously discussed studies have made invaluable contributions to an area that has historically received little attention, research on IPV among LGBT youth and young adults remains in its infancy, with numerous authors calling attention to the need for well-controlled, longitudinal studies that match the methodological rigor of research using predominately heterosexual populations (Burke & Follingstad, 1999; Dank et al., 2014). Most studies examining IPV in LGBT young people are qualitative or quasi-empirical, cross-sectional, assess IPV through a single item, lack ethnic diversity, do not include transgender participants, and/or study one gender exclusively. Beyond confirming that LGBT youth and young adults report higher rates of IPV compared to heterosexual youth, it is important for research to further unpack this finding and explore factors that are associated with IPV in this population. As previous authors have pointed out (Luo et al., 2014), it remains unclear whether traditional correlates of IPV derived from heterosexual samples are relevant to LGBT young people.

Purpose of the Present Study

Given the robust association between IPV and a host of serious consequences in heterosexual samples, the present study aimed to examine both cross-sectionally and longitudinally the relation between IPV and health behaviors in an ethnically diverse, community sample of LGBT young adults. Specifically, we hypothesize that 1) female-identified and racial/ethnic minorities will experience higher rates of IPV than male-identified and White young people; 2) IPV will be concurrently associated with higher depression and anxiety scores, substance use (e.g., marijuana use, binge drinking), and sexual risk taking behaviors (e.g., number of sexual partners, number of condomless vaginal or anal sex acts with a serodiscordant partner); and 3) IPV will prospectively predict depression and anxiety, substance use, and sexual risk taking one year later.

Methods

Participants and Procedures

The sample included 172 ethnically diverse young adults in the Chicago area, drawn from a larger longitudinal study of LGBT youth (ages 16–20 at baseline) involving eight waves of data collection over 5 years (Mustanski, Garofalo, & Emerson, 2010; Newcomb & Mustanski, 2014). This study was approved by the appropriate Institutional Review Boards with a waiver of parental permission for minor participants. Recruitment was achieved through an initial convenience sample (i.e., flyers posted in LGBT community centers and neighborhoods; emails sent to college LGBT organization listservs; 38%) followed by incentivized peer recruitment (62%). All recruitment materials (i.e., flyers, emails, peer recruitment coupons) described the project as a study examining issues affecting lesbian, gay, bisexual, transgender, queer, questioning, or same-sex attracted youth. Participants completed self-report measures of health behaviors, mental health, and psychosocial variables at each wave and were compensated $25 to $50 at each time point. Assessments were conducted in private rooms at university and community locations. In order to address the aims of the present study, we used Wave 4 and 5 data because these were the first waves to include measures that assessed multiple forms of IPV (e.g. physical, verbal). Therefore, data for analyses were from the 4- and 5-year follow-up time points (2011–2014, and retention was 83% at 4-year follow-up and 82% at 5-year follow-up). Retention rates may differ from previous reports based on differences between analytic samples. Average age at 4-year follow-up was 22.48 (SD = 1.61). Only participants who indicated having a sexual and/or romantic partner at 4-year follow-up answered questions assessing IPV and were included in current analyses (N = 172). Breakdown of demographics by birth sex, gender identity, sexual orientation, race, living situation, and education at baseline are displayed in Table 1.

Table 1.

Description of lesbian, gay, bisexual, and transgender youth sample at baseline (analytic sample; N = 172), n (%)

Variable	Value
Birth Sex
Male	72 (41.9)
Female	100 (58.1)
Gender Identity
Male	62 (36.5)
Female	92 (54.1)
Male-to-Female	9 (5.3)
Female-to-Male	7 (4.1)
Sexual Orientation
Lesbian	50 (29.4)
Gay	51 (30.0)
Bisexual	56 (32.9)
Questioning/Unsure/Other	13 (7.7)
Race
Latino	20 (11.6)
Black	100 (58.1)
White	23 (13.4)
Other	29 (16.8)
Living Situation
Living with Parents	111 (65.3)
Other Stable Housing	50 (29.4)
Unstable Housing	9 (5.3)
Highest Level of Education
College	10 (5.9)
Partial College	36 (21.2)
High School	47 (27.6)
Partial High School	72 (42.4)
Less than High School	5 (2.9)

Measures

Demographics

A demographics questionnaire was administered at each wave to assess participant age, birth sex, gender identity, self-reported sexual orientation, race, living situation, and education.

Intimate Partner Violence Victimization

At each wave, participants completed the HIV-Risk Assessment of Sexual Partnerships (H-RASP; Mustanski, Starks, & Newcomb, 2014), which assessed sexual behaviors and associated partner/relationship characteristics for up to three sexual and/or romantic partners during the preceding six months. For each partnership, participants were asked whether they had ever been the victim of physical, sexual, or verbal abuse. Physical abuse was assessed by asking, “Has this partner ever hit, slapped, punched, or hurt you?” Forced sex was assessed by asking, “Has this partner ever forced you to have vaginal, anal, or oral sex when you did not want to?” Verbal abuse was assessed by asking, “Did this partner call you names, insult you, or treat you poorly?”, “Did this partner swear at you?”, and “Did this partner threaten you with violence?” A dichotomous variable indicated positive endorsement of physical IPV if participants answered yes to the physical abuse item. A dichotomous variable indicated positive endorsement of verbal IPV if participants answered yes to any of these verbal abuse items. Finally, we created a dichotomous “any IPV” variable, which was scored positive if participants answered yes to any of the IPV items (physical, sexual, or verbal). Due to the small number of participants reporting sexual abuse (N = 6), we do not examine demographic differences in sexual IPV or whether this variable was concurrently or prospective associated with health outcomes.

Sexual Risk Taking

The H-RASP also assessed sexual behaviors during the preceding six months. Participants were asked to report their total number of sexual partners and total number of condomless vaginal or anal sex acts with a serodiscordant partner in the past six months (female/female condomless sex acts excluded).

Mental Health Symptoms

The Brief Symptom Inventory (BSI-18; Derogatis, 2001) is an 18-item self-report measure of psychological distress for the previous week that includes three domains: somatization (e.g., “feeling weak”), depression (e.g., “feeling blue”), and anxiety (e.g., “feeling tense”). It is a widely used screening tool in epidemiological research and clinical settings with adequate reliability and validity (Zabora et al., 2001). For the present study, the BSI 18 demonstrated strong internal consistency across waves (Cronbach alpha ranged .91 to .94).

Binge Drinking and Marijuana Use

Binge drinking and marijuana use were assessed at each visit using a self-report measure. Participants were first asked if they had drank alcohol and used marijuana during the past six months. Participants who endorsed any alcohol or marijuana use during the past 6 months received the following items, respectively. Binge drinking was assessed by asking, “Over the last six months, on how many days did you drink five or more drinks in a row, that is, within a couple of hours?” Marijuana use was assessed by asking, “Over the last six months, how many times did you use marijuana?” Participants who reported no prevalent alcohol or marijuana use were coded 0 for those variables. Research has demonstrated that youth are generally reliable and valid sources for reporting their substance use behaviors (Johnston, O’Malley, Bachman, & Schulenberg, 2010).

Data Analytic Strategy

For the first hypothesis stating that female-identified and racial/ethnic minorities will experience higher rates of IPV than male-identified and White youth, a series of Chi-square analyses were conducted with categorical variables of IPV victimization (i.e., verbal, physical, any) and demographic variables (i.e., birth sex, gender identity, sexual orientation, race). For the second hypothesis stating that IPV will be concurrently associated with various health behaviors, a series of independent samples t-tests were conducted with victimization (i.e., verbal, physical, any) as categorical variables and depression, anxiety, binge drinking, marijuana use, condomless sex acts, and number of sexual partners as dimensional variables. For the third hypothesis, analyses were identical to the previous, except using 5-year follow-up scores (instead of 4-year follow-up scores) as dimensional variables. For this hypothesis, all participants who did not complete both 4- and 5-year follow-up were excluded from analysis (total included N = 162)

Results

Demographics Differences in IPV

Of those with a sexual and/or romantic partner at 4-year follow-up (N = 172), 63 (36.6%) reported experiencing verbal IPV, 32 (18.6%) reported experiencing physical IPV, 6 (3.5%) reported experiencing sexual IPV, and 71 (41.3%) reported experiencing any IPV. As hypothesized, participants who experienced verbal IPV were more likely to identify as female or male-to-female transgender (χ² = 10.13, p = .017) compared to male and female-to-male transgender, and they were more likely to identify as Black/African-American (χ² = 10.74, p = .013) compared other races. Participants who experienced physical IPV were more likely to identity as female or male-to-female transgender (χ² = 12.48, p = .006), Black/African-American (compared to White, Latino and other racial groups; χ² = 10.29, p = .016), and lesbian (compared to gay, bisexual and other identities; χ² = 8.36, p = .039). Participants who experienced any IPV were more likely to identity as female or male-to-female transgender (χ² = 13.03, p = .005) and Black/African-American (χ² = 9.20, p = .027). When gender was examined as the binary (i.e., male and FTM; female and MTF), female gender was associated with greater risk across verbal (χ² = 6.57, p = .015), physical (χ² = 10.63, p = .001), and any IPV (χ² = 8.52, p = .004). No differences in IPV were found for birth sex. Except for the finding that lesbian-identified young adults were at greater risk for physical IPV (compared to gay, bisexual and other identities), no other differences in IPV were found for self-identified sexual orientation. Given emerging research suggesting bisexual (and questioning) young people have poorer psychosocial outcomes compared to monosexual (gay and lesbian) young people, we conducted follow-up analyses to directly compare gay/lesbian (N = 101) and bisexual/questioning (N = 69) youth across all forms of IPV victimization (verbal, physical, any). No significant differences were found. Further, all findings are based on IPV at 4-year follow-up, though analysis of IPV at 5-year follow-up yielded similar demographic differences, except birth sex. These results are summarized in Table 2.

Table 2.

Percentage of young adults experiencing various types of partner violence by demographics

Variable		Any Verbal		χ2	p	V	Any Physical		χ2	p	V	Any Abuse		χ2	p	V
		N	%				N	%				N	%
Birth Sex	Male	22	30.6	1.97	.200	.107	9	12.5	3.05	.111	.133	25	34.7	2.20	.159	.113
	Female	41	41.0				23	23.0				46	46.0
Gender Identity	Male	17	27.4	10.13*	.017	.244	5	8.1	12.48**	.006	.271	19	30.6	13.03**	.005	.277
	Female	41	44.6				23	25.0				46	50.0
	MTF	5	55.6				4	44.4				6	66.7
	FTM	0	0.0				0	0.0				0	0.0
Gender Identity (Binary)	Male (+FTM)	18	25.7	6.57*	.015	.197	5	7.1	10.63**	.001	.250	20	28.6	8.52**	.004	.224
	Female (+MTF)	45	45.0				27	27.0				51	51.0
Sexual Orientation	Lesbian	23	46.0	2.47	.480	.121	16	32.0	8.36*	.039	.222	26	52.0	3.36	.340	.140
	Gay	17	33.3				7	13.7				20	39.2
	Bisexual	19	33.9				8	14.3				21	37.5
	Other	4	30.8				1	7.7				4	30.8
Race	Latino	3	15.0	10.74*	.013	.250	2	10.0	10.29*	.016	.245	4	20.0	9.20*	.027	.231
	Black	45	45.0				26	26.0				50	50.0
	White	4	17.4				0	0				6	26.1
	Other	11	37.9				4	13.8				11	37.9

Note:

^*p < .05,

^**p < .01;

% refers to percentage within demographic variable; V refers to effect size for Chi-square test (Cramer’s V)

The Relation between IPV and Health Behaviors

Verbal IPV at 4-year follow-up was not associated with concurrent anxiety symptoms, but was associated with higher levels of anxiety symptoms one year later (t = 2.04, p = .044). Verbal IPV was not associated with depression concurrently, but these variables were marginally associated one year later (t = 1.70, p = .092). Physical IPV at 4-year follow-up predicted higher levels of depression (t = 2.42, p = .020) one year later and was marginally associated with anxiety (t = 1.88, p = .069) one year later, but it was not concurrently associated with mental health symptoms. We observed no differences in binge drinking or marijuana use for any of the IPV variables either concurrently or one year later. Regarding sexual risk taking, verbal IPV was concurrently associated with a higher number of condomless sex acts (t = 2.29, p = .029), but it was not associated with condomless sex one year later. No differences in number of sexual partners were found at either baseline or one year follow-up. Note that for all prospective analyses, the same pattern of effects was observed when controlling for Year 4 health. Together, these findings provide mixed support for hypotheses 2 and 3. Results are summarized in Table 3.

Table 3.

Results of independent samples t-tests between various types of partner violence and health behaviors

Variable	Year	Any Verbal		t	p	d	Any Physical		t	p	d	Any Abuse		t	p	d
		Yes	No				Yes	No				Yes	No
Anxiety	4	.50 (.80)	.38 (.70)	0.96	.338	.160	.47 (.67)	.42 (.76)	0.34	.737	.070	.54 (.81)	.35 (.68)	1.62	.109	.254
	5	.69 (.83)	.44 (.64)	2.04*	.044	.337	.80 (.92)	.47 (.66)	1.88	.069	.412	.76 (.85)	.37 (.56)	3.31**	.001	.542
Depression	4	.62 (.82)	.48 (.76)	1.05	.296	.177	.64 (.78)	.51 (.79)	0.85	.401	.166	.67 (.83)	.44 (.74)	1.83	.069	.293
	5	.80 (.94)	.56 (.70)	1.70	.092	.290	.99 (.90)	.56 (.76)	2.42*	.020	.516	.87 (.93)	.48 (.65)	2.95**	.004	.486
Binge Drinking	4	2.20 (1.61)	2.53 (1.53)	−1.19	.235	−.210	2.12 (1.87)	2.47 (1.49)	−0.91	.369	−.207	2.15 (1.57)	2.59 (1.54)	−1.67	.098	−.283
	5	2.43 (1.72)	2.31 (1.50)	0.40	.691	.074	3.00 (2.06)	2.24 (1.46)	1.54	.138	.426	2.43 (1.74)	2.29 (1.46)	0.49	.627	.087
Marijuana Use	4	40.33 (60.50)	34.10 (59.13)	0.64	.527	.104	46.68 (63.99)	34.11 (58.51)	0.96	.344	.205	41.33 (61.27)	32.98 (58.39)	0.87	.386	.140
	5	48.41 (65.70)	39.04 (65.86)	0.86	.389	.142	45.24 (59.65)	41.86 (67.24)	0.27	.789	.053	50.09 (67.42)	37.16 (64.38)	1.21	.229	.196
Condomless Sex Acts┼	4	28.00 (47.78)	8.69 (12.57)	2.29*	.029	.553	34.29 (44.33)	11.82 (26.25)	1.85	.085	.617	26.87 (44.67)	7.86 (12.38)	2.57*	.014	.580
	5	21.38 (38.20)	15.03 (30.08)	0.83	.409	.185	32.07 (53.88)	14.75 (27.78)	1.18	.259	.404	23.00 (41.61)	13.28 (25.17)	1.30	.200	.283
Number of Sexual Partners	4	1.40 (.64)	1.41 (.78)	−0.15	.879	−.014	1.38 (.55)	1.41 (.77)	−0.34	.736	−.045	1.42 (.65)	1.40 (.79)	0.24	.814	.028
	5	1.35 (1.12)	1.37 (1.20)	−0.09	.928	−.017	1.13 (.94)	1.41 (1.21)	−1.38	.174	.258	1.35 (1.08)	1.36 (1.23)	−0.07	.944	−.009

Note:

^*p < .05,

^**p < .01,

^┼Female/female excluded; d refers to effect size for t-test (Cohen’s d)

Discussion

The present study aimed to examine the relation between IPV and health behaviors in a community sample of LGBT young adults. Several findings merit discussion. First, as hypothesized, IPV was experienced non-uniformly across particular demographic groups, specifically race and gender identity. Second, being a victim of IPV was associated with concurrent sexual risk taking and future mental health outcomes. Third, contrary to expectations, experiences of IPV were not associated with substance use.

Demographic differences in IPV, particularly those by race and gender, have remained debated topics in the empirical literature. Although findings from these studies are historically mixed, evidence generally finds that IPV, specifically victimization, affects women and racial/ethnic minorities disproportionately (Caetano, Cunradi, Schafer, & Clark, 2000; Rennison & Welchans, 2000; Tjaden & Thoennes, 2000). Our findings are consistent with this literature, even though previous research has used predominately heterosexual samples. That is, although nearly all participants from the present study endorsed a non-heterosexual orientation, similar demographic differences in IPV were still observed, such that female, male-to-female transgender, and Black/African-American youth were at greatest risk. It is interesting, too, that differences in IPV were observed for gender identity, but not birth sex, suggesting transgender participants are driving this finding. Indeed, it is striking that none of the female-to-male participants reported any IPV, while over half of the male-to-female participants reported some form of IPV. However, given the small sample of transgender youth, this explanation is merely speculative, though recent evidence suggests transgender youth are at particularly high risk for IPV victimization and perpetration (Dank et al., 2014). Future empirical work that closely examines differences in health behaviors by gender identity, and change in gender identity, appears warranted.

Various explanations have been proposed why disproportional rates of IPV exist amongst women and racial/ethnic minorities, including differing cultural views on acceptance of violence, patriarchy and sexism, and structural inequalities such as poverty, discrimination, and lower educational attainment (Dobash & Dobash, 1977; Rennison & Planty, 2003). Although many of these explanations have received varying degrees of empirical support, mounting evidence points to a more nuanced understanding of IPV, namely an intersectional perspective that highlights both the individual and structural contributions of race, class, gender, and sexual identity (Caetano, Schafer, & Cunradi, 2001; Sokoloff & Dupont, 2005). One such explanation that involves sexual minorities in particular is a minority stress model (Meyer, 2003), which posits that the relation between minority status and majority values often leads to conflict with the social environment (e.g., racism, sexism), resulting in increased stress and poor health outcomes (Meyer, 2003). Indeed, in addition to stressors associated with a sexual minority status, youth of color and female-identified individuals are perhaps at even greater risk for negative health outcomes due to cumulative stress from a combination of gender, racial, and/or sexual minority identities (Crawford, Allison, Zamboni, & Soto, 2002; Meyer, Dietrich, & Schwartz, 2008). In conjunction with this model, researchers have posited that certain general psychological processes (e.g., emotion dysregulation) are strong predictors of health behaviors, and that these processes are typically elevated in sexual minorities as a result of minority stress (Hatzenbuehler, 2009). With a strong literature demonstrating links between poor emotion regulation and IPV (Bouchard, Sabourin, Lussier, & Villeneuve, 2009; Kinsfogel & Grych, 2004), perhaps LGBT youth are at greater risk for victimization due to these maladaptive social, emotional, and cognitive processes.

The finding that youth who experienced IPV are at greater risk for future depression and anxiety (though the association with anxiety was marginally significant) is congruent with a large literature demonstrating strong links between IPV and internalizing psychopathology (Ellsberg, Jansen, Heise, Watts, & Garcia-Moreno, 2008). It is striking, however, that there were no associations between IPV and mental health at the cross-sectional level, a finding that is inconsistent with previous research from heterosexual samples (Callahan et al., 2003; Coker et al., 2000; Silverman et al., 2001). One explanation for this null finding is that rather than co-occurring, perhaps mental health problems develop over time as a result of IPV. It is possible that the dissolution of a violent romantic or sexual partnership and/or ongoing IPV may lead to a delayed onset of internalizing psychopathology, a pattern observed with other psychiatric disorders. Indeed, the very criteria of post-traumatic stress disorder (PTSD) requires at least one month between the stressor and onset of symptoms (American Psychiatric Association, 2013). However, while there certainly is overlap in etiology, course, and symptom presentation between depression, anxiety, and PTSD, these disorders are undoubtedly distinct. Future longitudinal research with frequent, comprehensive measurement of IPV is needed to unpack whether depression and anxiety are causes, consequences, or both.

Why IPV was associated with sexual risk taking at baseline, but not one year follow-up, warrants further study. One possible explanation is that individuals in violent relationships may have difficulty negotiating condom use while those relationships are ongoing, due to likely power differentials between partners, but these effects may not generalize to future relationships that are not characterized by violence. Previous studies have generally supported this claim at the cross-sectional level, with research consistently linking partner abuse victimization–both physical and psychological–with condom nonuse, fear of consequences negotiating condom use, and/or inconsistent condom use (He, McCoy, Stevens, & Stark, 1998; Silverman, Raj, & Clements, 2004; Teitelman, Dichter, Cederbaum, & Campbell, 2007). Evidence also suggests that severe violence in relationships tends to persist whereas moderate acts of violence tend to remit (Caetano, Field, Ramisetty-Mikler, & McGrath, 2005), and many individuals in violent relationships or marriages do end up leaving their partner (Fleury, Sullivan, & Bybee, 2000; Kurz, 1996). Perhaps IPV was not associated with future sexual risk taking because participants in violent relationships either left their partner or violence remitted. Given the significant lack of empirical literature on IPV revictimization and associated outcomes, additional longitudinal research is desperately needed in order to understand both the immediate and delayed consequences of IPV.

Finally, it is interesting that substance use was not associated with IPV in the present study, particularly because a strong literature has demonstrated a robust link between alcohol/drug use and conflict, aggression, and partner violence across both adults (Foran and O’Leary, 2008) and adolescents (Rothman et al., 2012; Temple, Shorey, Fite, Stuart, & Le, 2013). One possibility is evidence demonstrating higher rates of substance use among LGBT youth and young adults compared to their heterosexual counterparts overall (Marshal et al., 2008; Newcomb, Birkett, Corliss, & Mustanski, 2014; Thiede et al., 2003). Studies have referenced the “bar culture” being more common in LGBT populations, with research suggesting sexual minorities may have different drinking patterns, more permissive social norms, and positive expectancies for alcohol use (Hatzenbuehler, Corbin, & Fromme, 2008; Heffernan, 1998). It is possible that less variability coupled with higher baseline rates of substance use, likely driven by minority stressors, limits detectability of differences, which may explain this null finding. Finally, perhaps the current theoretical model of IPV, which has been formulated from decades of studies using predominately heterosexual samples, does not map onto the dating experiences of LGBT populations.

Limitations

As with all research, these findings should be interpreted in light of several limitations. First, questions about IPV assessed only the frequency of abusive behaviors, not the context. Given the importance of situational factors (e.g., presence of others, location, availability of weapon, emotional distress, intoxication), it is unknown where and when young people are at greatest risk of being victimized by a romantic partner. Subsequent research should investigate not only the influence of the immediate environment on IPV, but also of the environment at multiple ecological levels. Second, only experiences of IPV victimization were measured. Because bi-directionality is more often the rule than the exception in violent relationships (Caetano, Ramisetty-Mikler, & Field, 2005; Langhinrichsen-Rohling et al., 2012), future research should also assess for IPV perpetration. Finally, the present study used a sample of convenience that had low rates of IPV in certain groups. Future studies should strive for a larger, non-convenience sample of sexual minority youth in order to increase generalizability and confirm these findings.

Research Implications

Despite these limitations, findings from the present study are strengthened in a number of important ways. First, a large, diverse sample across race, sexual orientation, and gender identity improves the generalizability of our findings, as well as contributes to the limited literature on intersectionality of multiple identities (Cole, 2009). Second, the study differentiated between non-physically injurious (i.e., verbal) and physical types of violence. Third, the effects of IPV on health behaviors was examined both cross-sectionally and longitudinally, allowing for a more fine-tuned examination of onset and course of mental health symptoms and risk behaviors. Future studies could benefit from more rigorous longitudinal analyses that investigate how IPV impacts negative outcomes and revictimization within participants. Subsequent research should also assess situational factors in order to better understand where and when violence occurs, as well as include measures of IPV perpetration. In addition, although the present study found few differences in IPV victimization between various sexual orientations, researchers should continue to be mindful of the heterogeneity that exists within the LGBT population and examine whether these differences in sexual identity are associated with better (or worse) health outcomes. Broadly, increased empirical attention examining the link between IPV and health behaviors is needed to clarify why more traditional risk factors found in heterosexual youth (e.g., substance use) may not generalize to LGBT populations.

Clinical and Policy Implications

Findings from the present study suggest IPV is associated with particular demographics and short and long-term health behaviors in LGBT young adults. Interventions targeting IPV in LGBT young people may benefit from programs that are tailored to address the specific needs of this community, while also being mindful of the heterogeneity within this population, such that individuals with multiple gender, racial, and/or sexual minority identities may be at even greater risk for IPV. Also important is that, until recently, the overwhelming majority of empirical studies addressing IPV in youth have utilized predominately heterosexual samples. Clinicians and policy makers should therefore be aware that risk factors proven to be tried-and-true in heterosexual youth may not readily translate to LGBT youth, as also evidenced in the present study. Taken together, given high prevalence, serious outcomes, and limited existing empirical data, future research investigating the correlates and consequences of IPV in this population is warranted.