Intimate Partner Violence Perpetration and Victimization among Young Adult Sexual Minorities in a Nationally Representative US Sample

Abstract

Context:

Despite health risks associated with minority stress, few studies have examined intimate partner violence (IPV) among sexual minorities; those that do have significant limitations. This study addresses key limitations to generate victimization and perpetration estimates in a national sample of young adults (ages 24–32).

Methods:

Data are from Wave IV of the National Longitudinal Study of Adolescent to Adult Health (n=13,668). Sexual identity categories were: 100% heterosexual (94% of males, 80% of females), mostly heterosexual (4% males, 16% females), bisexual (1% males, 2% females), mostly homosexual, and 100% homosexual (2% males, 2% females – mostly and 100% combined). IPV outcomes included physical violence (slapping, hitting, or kicking); threatened or mild violence (threats, pushing or shoving); and forced sex. Multivariate logistic regression, stratified by biological sex, was used to estimate odds of IPV victimization and perpetration.

Results:

Compared to same-sex heterosexual peers, adjusted odds of perpetrating (1.9) and experiencing (1.6) physical IPV were higher for mostly heterosexual females and mostly heterosexual males (3.1, 1.8 respectively). Mostly heterosexual females also had higher odds of experiencing threatened violence (1.6) and forced sex (1.6), and threatening violence (2.0). Mostly heterosexual males had higher odds of perpetrating forced sex (2.0) and threatening violence (1.8). Bisexual males had higher odds of forced sex victimization (4.9) and perpetration (5.0), and physical violence victimization (3.3).

Conclusions.

National prevalence estimates indicate heterogeneity in IPV perpetration and victimization by sexual identity and biological sex. Heterogeneity has important policy and prevention implications that require further study of stressor mechanisms.

Introduction

Intimate partner violence (IPV), which includes physical, psychological, and sexual violence towards a romantic partner, is a prevalent public health issue both globally and in the United States (US). IPV has well-documented associations with reproductive and sexual health indicators such as sexual risk taking, unplanned pregnancy or abortion, and sexually transmitted infections (1, 2). However, relatively few studies have examined intimate partner violence, especially perpetration of IPV, among sexual minorities. Recent studies using US data from both national surveillance and convenience samples suggest that rates of IPV among sexual minority individuals are greater than those of their heterosexual peers (3). Unfortunately, however, the prevalence estimates of IPV perpetration and victimization among sexual minorities vary widely, ranging from 11% to 40% (4–6). Such differences are likely a function of variations in sampling, violence definitions, or time periods examined. Regarding sampling, most studies that have focused on IPV among sexual minorities have used convenience samples, which limits their generalizability. Researchers also often collapse across categories of sexual orientation (i.e., “sexual minority” vs. “gay,” “lesbian,” and “bisexual”), which does not capture the continuum of orientation nor the likely heterogeneity of experiences among members of this population (3). For instance, one of the largest national studies on IPV among sexual minorities, the 2010 National Intimate Partner and Sexual Violence Survey, analyzed patterns of IPV victimization among three categories: gay/lesbian, bisexual, and heterosexuals (7). However, emerging research supports the existence of “in-between” sexualities, such as mostly straight or mostly homosexual, and the need to analyze populations using a continuum-based perspective of sexual orientation (8).

Although there has been a growing focus on IPV among sexual minorities in the public health literature, most of this work focuses on victimization and neglects violence perpetration. For victimization, a study using the population-based National Violence Against Women Survey found that non-heterosexuals had higher odds of experiencing IPV victimization than heterosexuals in an intimate relationship but, because of data limitations, did not examine perpetration (9). In another analysis using the same data, Messinger found that respondents with a history of same-sex cohabiting relationships were more likely to experience verbal, psychological, physical, and sexual IPV victimization (10). Using the National Epidemiologic Survey on Alcohol and Related Conditions, Roberts et al. documented that lesbians and gay men, bisexuals, and heterosexuals who reported any same-sex partners over their lifetime had greater odds of interpersonal violence including domestic violence victimization (11). One of the few studies that has reported same-sex IPV perpetration found that 29% to 57% of lesbian and gay college students recounted physical and sexual dating violence perpetration within the past 12 months (12). Unfortunately, no such studies on IPV perpetration have been conducted in a population-based sample. Such research highlights the critical need for more population-based studies of sexual minority IPV victimization and perpetration to identify targets for public health prevention and intervention.

Lastly, the majority of existing population-based research studies on IPV among sexual minorities focuses on younger age groups. Using data from Wave II of the National Longitudinal Study of Adolescent to Adult Health (Add Health), Halpern et al. found that 23.3% of adolescents in same-sex relationships reported some type of partner violence victimization in the previous 18 months (5). Perpetration was not measured at Wave II. No such studies have been conducted with young adult sexual minorities, who are more likely than adolescents to be in established, longer-term romantic relationships (13). Accordingly, this paper builds on prior research by Halpern et al. and the larger literature by describing associations between self-identified sexual minority status and three different types of IPV perpetration and victimization using young adult data from Add Health.

Prior research suggests that these and similar findings documenting the sexual minority IPV disparity may be explained using the minority stress model. Although a number of factors are associated with partner violence across people of all orientations, the sexual minority population may experience additional factors related to sexual minority status (14). Minority stress theory posits that individuals who identify as sexual minorities or identify outside of the heterosexual paradigm experience minority stress, or stress resulting from experienced and internalized stigma (15). For example, general sexual orientation related victimization and discrimination, internalized homonegativity, and identity concealment have been reported as minority stressors (16, 17). These types of unique psychosocial stressors may breed or exacerbate partner violence or encourage victims to remain silent out of fear of perceived discrimination among sources of help (17). In addition, an emerging body of literature that examines the sexual continuum suggests that all nonexclusive groups (including mostly heterosexuals, bisexuals, and mostly homosexuals) may be at elevated risk of negative health outcomes compared to the two exclusive orientation groups (heterosexuals and homosexuals) (18–20). In many studies, bisexuals have been shown to be particularly vulnerable to violence (4, 7, 21, 22). “In-between” sexual minorities may be at similar elevated risk and may experience fewer protective factors such as group social support and psychosocial benefits of identifying with an exclusive sexual orientation group (23). Accordingly, we hypothesized that identifying as any type of sexual minority would be associated with increased odds of both IPV victimization and perpetration in our sample of young adults. Because our analyses report IPV in the current or most recent relationship, our prevalence estimates should be lower than other studies that report lifetime violence estimates.

Methods:

Sample

Data are from Add Health, a large, longitudinal survey of a nationally representative sample of in-school adolescents (7^th-12^th grades) during the 1994–1995 school year. At the time of this analysis four complete waves of data had been collected and released, the most recent of which was completed in 2008–2009 (Wave IV; n=15,701, ages 24–32; response rate=80.3%)(24)¹. In Wave IV, 92.5% of the original sample were eligible for follow up. The remaining participants had died or were incarcerated during the interview period. Of those eligible, 80.3% were interviewed. We restricted our analytic sample to a subset of respondents who: were re-interviewed at Wave IV (n=15,701), reported a current or recent romantic partnership (n=15,315), and had valid sampling weights (14,800). Among these eligibles, 13,653 respondents had complete data on all variables of interest, reflecting an 8% sample loss from eligible participants.

Measures

Outcomes:

Intimate Partner Violence Victimization and Perpetration

Information about IPV victimization and perpetration in a current or most recent relationship was collected at Wave IV. For each of the outcomes described below, response choices included never, happened before last year, once in the last year of the relationship, twice, 3 to 5 times, 6 to 10, 11 to 20, or over 20 times. For our analyses, we collapsed the response categories to create two dichotomous variables (yes, no), one each for victimization and perpetration. Specifically, a response of “never” was coded as “no,” while a report of any violence was coded as “yes.” Therefore a “yes” response may reflect past year violence or earlier violence for multiyear relationships.

Physical and Injury:

Physical violence outcomes were measured with items that asked how often a romantic partner had slapped, hit, or kicked the respondent (victimization) and how often the respondent had slapped, hit, or kicked their partner (perpetration). Among those who reported physical violence, injury was measured with items that asked how often the respondent (victimization) or the respondent’s partner (perpetration) had an injury, such as a sprain, bruise, or cut because of a fight with a romantic partner.

Threatened or Mild Physical Violence:

One item asked about both threats and/or mild physical dating violence. Specifically, the item asked how often the respondent (perpetration) or the respondent’s romantic partner (victimization) had threatened violence, pushed or shoved, or thrown something at their romantic partner. For brevity, we label the variable as “threatened.”

Forced Sex:

Forced sex was measured using items that asked how often the respondent (perpetration) or the respondent’s romantic partner (victimization) had insisted on or made the other have sexual relations when they did not want to.

Main Predictors

Sexual orientation:

Sexual orientation identity was self-reported at Wave IV. Specifically, respondents were asked to choose the description that best fit how they thought of themselves from the following: 100% heterosexual, mostly heterosexual, bisexual, mostly homosexual, and 100% homosexual. Given sparse sample sizes for “mostly homosexual” and since the regression results did not differ between the mostly homosexual and 100% homosexual groups, we collapsed these to increase statistical power.

Biological sex:

Biological sex of the respondent was taken from adolescent school records and confirmed by the interviewer at Wave IV.

Covariates

Age:

Age was calculated by subtracting the respondent’s birth date from the Wave IV interview date.

Race/ethnicity:

Race/ethnicity (Hispanic, non-Hispanic [NH] Black, NH White, and NH Other) was self-reported at Wave I.

Parent education:

Parent education is a proxy for socioeconomic status (SES) during adolescence, and was documented as the highest level of education achieved by either parent at the time of the Wave I interview. Categories included less than high school, graduated high school or received GED, some college, and college graduate and beyond a 4-year degree.

Relationship characteristics:

To account for the context of the romantic relationship, we also controlled for relationship type (married, cohabiting, dating, pregnancy partner², or most recent³ partner), relationship duration in months, and biological sex of the partner (other or same sex).

Analysis:

Previous research indicates dimensions of sexual orientation may have differential health effects by sex (25). In an effort to consider heterogeneity in health risks and outcomes, all analyses were stratified by biological sex. Sample characteristics were assessed using weighted percentages. This is reported in Table 1 and in the text below. Associations between sexual orientation and violence victimization and perpetration were tested using logistic regression models. The unadjusted model for each outcome included sexual orientation as the only predictor, and the adjusted model included sexual orientation and all other covariates. All analyses used sampling weights and adjusted variance estimates to account for the Add Health complex survey design and were conducted using Stata 14.0.

Table 1.

Descriptive statistics by biological sex and sexual orientation

	Males					Females
Weighted %	100% Heterosexual (N=5,947)	Mostly Heterosexual (N=208)	Bisexual (N=41)	Homosexual (N=166)	Total (N=6,362)	100% Heterosexual (N=5,850)	Mostly Heterosexual (N=1,149)	Bisexual (N=164)	Homosexual (N=128)	Total (N=7,291)
% in Category	93.6	3.5	0.6	2.3	100	79.9	16.2	2.3	1.6	100
Mean Age	28.4	28.4	28.4	28.5	28.4	28.3	27.9	27.7	28.0	28.2
Race
White	67.2	73.2	87.1	59.3	67.4	65.7	72.8	64.2	64.8	66.8
Black	14.6	7.3	5.3	14.1	14.3	17.0	8.8	16.2	18.2	15.7
Hispanic	11.4	12.9	5.8	19.6	11.6	11.6	10.5	12.7	13.4	11.5
Other	6.8	6.5	1.9	7.0	6.7	5.8	7.9	6.9	3.6	6.1
Parent Education
Less than HS	11.7	7.9	3.7	15.2	11.6	12.1	9.1	18.9	11.2	11.8
HS/GED	26.5	20.7	41.8	21.7	26.3	28.8	23.8	37.2	29.2	28.2
Some college	31.3	24.9	20.7	25.3	30.8	28.7	32.9	19.2	26.1	29.1
College graduate	30.6	46.4	33.8	37.8	31.3	30.4	34.2	24.7	33.6	31.0
Relationship Type
Marriage	42.3	29.7	27.2	3.0	40.8	51.7	42.7	32.7	13.0	49.2
Cohabitating	28.4	25.4	29.2	43.0	28.6	24.0	34.4	43.5	51.2	26.5
Pregnant	1.7	1.9	0	0	1.7	2.8	2.8	5.2	1.1	2.8
Dating	17.8	23.8	10.7	20.9	18.0	14.2	12.8	13.0	23.1	14.1
Most Recent	9.9	19.3	32.9	33.1	10.9	7.4	7.3	5.6	11.6	7.4
Mean Relationship
Duration (months)	53.0	42.0	40.5	20.7	51.8	66.7	58.7	54.1	36.7	64.6
Same Sex Partner	0.2	0	6.5	88.4	2.3	0.5	1.1	1.1	79.0	2.1
Violence Perpetration
Physical	4.9	10.3	11.8	6.5	5.2	10.8	18.9	15.2	8.0	12.2
Injury	2.5	5.2	5.1	3.8	2.6	2.9	5.0	8.8	2.6	3.4
Threatened	10.0	12.7	18.5	9.7	10.1	13.7	23.5	17.2	9.7	15.3
Forced Sex	4.9	8.6	17.1	6.5	5.1	1.4	1.8	2.6	2.7	1.5
Violence Victimization
Physical	17.3	23.0	34.8	12.4	17.5	7.4	10.6	11.8	5.7	8.0
Injury	4.9	7.1	9.4	8.6	5.1	4.4	6.6	9.4	3.4	4.9
Threatened	23.9	20.8	36.8	15.3	23.7	16.9	23.2	22.9	15.4	18.0
Forced Sex	6.2	9.6	24.8	5.6	6.4	5.0	7.1	5.4	4.5	5.3

Notes: Percentages are weighted to yield national probability estimates

Results:

Descriptive statistics for the study population by biological sex and sexual orientation are presented in Table 1. The sample was almost evenly split between males and females. Among males, 94% identified as 100% heterosexual, 4% as mostly heterosexual, 1% as bisexual, and 2% as homosexual. Among females, 80% identified as 100% heterosexual, 16% as mostly heterosexual, 2% as bisexual, and 2% as homosexual. The average age was 28.4 for males and 28.2 for females. A majority of the sample’s males and females were White (67% and 67%, respectively). 14% of males and 16% of females were Black, while 12% of males and 12% of females were Hispanic. A small portion of the sample, 7% of males and 6% percent of females, described their race/ethnicity as “Other.” Over 60% had parents who completed at least some college. The participants described their current or most recent relationship (if not in a relationship at the time of the survey). Of the total sample of males and females, between 40–50% reported on a married relationship, approximately 28% reported about a cohabiting relationship, 14–18% reported about a dating partner, 7–11% reported on a recent partner, and 1–3% reported on a relationship with a pregnancy partner.⁴ Roughly 90% of respondents reported on a current rather than a recent partner. Just over two percent of both males and females reported on a relationship with a same-sex partner. The average relationship duration was 51.8 months for males and 64.6 months for females.

The prevalence of victimization was higher than perpetration for males. Overall, 5%, 10% and 5% of males reported physical, threatened, and forced sex perpetration, respectively, and 18%, 24%, and 6% of males reported these types of victimization. Among females, prevalence of perpetration and victimization were more similar and somewhat higher than for males. Twelve percent, 15% and 2% of females reported physical, threatened, and forced sex perpetration, respectively, and 8%, 18%, and 5% of females reported physical, threatened, and forced sex victimization.

Among sexual minority males, prevalence rates of physical, threatened violence, and forced sex perpetration outcomes were typically higher among mostly heterosexual and bisexual males compared to the other groups. Physical and forced sex victimization were also higher. A few differences were notable in size. For example, 35% of bisexual males had experienced physical violence victimization compared to 12% of homosexuals, 17% of 100% heterosexuals, and 23% of mostly heterosexual males. Among sexual minority females, physical and threatened violence perpetration were higher for mostly heterosexual and bisexual females. Patterns were similar for these types of victimization. Forced sex perpetration was elevated for both bisexual and homosexual females; forced sex victimization was more similar across groups.

Logistic Regression Models

Table 2 presents the results for physical violence victimization and perpetration for both males and females. Among males, in adjusted models mostly heterosexual men had 1.8 times greater odds of being the victim of and 3.1 times the odds of perpetrating physical dating violence compared to the odds of the 100% heterosexual male group (referent). Although they did not differ significantly in perpetration, bisexual males had 3.3 times greater odds of being the victim of physical violence compared to the odds of their 100% heterosexual male peers. Mostly heterosexual females followed a pattern similar to their male counterparts; they had 1.6 times the odds of being the victim of and 1.9 times the odds of perpetrating physical violence compared to 100% heterosexual females. No differences in physical dating violence victimization or perpetration emerged for bisexual females, nor for the male and female homosexual groups compared to 100% heterosexuals.

Table 2:

Results of bivariate and multivariate logistic regression models examining associations between sexual minority identification and physical dating violence, stratified by biological sex

	Males (N=6,362)				Females (N=7,291)
aOR (95% CI)	Victimization		Perpetration		Victimization		Perpetration
	Unadjusted	Adjusted	Unadjusted	Adjusted	Unadjusted	Adjusted	Unadjusted	Adjusted
Sexual Orientation
100% Heterosexual	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00
Mostly Heterosexual	1.43 (0.90–2.26)	1.77 (1.11–2.80)*	2.23 (1.13–4.41)*	3.10 (1.53–6.24) **	1.48 (1.14–1.94)**	1.55 (1.17–2.06)**	1.92 (1.52–2.43)***	1.92 (1.49–2.47)***
Bisexual	2.55 (1.03–6.33)*	3.32 (1.08–10.22)*	2.58 (0.79–8.44)	3.82 (0.96–15.20)	1.67 (0.93–3.00)	1.41 (0.69–2.88)	1.48 (0.86–2.56)	1.38 (0.77–2.47)
Homosexual	0.68 (0.33–1.39)	0.64 (0.21–1.93)	1.34 (0.53–3.36)	2.15 (0.44–10.45)	0.75 (0.38–1.50)	0.39 (0.11–1.38)	0.72 (0.32–1.63)	0.88 (0.30–2.60)
Age		0.93 (0.88–0.99)*		0.95 (0.88–1.04)		1.03 (0.95–1.12)		0.96 (0.90–1.02)
Race
NH White		1.00		1.00		1.00		1.00
Hispanic		1.34 (1.00–1.81)		1.34 (0.87–2.06)		1.27 (0.81–1.98)		1.58 (1.19–2.10)**
NH Black		1.87 (1.35–2.60)***		2.52 (1.72–3.70)***		1.92 (1.41–2.61)***		1.67 (1.30–2.14)***
NH Other		1.12 (0.80–1.58)		1.15 (0.61–2.16)		2.03 (1.25–3.30)**		1.68 (1.12–2.51)*
Parent Education (SES)
College Grad		1.00		1.00		1.00		1.00
Less than HS		0.75 (0.55–1.00)		1.27 (0.69–2.34)		2.24 (1.50–3.30)***		1.23 (0.87–1.74)
HS grad/GED		0.90 (0.72–1.13)		0.88 (0.59–1.31)		1.84 (1.36–2.50)***		1.22 (0.94–1.58)
Some college		0.89 (0.70–1.14)		0.86 (0.57–1.31)		2.01 (1.46–2.77)***		1.24 (0.95–1.62)
Relationship Duration (months)		1.00 (1.00–1.01)*		1.01 (1.00–1.01)*		1.00 (1.00–1.01)		1.00 (1.00–1.01)*
Relationship Type
Married		1.00		1.00		1.00		1.00
Cohabiting		1.47 (1.14–1.90)**		1.40 (0.92–2.13)		1.74 (1.27–2.38)**		1.72 (1.34–2.19)***
Pregnancy partner		0.87 (0.45–1.69)		1.06 (0.32–3.47)		1.69 (0.94–3.06)		0.77 (0.43–1.38)
Dating		0.46 (0.31–0.68)***		0.31 (0.13–0.74)**		0.30 (0.18–0.52)***		0.54 (0.36–0.80)**
Most Recent partner		0.41 (0.28–0.60)***		0.34 (0.15–0.78)*		0.49 (0.25–0.99)*		0.39 (0.20–0.77)**
Sex of Partner
Opposite Sex		1.00		1.00		1.00		1.00
Same Sex		1.39 (0.38–5.04)		0.89 (0.20–4.04)		2.18 (0.69–6.86)		0.71 (0.27–1.92)

^*p < 0.05

^**p < 0.01

^***p < 0.001

The results for threatened violence victimization and perpetration are shown in Table 3. Sexual minority males did not differ from 100% heterosexual males for threatened violence victimization; however, mostly heterosexual males had 1.8 times the odds of perpetrating threatened violence compared to the odds of heterosexual males, similar to the odds of actual violence perpetration. Bisexual and homosexual males showed no statistically significant differences in threatened violence perpetration compared to heterosexual males. Among females, mostly heterosexuals reported 1.6 times the odds of experiencing and 2.0 times the odds of perpetrating threatened violence compared to the odds of 100% heterosexual females. Bisexual and homosexual females showed no statistically significant differences in threatened dating violence perpetration compared to heterosexual females.

Table 3:

Results of bivariate and multivariate logistic regression models examining associations between sexual minority identification and mild or threatened dating violence, stratified by biological sex

	Males (N=6,362)				Females (N=7,291)
aOR (95% CI)	Victimization		Perpetration		Victimization		Perpetration
	Unadjusted	Adjusted	Unadjusted	Adjusted	Unadjusted	Adjusted	Unadjusted	Adjusted
Sexual Orientation
100% heterosexual	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00
Mostly heterosexual	0.84 (0.50–1.40)	1.04 (0.62–1.75)	1.31 (0.79–2.19)	1.78 (1.06– 2.97)*	1.49 (1.20–1.85)***	1.61 (1.29– 2.00)***	1.94 (1.57–2.40)***	1.95 (1.56–2.44)***
Bisexual	1.85 (0.79–4.32)	2.34 (0.81–6.75)	2.06 (0.71–5.97)	3.27 (0.99–10.75)	1.46 (0.92–2.31)	1.33 (0.80–2.20)	1.31 (0.82–2.10)	1.19 (0.7–2.00)
Homosexual	0.57 (0.30–1.12)	0.40 (0.13–1.28)	0.97 (0.45–2.09)	1.83 (0.56–5.92)	0.90 (0.53–1.53)	0.70 (0.28–1.73)	0.68 (0.33–1.40)	0.71 (0.21–2.37)
Age		0.95 (0.91–1.00)*		0.97 (0.91–1.03)		1.00 (0.95–1.05)		0.92 (0.87–0.97)**
Race
NH White		1.00		1.00		1.00		1.00
Hispanic		1.30 (0.95–1.79)		1.23 (0.87–1.73)		1.27 (0.98–1.65)		1.5 (1.15–1.97)**
NH Black		1.90 (1.55–2.33)***		1.87 (1.40–2.49)***		1.98 (1.59–2.47)***		1.74 (1.37– 2.2)***
NH Other		1.21 (0.85–1.74)		1.35 (0.75–2.42)		1.25 (0.87–1.80)		1.75 (1.29–2.39)***
Parent Education (SES)
College Grad		1.00		1.00		1.00		1.00
Less than HS		1.15 (0.85–1.56)		1.09 (0.66–1.80)		2.23 (1.60–3.10)***		1.08 (0.72–1.60)
HS grad/GED		1.25 (1.03–1.51)*		0.80 (0.58–1.11)		1.72 (1.35–2.19)***		1.5 (1.20–1.86)***
Some college		1.08 (0.85–1.36)		1.09 (0.81–1.47)		1.81(1.43–2.29)***		1.43 (1.12–1.82)**
Relationship Duration (months)		1.01 (1.00–1.01)***		1.01 (1.00–1.01)***		1.00 (1.00– 1.01)**		1.00 (1.00–1.01)**
Relationship Type
Married		1.00		1.00		1.00		1.00
Cohabiting		1.58 (1.29–1.94)***		1.46 (1.10–1.93)**		1.43 (1.16–1.77)**		1.64 (1.30–2.05)***
Pregnancy partner		1.15 (0.57–2.30)		1.58 (0.68–3.70)		1.13 (0.71–1.80)		0.94 (0.59–1.50)
Dating		0.57 (0.42–0.79)**		0.29 (0.17–0.47)***		0.35 (0.24–0.51)***		0.45 (0.30–0.67)***
Recent dating		0.52 (0.37–0.74)***		0.17 (0.08–0.35)***		0.40 (0.24–0.67)**		0.32 (0.18–0.56)***
partner
Sex of Partner
Opposite		1.00		1.00		1.00		1.00
Same Sex		1.91 (0.67–5.47)		0.81 (0.27–2.42)		1.45 (0.61–3.45)		0.92 (0.31–2.76)

^*p < 0.05

^**p < 0.01

^***p < 0.001

Table 4 displays results for forced sex victimization and perpetration. Bisexual males had 4.9 times the odds of being the victim and 5.0 times the odds of perpetrating forced sex compared to heterosexual males. Although mostly heterosexual males did not differ from 100% heterosexuals in experiencing forced sex, they had 2.0 times higher odds of perpetrating forced sex compared to the odds of 100% heterosexual males. Homosexual males were not statistically different from heterosexual males in their experiences of forced sex victimization or perpetration. Among females, mostly heterosexuals had 1.6 times the odds of experiencing forced sex compared to the odds of 100% heterosexual females. Bisexual and homosexual females did not statistically differ in reported victimization compared to the 100% heterosexual female group. For perpetration, there were no statistically significant differences between any of the sexual minority identification categories compared to the 100% heterosexual female reference group. In addition, for all forms of IPV, mostly heterosexual, bisexual, and homosexual groups had overlapping confidence intervals for both males and females. This suggests that the differences among sexual minority groups may not be statistically significant from each other.

Table 4:

Results of bivariate and multivariate logistic regression models examining associations between sexual minority identification and forced sex, stratified by biological sex

	Males (N=6,362)				Females (N=7,291)
aOR (95% CI)	Victimization		Perpetration		Victimization		Perpetration
	Unadjusted	Adjusted	Unadjusted	Adjusted	Unadjusted	Adjusted	Unadjusted	Adjusted
Sexual Orientation
100% heterosexual	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00
Mostly heterosexual	1.60 (0.86– 2.97)	1.54 (0.80–2.98)	1.84 (1.06–3.19)*	2.03 (1.16–3.56)*	1.45 (1.11–1.90)**	1.59 (1.19–2.13)**	1.36 (0.73, 2.54)	1.43 (0.74– 2.78)
Bisexual	4.98 (1.84–13.50)**	4.94 (1.52–16.10)**	4.04 (1.36–12.06)*	4.96 (1.55–15.83)**	1.08 (0.48–2.41)	1.03 (0.42,–2.48)	1.95 (0.42, 9.16)	1.79 (0.34–9.44)
Homosexual	0.90 (0.36–2.24)	0.29 (0.04–2.16)	1.36 (0.56–3.33)	1.37 (0.68–2.77)	0.89 (0.31–2.53)	0.80 (0.12–5.40)	1.99 (0.44, 8.92)	0.85 (0.07–10.00)
Age		1.01 (0.93–1.10)		1.01 (0.92–1.11)		1.03 (0.95–1.12)		1.02 (0.87–1.18)
Race
NH White		1.00		1.00		1.00		1.00
Hispanic		1.67 (1.08–2.56)*		1.61 (1.04–2.47)*		2.03 (1.33–3.12)**		3.43 (1.65–7.15)**
NH Black		1.88 (1.33–2.67)***		1.86 (1.26–2.74)**		2.00 (1.41–2.84)***		2.57 (1.26–5.23)*
NH Other		1.53 (0.93–2.51)		2.23 (1.43–3.48)***		2.06 (1.26–3.35)**		4.50 (1.91–10.60)**
Parent Education (SES)
College Grad		1.00		1.00		1.00		1.00
Less than HS		1.35 (0.84–2.17)		1.32 (0.81–2.17)		1.49 (0.92–2.41)		0.88 (0.34–2.25)
HS grad/GED		0.98 (0.68–1.40)		1.08 (0.67–1.74)		1.15 (0.80–1.65)		0.67 (0.32–1.42)
Some college		0.74 (0.50–1.10)		0.94 (0.62–1.43)		1.00 (0.72–1.40)		1.20 (0.62–2.32)
Relationship Duration (months)		1.00 (0.99–1.00)*		1.00 (0.99–1.01)		1.00 (1.00–1.01)**		1.01 (1.00–1.01)
Relationship Type
Married		1.00		1.00		1.00		1.00
Cohabiting		1.78 (1.2–2.65)**		1.21 (0.82–1.79)		1.43 (1.01–2.02)*		1.53 (0.70–3.33)
Pregnancy partner		2.76 (1.16– 6.56)*		2.25 (0.92–5.49)		1.79 (0.88–3.61)		2.76 (0.62–12.28)
Dating		1.22 (0.72–2.08)		0.72 (0.38–1.39)		0.37 (0.19–0.69)**		1.42 (0.47–4.31)
Recent dating		1.54 (0.90–2.63)		0.95 (0.49–1.83)		2.00 (1.16–3.48)*		3.60 (1.10–11.70)*
partner
Sex of Partner
Opposite Sex		1.00		1.00		1.00		1.00
Same Sex		2.41 (0.33–17.54)		1.02 (0.49–2.12)		1.24 (0.22–7.00)		3.17 (0.41–24.7)

^*p < 0.05

^**p < 0.01

^***p < 0.001

IPV outcomes also varied by race/ethnicity and socioeconomic status. Specific results are shown in Tables 2 - 4. Among males, identifying as NH Black was associated with higher odds of physical and threatened violence perpetration and victimization compared to NH White males. Virtually all race/ethnic groups had higher odds of forced sex perpetration and victimization compared to NH Whites; the exception was forced sex victimization for NH Other. Among females, most race/ethnic groups had elevated odds of perpetration and victimization for all types of violence. The exception was threatened violence victimization, for which only NH Black females showed higher odds that NH Whites.

The highest level of parental education was used as a proxy for childhood socioeconomic status. Among males, there was little relationship with IPV. However, among females, lower parental education was linked to higher odds of physical victimization, and to threatened violence victimization and perpetration.

Regarding relationship type, odds of violence in cohabiting relationships were generally higher than in marriages, and dating and most recent relationships tended to have lower odds. There were a couple of exceptions to this pattern. Among males, odds of forced sex victimization were higher among both cohabiting and pregnancy partner relationships. Among females, those reporting on relationships with most recent partners had elevated odds of forced sex victimization and perpetration compared to current marriages.

Discussion:

To our knowledge, this paper is the first to describe IPV perpetration in addition to specific types of victimization among several distinct sexual minority groups in a population-based sample of young adults. The present findings show that a mostly heterosexual self-identification is associated with greater odds of both violence victimization and perpetration among males and females compared to the experiences of individuals identifying as 100% heterosexual. In addition, among males only, a bisexual identity is associated with greater odds of physical victimization, and greater odds of both forced sex victimization and perpetration. Homosexual males and females had similar odds of different forms of IPV victimization compared to heterosexuals. These findings corroborate emerging literature indicating that some sexual minorities face greater odds of experiencing IPV victimization than heterosexuals. We further add to the literature by providing prevalence estimates of IPV perpetration among young adult sexual minorities in a population-based sample. Particularly, females had a higher prevalence of physical and mild or threatened violence perpetration than males. Although the context of this perpetration is unknown, sex differences have been seen in other studies (26). It is notable, however, that current estimates for perpetration are consistently lower than the 29% to 57% prevalence estimates of violence perpetration in the past year among college students (12). This may be partly a function of age differences.

Our results varied by sex, sexual orientation group, and IPV outcome. One major finding is that mostly heterosexual males and females had consistently higher odds across most reported types of violence victimization and perpetration than 100% heterosexuals. Although traditional research examines three categories of sexual orientation – heterosexual, bisexual, and homosexual - studies over the past decade have begun to investigate the mostly heterosexual orientation group specifically (27–31). In fact, Savin-Williams et al. found that persons who identify as mostly heterosexual significantly outnumber both bisexuals and homosexual populations in the United States, with about 4% of males and 9% of females in the general population identifying as mostly heterosexual (29). We see similar patterns in the Add Health data, although the percentage of Add Health females identifying as mostly heterosexual is higher. A systematic review of the health outcomes among mostly heterosexuals shows that these populations consistently report higher levels of risk in IPV victimization outcomes compared to heterosexuals (28). To our knowledge, our study is the first to find that mostly heterosexual males and females are also more likely to report perpetration, and for most violence types the differences are large. The results from this paper are consistent with emerging research on risk factors associated with identifying with a minority sexual orientation group and suggest that more studies are needed to describe the experiences and needs of neglected sexual minority populations (22, 32).

Our examination of IPV among bisexual men resulted in another major finding. Bisexual males reported greater odds of physical IPV victimization and greater odds of forced sex perpetration and victimization compared to 100% heterosexual males. Odds for bisexual males also tend to be even greater than those for mostly heterosexual males, sometimes markedly so. These results are consistent with those of other recent research showing that bisexual adults are more likely than either homosexual or heterosexual orientation groups to experience IPV victimization (4, 7, 22), or in the case of a convenience sample, victimization and perpetration (17). However, most of the past research on the bisexual population has focused on females due to small bisexual male sample sizes (7). Although other studies have noted that bisexual females are at greater risk of sexual abuse by a partner (3, 7), our study did not find such associations. There may be several reasons for this discrepancy, including current analysis factors such as adjusting for partner sex, and examining violence within a specific relationship versus lifetime experience. For example, several studies that examine IPV among bisexuals do not account for sex of the partner in those relationships (4, 10, 17, 22). The 2010 National Intimate Partner and Sexual Violence Survey did account for sex of the partner, and found that bisexual women who experienced IPV victimization reported only male perpetrators (7). A 2018 study found that perpetrator bi-negativity and perceived or real infidelity were the variables most related to perpetration and victimization of IPV among bisexuals, especially if the perpetrator was male (22).

Having a same-sex partner in our models was not a significant predictor of IPV. However, there were few persons in our analysis sample who reported a current or most recent same-sex partner, resulting in less precision in our estimates. It is interesting to note that although not a statistically significant predictor of IPV, the patterns of odds ratios are similar across physical and threatened violence for males and females, and suggest the possibility that having a same-sex partner may be linked to greater odds of victimization and lower odds of perpetration. Patterns for forced sex outcomes were different, however, the confidence intervals are quite large. To better understand these patterns, other characteristics linked to violence may need to be considered. For example, other studies examining violence in same-sex partnerships suggest the importance of gender and gender expression. McKenry found that higher scores of masculinity were associated with higher levels of violence among both lesbians and gay men (33). In contrast, other studies have found that lower socioeconomic status is more important than gender as a determinant of IPV (34). Additional studies are needed to explore the underlying relationships among intimate partner violence, gender, and social stratification among sexual minorities.

These findings are mostly consistent with theories on minority stress; individuals in our sample who identified as “mostly heterosexual” or “bisexual” had greater odds of some forms of IPV compared to those individuals who identified as 100% heterosexual. It is interesting to note that the relative degree of elevated risk for these two minority identities varies by sex. Among males, bisexuals appear to be at particular risk. Among females, mostly heterosexual females appear to be at greatest risk. Overall, these findings are consistent with the hypothesis that sexual minorities may experience stigma and discrimination related to their orientation that produces psychological distress linked to negative health behaviors (15, 35). However, in contrast to what would be predicted by Minority Stress Theory, individuals in our sample who identified as homosexual did not experience greater odds of IPV compared to 100% heterosexuals. Rather, it is the “non-monosexuals” who appear at greater risk for both perpetration and victimization. This may be due to the development of a distinct “monosexual” identity and sense of purpose through group affiliation based on sexual identity, which can provide protection against minority stress (23). This may explain some of the differences in health risk behaviors between those who identify in the middle of the spectrum of sexual identity and those who strongly identify on the poles of sexual identity. Further research is needed to understand this phenomenon.

Limitations and Strengths

This study has limitations that may temper our findings. Even with our large population-based sample, the numbers of participants who identified as bisexual, mostly homosexual and homosexual were relatively small, especially among men. These small numbers may have limited our ability to detect significant associations. We made the decision to collapse the original variable categories “mostly homosexual” and “homosexual” to increase statistical power. This diminished our ability to thoroughly examine whether individuals who identify as “nonexclusive” at the homosexual end of the spectrum have worse violence outcomes than either the heterosexual or homosexual groups. However, we were able to identify previously unobserved associations, including elevated experiences of IPV perpetration among bisexual males. Second, this analysis is cross-sectional and only generalizable to individuals in Grades 7 through 12 in the 1994–1995 school year. There may be differences between this cohort and contemporary cohorts, given changes in population attitudes about sexual minorities and associated stressors. Third, our measure of sexual orientation was limited. Although the survey question asked about self-identification, the survey question could be interpreted as reporting degrees of attraction to males and females and not necessarily how that individual identifies socially. Despite the limitations of this measure, we were able to capture a more nuanced and diverse sexual orientation identification than previous studies’ categories of heterosexual, bisexual, and homosexual.

The paper has several strengths that advance the literature on sexual minorities and intimate partner violence. First, because these data come from a nationally representative sample of adolescents in 9^th-12^th grades in the 1994–95 school year, our results offer prevalence estimates that are generalizable to sexual minority populations from this cohort. In addition, most IPV research linked to sexual orientation collapses across all categories that are not 100% heterosexual. Our more nuanced measure of sexual minority identity allowed us to examine the diversity of experiences among young adult sexual minorities. Although the mostly heterosexual group has largely been ignored in research until recently, this study is consistent with emergent literature showing that mostly heterosexuals, especially women, are a distinct population who are at greater risk for many negative health outcomes, including IPV (28). Finally, our paper is the first to examine sexual minority IPV perpetration in a population-based sample, and to do so in the context of sex of partner. Perpetration has rarely been examined, and our findings suggest that “mostly heterosexual” individuals and bisexual males are at elevated risk for perpetration. Given that promising intervention and prevention efforts are typically heteronormative (36), they may not provide the necessary tools to address perpetration in these populations. Overall, our study provides important information regarding the diversity of IPV experiences within the larger sexual minority community.

Conclusion

This study is the first of its kind to examine IPV perpetration among sexual minorities in a nationally representative sample. The results showed that males and females who identify as mostly heterosexual (or bisexual among males) have increased odds of both victimization and perpetration for some forms of IPV. These findings indicate that some sexual minorities in the United States are disproportionately affected by violence in their romantic relationships, pointing to the need for increased IPV awareness and prevention efforts, including in the context of comprehensive and inclusive sexual health education. In addition, some sexual minorities – those who identified as homosexual – had the same or even decreased odds of experiencing IPV. It is important to study the protective factors of this group and recognize considerable heterogeneity among sexual minorities experiences, which has an important policy and prevention implications. Further studies examining the reasons underlying patterns of IPV among sexual minorities are needed. Identifying the latent mechanisms underlying partner violence that are common to persons of all orientations and those mechanisms that may be unique to sexual minorities would inform prevention and intervention efforts and provide critical perspectives to inform inclusive changes in education and prevention programs, as well as policy.