Anti-LGBT Victimization, Fear of Violence at School, and Suicide Risk Among Adolescents

Abstract

We investigated whether strengths of the relationships between anti–lesbian, gay, bisexual, and transgender (LGBT) victimization, fear of violence at school, and suicide risk differ by sexual orientation among a predominantly ethnic minority sample of adolescents. Using the 2012 District of Columbia Youth Risk Behavior Survey high school data set, we performed a 4-stage, stepwise logistic regression for suicide attempts, suicidal ideation, and suicide planning. First, we tested the independent variables, sexual orientation and anti-LGBT victimization. Second, we added an anti-LGBT victimization by sexual orientation interaction term. Third, we tested the independent variable, fear of violence at school, in an additive model. Fourth, we added a fear of violence at school by sexual orientation interaction term. In Model 1, sexual orientation and anti-LGBT victimization were both significantly associated with each suicide risk behavior. In Model 2, the anti-LGBT victimization by sexual orientation interaction term was not significant for any of the dependent variables. In Model 3, fear of violence at school was significantly associated with each suicide risk behavior. In Model 4, the fear of violence at school by sexual orientation interaction term was not significant for suicide attempts or suicidal ideation but was significant for suicide planning in the direction opposite to our hypotheses. Anti-LGBT victimization, sexual orientation, and fear of violence at school were associated with suicidal ideation, suicide planning, and suicide attempts. The strength of the association between fear of violence at school and suicide planning was weaker for sexual minority adolescents than for heterosexual adolescents.

Compared with heterosexual peers, sexual minority adolescents report significantly higher rates of suicide risk behaviors (Kann et al., 2016; Marshal et al., 2011). Meyer (2013) postulated that minority stress accounts for such disparities among sexual minority populations. Two important features of minority stress for sexual minorities are that the stresses are (a) unique to these individuals and thus not experienced by heterosexuals and (b) additive to general stressors that can impact all individuals. Meyer’s minority stress model distinguishes between distal processes, defined as objective events and conditions, and proximal processes, which relate to the effects of external stress on the individuals’ internal view of themselves and their world. Proximal minority stress takes three forms: expectations of rejection, concealment, and internalized anti–lesbian, gay, and bisexual (LGB) stigma.

Anti-LGB bullying and other forms of peer victimization constitute one type of distal minority stress processes for sexual minority adolescents. Studies have found that sexual minority adolescents report higher levels of peer victimization and victimization based on their actual or perceived sexual orientation than their heterosexual peers (Almeida, Johnson, Corliss, Molnar, & Azrael, 2009; Darwich, Hymel, & Waterhouse, 2012; Espelage, Aragon, & Birkett, 2008; Friedman et al., 2011; Kann et al., 2016; Mueller, James, Abrutyn, & Levin, 2015; Russell, Everett, Rosario, & Birkett, 2014; Shields, Whitaker, Glassman, Franks, & Howard, 2012). Although not all peer victimization research identifies anti-LGB discrimination as the thematic content, the consistent disparity in reported victimization between sexual minority and heterosexual youth suggests that sexual orientation may play a role.

Consistent with the minority stress model, peer victimization has been found to be associated with suicide risk behaviors for sexual minority youth (Goodenow, Szalacha, & Westheimer, 2006), and the association between victimization and suicidality has been reported to be stronger for sexual minority than heterosexual youth (Bontempo & D’Augelli, 2002). Research has also found evidence of an association between tacitly anti-LGB victimization and suicide risk behaviors among sexual minority youth and young adults (Hightow-Weidman et al., 2011; Liu & Mustanski, 2012; Russell, Ryan, Toomey, Diaz, & Sanchez, 2011). Further supporting the model, there is evidence that homophobic teasing moderates the relationship between sexual orientation and suicidality (Birkett, Espelage, & Koenig, 2009). One study reported differences by race: Anti-LGB victimization was associated with suicidality for white lesbian, gay, bisexual, transgender, and questioning (LGBTQ) youth but not LGBTQ youth of color (Poteat, Mereish, DiGiovanni, & Koenig, 2011). The present study builds on prior research by investigating the effect of anti–lesbian, gay, bisexual, and transgender (LGBT) victimization among sexual minority adolescents when accounting for other forms of bullying in a predominantly ethnic minority population-based sample.

The extant research on school climate and suicide risk for sexual minority youth relates to the theorized pathway between proximal minority stress processes and mental health outcomes. Sexual minority adolescents’ perception of their school’s acceptance of sexual minority students directly relates to their expectations of rejection in that setting. Positive school climate, living in a jurisdiction with a more protective school climate, and living in a school district with LGBT-inclusive antibullying policies have been associated with lower levels of depressive symptoms or suicidality for sexual minority students (Birkett et al., 2009; Hatzenbuehler, Birkett, Van Wagenen, & Meyer, 2014; Hatzenbuehler & Keyes, 2013). Conversely, for sexual minority students, attending a school with a Gay-Straight Alliance was not associated with suicidality when controlling for victimization (Goodenow et al., 2006), nor was an association found between Gay-Straight Alliance presence or participation and suicidality (Toomey, Ryan, Diaz, & Russell, 2011). The present study seeks to further contribute to the literature by assessing the relationship between suicidality and individual LGB students’ perceptions of climate, which would be expected to vary substantially, even within a particular school.

Using a predominantly ethnic minority sample of adolescents, we investigated the association between sexual orientation, distal and proximal minority stress processes, and suicide risk behaviors. Based on the minority stress model and existing literature, we developed two hypotheses: (a) anti-LGBT victimization is more strongly associated with suicide risk behavior for sexual minority students than for heterosexual students and (b) fear of violence at school is more strongly associated with suicide risk behavior for sexual minority students than for heterosexual students.

Method

The present study was conducted using the 2012 District of Columbia Youth Risk Behavior Survey (YRBS) for high school students. The Institutional Review Board of the George Washington University approved the study. The YRBS is conducted biennially by the Centers for Disease Control and Prevention to assess health risk behaviors among adolescents (Centers for Disease Control and Prevention, 2011). The District of Columbia (DC) YRBS is administered by the DC Office of the State Superintendent for Education in collaboration with the DC Department of Health (Ost & Maurizi, 2013). Data were collected October 2012 through January 2013. Eligibility criteria required students to be enrolled in a traditional public or public charter high school (grades 9–12). Passive consent was utilized. Participants were able to discontinue at any time.

Sample

Sampling for the 2012 DC YRBS used a two-stage cluster methodology at the school and classroom level, and the total response rate was 68% (Ost & Maurizi, 2013). Inclusion criteria for the present study required completion of the sexual orientation item, resulting in an analytic sample size of 10,593.

Measures

Demographics.

Single items assessed age, grade, sex, race, and ethnicity. Race and ethnicity were collapsed into one variable with the following categories: American Indian/Alaska Native, Asian, Black or African American, Native Hawaiian or other Pacific Islander, White, Hispanic/Latino, multiracial/Hispanic, and multiracial/non-Hispanic.

Sexual orientation.

Sexual orientation was measured through a single item with the response options heterosexual (straight), gay or lesbian, bisexual, or not sure. Participants who selected heterosexual (straight) were classified as heterosexual; any other response was classified as lesbian, gay, bisexual, or questioning (LGBQ).

Peer victimization.

Peer victimization was assessed using two dichotomous items assessing experiences of being bullied on school property or electronically bullied during the previous 12 months. These items were combined, with participants who reported experiencing either or both types of bullying coded as 1 and those who responded no to both items coded as 0.

Anti-LGBT harassment was assessed with a single item assessing frequency participants had been harassed on school property because someone thought they were gay, lesbian, bisexual, or transgendered. The variable was dichotomized into participants reporting any anti-LGBT victimization and those reporting none.

Fear of violence at school was assessed with a single item assessing frequency participants had been afraid of being beaten up at school during the previous year. This variable was dichotomized into participants reporting having been afraid at least once and those reporting no such instances.

Suicide risk behaviors were assessed with three items. Participants were asked about experiencing suicidal ideation, making a suicide plan, and attempting suicide in the previous year. The suicidal ideation and suicide plan items were both dichotomous. The suicide attempts item assessed frequency of this behavior and was dichotomized into participants who reported having made any suicide attempts and those who reported none.

Analytic Plan

Analyses were conducted using SAS version 9.3 (SAS Institute, Cary, NC) and followed recommendations for YRBS data (Centers for Disease Control and Prevention, 2016). Descriptive statistics were obtained using PROC SURVEYFREQ. PROC SURVEYLOGISTIC was used to test bivariate associations and the study hypotheses. Separate analyses were run for each suicide risk behavior. The variables were entered into a logistic regression in a four-stage, stepwise process. Model 1 included the independent variables sexual orientation and anti-LGBT victimization in an additive model, with peer victimization included as a control. In Model 2, an anti-LGBT victimization by sexual orientation interaction term was added. Model 3 included the independent variable fear of violence at school in an additive model. In Model 4, a fear of violence at school by sexual orientation interaction term was added. For each analysis, list-wise deletion was used for surveys with missing data, as has been used in previous analyses using YRBS and similar population-based data (Duncan & Hatzenbuehler, 2014; Mueller et al., 2015; Russell et al., 2014). A Bonferroni adjustment was calculated based on 12 hypotheses (three outcome variables by four models) for an adjusted α = .0042.

Results

Descriptive Statistics

The descriptive statistics of the sample are reported in Table 1. The sample was majority female and Black or African American. The Rao-Scott χ² test was performed to test whether each demographic characteristic, independent variable, and dependent variable was independent of sexual orientation. Significant associations were found for sex, age, race, grade, and each of the independent and dependent variables. The majority of participants reported a heterosexual identity (see Table 2).

Table 1.

Descriptive Statistics of Analytic Sample

	Total			LGBQ		Heterosexual
Statistics	n	Wt. n	Wt. %	Wt. n	Wt. %	Wt. n	Wt. %	p value
Total sample	10,593	15,874	100.00	2,435	15.34	13,439	84.66	<.0001
Sex
Male	4,778	7,583	48.18	447.24	5.90	7,136	94.10	<.0001
Female	5,733	8,156	51.82	1,925	23.60	6,231	76.40
Age, years
13 or youngera	198	341.39	2.16	93.94	27.52	247.45	72.48	<.0001
14	1,960	3,604	22.78	473.40	13.14	3,130	86.86
15	2,662	4,027	25.46	628.36	15.60	3,399	84.40
16	2,536	3,583	22.65	595.34	16.61	2,988	83.39
17	2,300	3,019	19.08	455.15	15.08	2,564	84.92
18 or older	902	1,246	7.87	182.20	14.63	1,063	85.37
Race
Black or African American	7,091	10,824	70.70	1,606	14.84	9,218	85.16	<.0001
White	518	667.95	4.36	69.84	10.46	598.11	89.54
Hispanic/Latino	882	1,293	8.44	127.81	9.89	1,165	90.11
Multiracial	1,323	1,912	12.49	374.20	19.57	1,538	80.43
Other race/ethnicity	422	613.32	4.01	126.65	20.65	486.67	79.35
Gradeb
9th	2,876	5,622	35.78	816.61	14.52	4,806	85.48	.0020
10th	2,746	3,764	23.95	606.62	16.12	3,157	83.88
11th	2,487	3,328	21.18	531.23	15.96	2,797	84.04
12th	2,336	2,928	18.63	418.73	14.30	2,509	85.70
Peer victimizationc	1,489	2,232	14.56	566.80	24.68	1,665	12.78	<.0001
Anti-LGBT victimizationc	916	1,429	9.10	514.31	21.33	914.42	6.88	<.0001
Fear of violence at schoolc	830	1,324	8.42	372.75	15.45	951.21	7.14	<.0001
Suicide attemptc	1,033	1,608	13.17	520.11	27.47	1,088	10.55	<.0001
Suicidal ideationc	1,527	2,247	14.69	703.44	30.51	1,544	11.88	<.0001
Suicide planningc	1,512	2,236	14.57	644.47	27.84	1,592	12.21	<.0001

^aParticipants reporting age 12 years and younger and age 13 years were combined into one group because of small group size.

^bParticipants reporting ungraded or other grade are not reported because of small cell size (>50) to prevent participant reidentification.

^cStatistics for participants reporting the type of behavior are presented (i.e. participants responding yes or at least once for each question).

Table 2.

Sexual Orientation of Participants

Sexual orientation	n	Wt. n	Wt. %
Heterosexual	8,961	13,439	84.66
Gay or lesbian	374	556.95	3.51
Bisexual	943	1,400	8.82
Not sure	315	477.36	3.01

To minimize the risk of participant reidentification, crosstabulations and analyses for cell sizes of less than 50 are not reported. For this reason, racial/ethnic groups were collapsed into Black/African American, White, Hispanic/Latino, multiracial, and other race/ethnicity, and the ungraded/other grade group was not reported. Bivariate associations were tested for the demographic variables and the independent variables (not shown). Race/ethnicity, sex, age, and grade were each related to at least one independent variable and one dependent variable and were therefore retained in the model for hypothesis tests.

Hypotheses Tests

Hypothesis 1.

First, we tested the association between sexual orientation, peer victimization, and anti-LGBT victimization and each suicide risk behavior in an additive model (see Tables 3, 4, and 5, Model 1). Each independent variable was significantly associated with each of the three suicide risk behaviors. Next, we tested the hypothesis by adding an anti-LGBT victimization by sexual orientation interaction term (see Tables 3-5, Model 2). For each dependent variable, the interaction term was not significant.

Table 3.

Multivariable Logistic Regression Models for Suicide Attempts

	Model 1, n = 7,797			Model 2, n = 7,797			Model 3, n = 7,759			Model 4, n = 7,759
Variables	OR	95% CI	p value	OR	95% CI	p value	OR	95% CI	p value	OR	95% CI	p value
Race (reference = Black/African American)
White	.31	(.19–.53)	<.0001	.32	(.19–.53)	<.0001	.31	(.18–.53)	<.0001	.31	(.18–.53)	<.0001
Hispanic/Latino	1.27	(.97–1.65)	.0804	1.27	(.97–1.65)	.0828	1.25	(.96–1.64)	.1014	1.25	(.95–1.63)	.1101
Multiracial	1.35	(1.10–1.65)	.0045	1.35	(1.10–1.65)	.0043	1.32	(1.07–1.62)	.0099	1.32	(1.07–1.63)	.0091
Other race/ethnicity	1.04	(.70–1.54)	.8498	1.04	(.71–1.54)	.8395	1.01	(.67–1.50)	.9769	1.02	(.69–1.51)	.9153
Age	1.21	(1.08–1.37)	.0013	1.21	(1.08–1.37)	.0012	1.20	(1.06–1.35)	.0029	1.20	(1.07–1.35)	.0027
Sex (reference = male)	1.42	(1.20–1.67)	<.0001	1.41	(1.19–1.66)	<.0001	1.46	(1.24–1.73)	<.0001	1.46	(1.23–1.73)	<.0001
Grade (reference = 9th)a
10th	.79	(.62–.99)	.0431	.79	(.62–.99)	.0443	.80	(.63–1.01)	.0565	.79	(.63–1.00)	.0540
11th	.52	(.37–.72)	<.0001	.52	(.37–.72)	<.0001	.54	(.39–.75)	.0002	.54	(.39–.74)	.0002
12th	.48	(.32–.72)	.0004	.48	(.32–.72)	.0004	.51	(.34–.76)	.0010	.50	(.34–.76)	.0009
Sexual orientation (reference = heterosexual)	2.33	(1.97–2.75)	<.0001	2.44	(2.03–2.94)	<.0001	2.28	(1.93–2.70)	<.0001	2.42	(2.02–2.90)	<.0001
Peer victimization	1.99	(1.65–2.41)	<.0001	1.99	(1.65–2.41)	<.0001	1.68	(1.37–2.05)	<.0001	1.68	(1.38–2.05)	<.0001
Anti-LGBT victimization	2.68	(2.13–3.38)	<.0001	2.93	(2.22–3.85)	<.0001	2.45	(1.94–3.10)	<.0001	2.47	(1.96–3.12)	<.0001
Anti-LGBT victimization × Sexual minority orientation				.79	(.50–1.25)	.3196
Fear of violence at school							2.21	(1.70–2.86)	<.0001	2.49	(1.87–3.32)	<.0001
Fear of violence × Sexual minority orientation										.67	(.40–1.12)	.1297

Note. OR = odds ratio; CI = confidence interval. Bold indicates significance at α = .0042.

^aUngraded/other grade not reported because of small cell size (n < 50).

Table 4.

Multivariable Logistic Regression Models for Suicidal Ideation

	Model 1, n = 9,624			Model 2, n = 9,624			Model 3, n = 9,574			Model 4, n = 9,574
Variables	OR	95% CI	p value	OR	95% CI	p value	OR	95% CI	p value	OR	95% CI	p value
Race (reference = Black/African American)
White	.72	(.52– 1.01)	.0604	.73	(.52–1.02)	.0650	.73	(.52–1.03)	.0721	.73	(.52–1.03)	.0729
Hispanic/Latino	1.08	(.87–1.33)	.4940	1.08	(.87–1.33)	.4990	1.06	(.85–1.31)	.6002	1.05	(.84–1.30)	.6869
Multiracial	1.60	(1.35- 1.90)	<.0001	1.60	(1.35–1.90)	<.0001	1.58	(1.33–1.88)	<.0001	1.59	(1.34–1.89)	<.0001
Other race/ethnicity	1.09	(.78–1.51)	.6205	1.09	(.79–1.51)	.6043	1.06	(.76–1.49)	.7188	1.08	(.78–1.51)	.6372
Age	.94	(.86–1.02)	.1151	.94	(.86–1.02)	.1155	.92	(.85–1.01)	.0689	.93	(.85–1.01)	.0752
Sex (reference = male)	1.81	(1.60–2.05)	<.0001	1.81	(1.59–2.05)	<.0001	1.87	(1.65–2.13)	<.0001	1.87	(1.64–2.12)	<.0001
Grade (reference = 9th)a
10th	1.19	(.99–1.44)	.0670	1.19	(.99–1.44)	.0663	1.20	(.99–1.45)	.0606	1.20	(.99–1.45)	.0614
11th	1.08	(.84–1.38)	.5431	1.08	(.84–1.38)	.5475	1.12	(.87–1.44)	.3795	1.11	(.87–1.42)	.4080
12th	1.23	(.89–1.68)	.2071	1.23	(.89–1.69)	.2066	1.30	(.94–1.79)	.1158	1.29	(.94–1.78)	.1205
Sexual orientation (reference = heterosexual)	2.31	(2.00–2.67)	<.0001	2.38	(2.03–2.79)	<.0001	2.29	(1.98–2.66)	<.0001	2.47	(2.13–2.87)	<.0001
Peer victimization	3.37	(2.91–3.89)	<.0001	3.37	(2.91–3.89)	<.0001	3.03	(2.60–3.53)	<.0001	3.02	(2.60–3.52)	<.0001
Anti-LGBT victimization	1.95	(1.59–2.41)	<.0001	2.09	(1.60–2.72)	<.0001	1.86	(1.50–2.30)	<.0001	1.89	(1.53–2.33)	<.0001
Anti-LGBT victimization × Sexual orientation			.83	(.56–1.25)	.3774
Fear of violence at school							1.72	(1.37–2.15)	<.0001	2.02	(1.56–2.61)	<.0001
Fear of violence × Sexual orientation										.57	(.37–87)	.0094

Note. OR = odds ratio; CI = confidence interval. Bold indicates significant at α = .0042.

^aUngraded/other grade not reported because of small cell size (n < 50).

Table 5.

Multivariable Logistic Regression Models for Suicide Planning

	Model 1, n = 9,632			Model 2, n = 9,632			Model 3, n = 9,583			Model 4, n = 9,583
Variables	OR	95% CI	p value	OR	95% CI	p value	OR	95% CI	p value	OR	95% CI	p value
Race (reference = Black/African American)
White	.70	(.52–.95)	.0233	.71	(.53–.96)	.0276	.71	(.53–.96)	.0265	.71	(.52–.96)	.0238
Hispanic/Latino	.90	(.71–1.15)	.3997	.90	(.71–1.15)	.3898	.89	(.69–1.13)	.3371	.87	(.68–1.12)	.2832
Multiracial	1.36	(1.13–1.64)	.0010	1.37	(1.14–1.64)	.0009	1.34	(1.11–1.62)	.0019	1.35	(1.12–1.63)	.0014
Other race/ethnicity	1.34	(1.00–1.79)	.0518	1.34	(1.01–1.80)	.0464	1.33	(.99–1.78)	.0602	1.35	(1.01–1.81)	.0449
Age	.99	(.91–1.08)	.8686	.99	(.91–1.08)	.8709	.99	(.91–1.08)	.7600	.99	(.91–1.08)	.7858
Sex (reference = male)	1.66	(1.45–1.90)	<.0001	1.65	(1.44–1.89)	<.0001	1.69	(1.48–1.94)	<.0001	1.69	(1.47–1.93)	<.0001
Grade (ref = 9th)a
10th	1.07	(.89–1.28)	.4925	1.07	(.89–1.28)	.4790	1.07	(.89–1.28)	.4847	1.07	(.89–1.28)	.4907
11th	.93	(.74–1.16)	.5053	.92	(.74–1.16)	.4975	.94	(.75–1.19)	.6050	.94	(.74–1.18)	.5673
12th	.88	(.65–1.19)	.3970	.88	(.65–1.19)	.4007	.91	(.67–1.23)	.5200	.90	(.67–1.23)	.5118
Sexual orientation (reference = heterosexual)	1.91	(1.65–2.21)	<.0001	2.05	(1.74–2.41)	<.0001	1.89	(1.63–2.19)	<.0001	2.07	(1.77–2.42)	<.0001
Peer victimization	2.44	(2.11–2.82)	<.0001	2.44	(2.11–2.82)	<.0001	2.27	(1.94–2.65)	<.0001	2.27	(1.94–2.65)	<.0001
Anti-LGBT victimization	2.16	(1.77–2.64)	<.0001	2.48	(1.95–3.16)	<.0001	2.09	(1.71–2.55)	<.0001	2.13	(1.74–2.60)	<.0001
Anti-LGBT victimization × Sexual orientation				.68	(.46–1.00)	.0476
Fear of violence at school							1.43	(1.13–1.81)	.0031	1.74	(1.34–2.27)	<.0001
Fear of violence × Sexual orientation										.51	(.32–.80)	.0032

Note. OR = odds ratio; CI = confidence interval. Bold indicates significant at α = .0042.

^aUngraded/other grade not reported because of small cell size (n < 50).

Hypothesis 2.

First, we tested the association between fear of violence at school and each suicide risk behavior in an additive model that also included sexual orientation, peer victimization, and anti-LGBT victimization (see Tables 3-5, Model 3). Fear of violence at school was significantly associated with suicide attempts, suicidal ideation, and suicide planning. Sexual minority orientation, peer victimization, and anti-LGBT victimization remained significantly associated with each suicide risk behavior. Next, we tested the hypothesis by including a fear of violence at school by sexual orientation interaction term (see Tables 3-5, Model 4). The interaction term was not significant for suicide attempts or suicidal ideation. The fear of violence at school by sexual orientation interaction term was significant for suicide planning, but not in the expected direction.

Discussion

Anti-LGBT Victimization

We found evidence of an association between anti-LGBT victimization and suicide attempts, suicide planning, and suicidal ideation. This finding is consistent with previous research showing an association between anti-LGBT victimization and suicidality (Hightow-Weidman et al., 2011; Liu & Mustanski, 2012; Russell et al., 2011). Importantly, we found evidence that anti-LGBT victimization was related to all three suicide risk behaviors tested independent of general peer victimization and sexual orientation. Among the suicide risk behaviors, the strongest association was found between anti-LGBT victimization and suicide attempts, with participants who reported anti-LGBT victimization being more than 2.5 times as likely to report a past-year suicide attempt. Anti-LGBT victimization was more strongly associated with suicide attempts than reporting other forms of peer victimization, suggesting that this type of victimization is particularly harmful. Our hypothesis that the association between anti-LGBT victimization and suicide risk behaviors is stronger for sexual minority students than for heterosexuals was not supported. Although inconsistent with the minority stress model, our finding is notable in adding to the evidence that anti-LGBT victimization has a deleterious impact on sexual minority and heterosexual adolescents (Poteat & Espelage, 2007).

Several possible explanations could account for the lack of support for the hypotheses. First, the measure of anti-LGBT victimization was limited to a single question about harassment; as a result, other types of anti-LGBT victimization were not assessed. Second, the differences in size between the sexual minority and heterosexual subsamples and between the participants who reported anti-LGBT victimization and those who did not may also have prevented the detection of the moderation effects. Finally, it is possible that the effect of anti-LGBT harassment on all students is similarly detrimental for both LGBT and heterosexual adolescents. If true, the mechanism through which this type of discrimination contributes to negative outcomes may differ for sexual minority and heterosexual adolescents, but the end result may be similar.

Meyer’s minority stress model holds that being subjected to discrimination and harassment based on one’s sexual orientation constitutes a unique form of stress for sexual minority adolescents (Meyer, 2013). Our findings suggest that harassment based on perceptions that one is a sexual minority or transgender is also a potent stressor for heterosexual adolescents that has an effect distinct from other forms of peer victimization. The harm associated with being targeted based on one’s presumed LGBT status for heterosexual students may be due to the stigma associated with being LGBT in contemporary U.S. culture.

The results of the present study are notable because of the racially diverse and majority African American sample. A previous study found that homophobic victimization was not associated with suicidality for LGBTQ and heterosexual youth of color when controlling for other forms of victimization (Poteat et al., 2011). Our findings suggest that anti-LGBT victimization is associated with suicidality among youth of color, even when controlling for reports of bullying.

Consistent with previous research, sexual minority orientations (Eisenberg & Resnick, 2006; Hatzenbuehler et al., 2014; Kann et al., 2016; Shields et al., 2012) and experiences of peer victimization (Bontempo & D’Augelli, 2002; Mueller et al., 2015) were associated with increased odds of reporting suicide attempts, suicide planning, and suicidal ideation. Sexual minority adolescents were more than twice as likely to report suicide attempts and suicidal ideation and more than 1.5 times as likely to report suicide planning than their heterosexual peers. Furthermore, our results suggest that this disparity is not solely a function of sexual minority adolescents being more likely to experience peer victimization or anti-LGBT victimization.

Fear of Violence at School

We found evidence that students’ expectations concerning their school environment are related to suicide risk. Fear of violence at school was associated with each suicide risk behavior independent of reported direct experiences of bullying and anti-LGBT victimization. This finding suggests that students’ perception that they will be subjected to violence at school is related to suicide risk, regardless of whether they have previously been victimized. As with anti-LGBT victimization, the strongest association was found between fear of violence at school and suicide attempts.

We found evidence that the relationship between fear of violence at school and suicide planning differed for sexual minority and heterosexual participants. However, contrary to our hypothesis, the results showed that the relationship was weaker for sexual minority than for heterosexual participants. Several possible explanations could account for this finding. One, the weaker relationship may denote a ceiling effect, given the strong associations between fear of violence at school and a sexual minority orientation and suicide planning. Another possible explanation relates to resilience among sexual minority youth. The present study did not include any measures of resilience, so it is possible that an unobserved factor such as LGBT community support serves as a buffer. In contrast, we found that the associations between fear of violence at school and suicide attempts and suicidal ideation did not differ between sexual minority and heterosexual participants.

As with our first hypothesis, methodological concerns may have contributed to our results. The measure of fear of violence at school did not specifically assess violence based on one’s sexual orientation. For this reason, participants may have endorsed this item if they had feared violence in the previous year based on factors unrelated to sexual orientation, thus obfuscating the unique effects of minority stressors on sexual minority participants. Furthermore, the use of a single item failed to assess the full domain of expectations of victimization at school. A more comprehensive measure would have assessed fear of violence as well as fear of other forms of rejection such as harassment and bullying. In addition, the relatively small sexual minority subsample may have prevented an interaction effect from being detected. A final possible explanation is that the relationship between fear of violence at school and suicide risk is the same for sexual minority adolescents as it is for their heterosexual peers. This finding is not incompatible with the minority stress model, which focuses solely on expectations of rejection related to sexual minority identity.

The current study included several methodological strengths. The participants in the sample were predominantly racial minorities (nearly 95% endorsed a racial identity other than White only). This sample composition differs from other studies examining suicide risk and sexual minority adolescents with majority White samples (Birkett et al., 2009; Bontempo & D’Augelli, 2002; Eisenberg & Resnick, 2006; Espelage et al., 2008; Hatzenbuehler, 2011; Poteat & Espelage, 2007). As a result, the findings of the present analysis contribute to the understanding of the associations among anti-LGBT victimization, fear of violence, and sexual orientation and suicide risk behavior among racial minority high school students.

An additional strength related to the sample is the inclusion of regular public high schools and public charter schools. The representation of students from both types of schools contributes to the heterogeneity of the sample. Public and public charter high schools may differ with respect to school policies regarding bullying, overall school climate, and acceptance of sexual minority students and populations. The results therefore may be more generalizable to students in other large urban school districts. A final strength of the current study is the use of the Bonferroni correction. As a result, we can be reasonably certain that the observed results reflect true relationships among the variables in the population despite the use of multiple analyses to test the hypotheses.

A number of important limitations should be noted in interpreting the results of the current analysis. The data were cross-sectional, which prevents the investigation of causal relationships. The present analysis assumed temporal relationships between the constructs of interest, namely that victimization experiences and fear of violence preceded suicide risk behavior. However, this assumption could not be tested and may be incorrect for any given participant. The missing data points in the variables of interest represent another limitation. In the analysis of the effect of fear of violence at school on suicide attempts, in which the largest number of data points were missing, 27% of the observations were excluded. List-wise deletion requires the assumption that the data are missing completely at random. It is possible that some of the data points were missing because of the stigma associated with the constructs being measured, making the assumption of missing completely at random untenable. However, if participants omitted items because of the associated stigma, one can assume that they did not answer the question because they would have endorsed the stigmatized response (i.e., having experienced peer victimization and/or fearing violence at school). If so, excluding these responses would result in a potential underestimation of effect.

A fourth limitation for the current study is the use of single items to measure the constructs of interest. Sexual orientation, anti-LGBT harassment, fear of violence at school, suicide attempts, suicidal ideation, and suicide planning were all operationalized with responses to single questions. Only two items were used to measure experiences of peer victimization. The measure of sexual orientation presents unique challenges. Responding to this item as a sexual minority requires that the participant be aware of their orientation, identify with one of the options presented (e.g., gay or lesbian), and be willing to disclose it on a survey administered at school. Youth who engage in same-sex sexual behavior and identify as straight may not be included in the sexual minority subsample in this analysis. It is also possible that a participant who has not come out at school would not disclose this information, even on an anonymous survey. For these reasons, the measurement of sexual orientation may not be accurate. The gender disparity in sexual minority orientations in the sample is notable and may reflect inaccurate reporting of sexual minority status.

The final significant limitation for the current study is the operationalization of minority stress processes. The proximal minority stress process was operationalized through an item about fear of violence at school without specifying that it be attributed to the participant’s sexual orientation. Meyer’s minority stress model suggests that this process be explicitly connected to the individual’s sexual minority orientation, at least in terms of their perceptions (Meyer, 2013). The item used constitutes a limited measure of the construct described in the model.

Although some of the hypotheses were not supported, the results of the present study have implications for clinical and educational practice as well as for future research. The results showed that sexual orientation was significantly associated with suicide risk behaviors independent of peer victimization, anti-LGBT harassment, and fears about future violence. For this reason, clinicians should continue to view sexual minority adolescents as a particularly vulnerable population. Peer victimization and fears about future violence were also identified as correlates of suicide risk behaviors, demonstrating that educators and school officials need to focus on addressing bullying and related behaviors at school and ensure targets of victimization receive appropriate services.

Anti-LGBT harassment was significantly associated with suicide risk behaviors for both sexual minority and heterosexual adolescents. These results demonstrate that harassment based on sexual identity can lead to negative outcomes for all students. Therefore, educators and school officials should recognize this form of victimization as particularly destructive and take measures to prevent it in educational settings for the benefit of all students.

Future research is needed to elucidate the relationships among sexual orientation, general and LGBT-specific peer victimization, expectations of victimization, and suicide risk behaviors. Several specific directions are indicated by the results of this study. Longitudinal research is needed to explore possible causal relationships between these variables. In addition, more research is needed to investigate the intersection of sexual and racial minority identities among adolescents, such as whether the associations between sexual minority stress processes and suicide risk operate differently for ethnic minority youth than for white youth. Finally, the use of more robust measures of these key constructs would lead to a greater understanding of the specific types of experiences associated with suicide risk. For example, anti-LGBT victimization could be measured through experiences of harassment, violence, and bullying; different forms of victimization may differ in their relationships with suicide risk.

Public Significance Statement.

The results of this study suggest that being harassed based on actual or perceived lesbian, gay, bisexual, or transgender status is associated with suicide risk behaviors among a predominantly ethnic minority sample of high school students. In addition, we found evidence of a positive association between fear of future violence at school and suicide risk for this population.