Bullying and depression serially mediate the association between perceived gender nonconformity and suicidality among US adolescents: a theory driven intersectional analysis

Abstract

Background

Adolescent suicidality has become an alarming public health concern. Perceived gender nonconformity has been identified as a risk factor for suicidality. Several factors (e. g. bullying and depression) have been posited as potential mechanisms through which perceived gender nonconformity is associated with suicidality; however, nascent research indicates instability in these associates due to youth’s exposure to intersectional forms of marginalization. This study aimed to examine an intersectional serial mediation model of adolescent suicidality by investigating the consistency of bullying and depression as prominent potential mediating mechanisms.

Methods

Hypotheses were tested using data derived from youths who participated in the 2019 Youth Risk Behavior Surveillance System (n = 70,047). We constructed an initial serial multiple mediation model that included all participants to examine whether the association of perceived gender nonconformity with suicidality among youth was mediated by bullying and/or depression. The goal of this model was to investigate the total and direct effects, reflected by the standardized regression coefficient and significance among the independent and dependent variables, and to assess 3 indirect effects, which that showed a change in suicidality for every 1-unit change in perceived gender nonconformity that the potential mediator mediated. We then conducted a multigroup analysis, using youth’s intersectional identities as a group variable.

Results

Overall, perceived gender nonconformity was positively associated with suicidality. This association was serially mediated by bullying and depression; however, these effects varied by youth’s intersectional social location. The direct association between perceived gender nonconformity and suicidality consistently emerged among all straight youth groups, with Black Straight Females and Males being notable exceptions. The full hypothesized serial mediation model could only be reproduced among Hispanic Straight Males; however, a partial mediation via bullying was demonstrated among White Bisexual Males.

Conclusions

This research has implications for understanding the potential underlying mechanisms that link perceived gender nonconformity to suicidality among adolescents. The hypothesized cascade of contextual risk factors for adolescent suicidality seems to be more harmful among Hispanic Straight Males. Studying contextual mechanisms can help develop therapeutic interventions that target adolescents most at-risk of suicidality.

Introduction

Suicide is the second-leading cause of death among people aged 15 to 24 in the U.S [1]. Rates of suicide have increased from 8.2 per 100,000 in 2018 to 11.2 per 100,000 in 2021, a rise of 36.6% [1]. Prior research has consistently demonstrated substantial health disparities by racial/ethnic identity and sexual orientation in suicidality, with racial/ethnic and sexual minority youth having higher prevalence of suicidality than their white or heterosexual peers [2, 3]. From 1999 to 2020, suicide-related mortality rates have steadily increased among Black, Asian/Pacific Islander, American Indian/Alaska Native, and Latinx communities, in comparison to White communities where rates have decreased [4]. A recent review of the literature indicated that sexual minority youth are at least twice as likely as heterosexual youth to contemplate suicide, and 2 to 7 times as likely to attempt suicide [5]. Considering the public health significance of adolescent suicidality and clear racial/ethnic and sexual minority disparities, an examination of the psychosocial antecedents of suicidality is warranted.

Gender identity-related stress and adolescent suicidality

Prior studies examining the antecedents of adolescent suicidality have identified gender identity-related stress as a psychosocial antecedent [6, 7]. Gender identity typically pertains to the internal and personal perception of one's gender, while gender expression relates to the way an individual publicly showcases or presents their gender [8]. Traditionally, gender identity and expression have been understood as binary concepts, classifying individuals as either male or female [8]. Terms such as"gender diversity"or"transgender"have emerged to encompass an individual's journey along a spectrum of potentialities beyond the traditional binary of male and female [9]. These terms aim to express the disparity between the sex assigned at birth and the present gender identity, signifying a gender identity that does not fully conform (and therefore nonconforming) [9]. Emerging research indicates a positive association between being perceived as gender nonconforming, i. e. being perceived as not conforming to tradition expressions of one’s assumed sex or gender, and adolescent suicidality [10, 11]. Gender theorists argue that the role of gender nonconformity as a predictor of suicidality is less representative of youth’s intrapsychic journey of discovery and instead a product of social gender policing. The notion that how one’s identity is perceived—or socially assigned— is not a novel concept but has only recently been highlighted as an important health determinant. Jones et al. examined socially assigned race and health status using the Behavioral Risk Factor Surveillance System (BRFSS), an ongoing state-based health survey administered in the U.S. to individuals aged 18 and older and found that being perceived as White was linked with better health outcomes in individuals who self-identified as a racial/ethnic minority [12]. Given that perceived gender nonconformity, like socially assigned race, may be an external cue used to judge an individual, it follows that socially assigned gender expression could be similarly linked to health disparities among adolescents.

Perceived gender nonconformity and minority stress theory

Perceived gender nonconformity remains a severely understudied vulnerability among youth, yet emergent research has linked this factor to major suicide outcome measures. For example, using the Youth Risk Behavior Surveillance (YRBS), Spivey and Prinstein found that gender nonconformity predicts higher odds of reporting suicidal ideation, suicide plans, and multiple suicide attempts among adolescents [13]. While novel in their attempts, this emergent area of adolescent research seldom explains this robust association. To address this gap, we rely on the theoretical lens of minority stress theory (MST) which links health disparities to excess exposure to social stress faced by minoritized and stigmatized populations [14]. Accordingly, MST suggests that those with stigmatized identities (e.g., gender minorities) are more likely to engage in health-harming behaviors (e.g., suicidality) [14]. This link is further exacerbated for those with multiple axes of marginalization (i.e., those who identify as both a racial/ethnic minority and a sexual minority), fewer resources, and less motivation to properly engage with health promoting-coping strategies. Intersectionality theory, which posits that individuals exist at the intersection of their social identities (e.g., race and sexual orientation) and their experiences are dependent upon the unique combination of privilege (or lack thereof) those identities afford them, is frequently invoked as a potential mechanism for understanding how multiple axes of marginalization synergize to convey greater or lesser degrees of adolescent suicidality risk [15]. For instance, a small but growing body of research has found that sexual minority youth of color are typically at higher risk of suicidal ideation than their heterosexual and White peers [16]. That said, more work is needed to fully comprehend suicidality risk across diverse groups of adolescents.

Bullying and depression

Prior literature suggests that exposure to bullying, and depression may be potential salient mechanisms by which perceived gender nonconformity is associated with adolescent suicidality. Although bullying can impact any adolescent regardless of sociodemographic and cultural identity, research has demonstrated that bullying is frequently motivated by discriminatory beliefs and attitudes that target adolescents who are perceived to be affiliated with certain marginalized groups or stigmatized identity statuses [17–19]. Racial or ethnic minority youth, sexual and gender minority youth, immigrant youth, youth with disabilities, and youth living in poverty have largely been identified as being more likely to experience victimization in school settings than their majority peers [20, 21].

At the same time, visibility of minority group status has been identified as an influential factor in exposure to victimization [22]. Highly visible members of minority groups, such as darker skinned racial minorities or individuals with visible disabilities, are more likely to experience bias-based interpersonal victimization than less visible members of minority groups, such as lighter skinned racial minorities or individuals with invisible disabilities [23, 24]. Emerging research has recognized this phenomenon in transgender youth as well. Visual gender conformity has been evidenced to impact risk of exposure to bullying and victimization among gender minority youth, with transgender adolescents who report being socially out at school and transfeminine adolescents who report not being able to pass as cisgender being most likely to experience bullying [25, 26].

Perceived gender nonconformity and suicidality among multiply marginalized adolescents

Due to visibility of minoritized attributes and perceived minority group status, adolescents with higher perceived gender nonconformity seem to be at greater risk of bullying. This increased risk of bullying significantly impacts mental health outcomes including alcohol and cannabis use [27, 28]. Research has consistently demonstrated that adolescents who experience bullying are at greater risk of developing depression and suicidality in both the short- and long-term [29, 30]. Other predictors of depression and suicidality include alcohol and cannabis use and exposure to sexual violence or forced sex [31]. Exposure to bullying, and the consequential stress it brings, causes psychological disruption in daily life and can, therefore, lead to depression, and even escalated levels of suicidality [32]. In line with this perspective, gender nonconforming individuals, or those who are perceived to be so, may face bullying from an early age due to gender socialization norms being taught and enforced very early in life. Consequently, these individuals may experience prolonged levels of elevated stress that may impact severity of depression and suicidality in this population [10]. Despite these evidenced associations, no study to date has examined the serial mediation that may exist between perceived gender nonconformity, bullying, depression, and suicidality, nor have they examined the degree to which youth’s social location (i.e., their identified race, sex, and sexual orientation) may amplify or reduce the direct and indirect pathways linking perceived gender nonconformity to adolescent suicidality.

The present study

Informed by prior empirical research and critical gender theorizing, we hypothesized that being perceived as not conforming to one’s assumed sex or gender (i.e. gender nonconforming), regardless of its accuracy, would expose adolescents to increased levels of bullying and feelings of depression thus increasing their risk of suicidality (i.e., perceived gender nonconformity → bullying → depression → suicidality). We also hypothesized that youth’s intersectional social location would influence the presence, and strength of all associations. The current study contributes to the knowledge on the potential mechanisms that proliferate suicidality among adolescents by examining a multivariate contextual vulnerability model that includes complex mediations and considers unique intersectional experiences.

Methods

Participants

Participants were drawn from the Local YRBS conducted by the Centers for Disease Control and Prevention (CDC). In its implementation, jurisdictions used a two-stage cluster sample designed to identify a sample of students [33]. In the first stage, schools were selected with a probability proportional to their enrollment; in the second stage, classes of a required subject or during a required period were randomly selected, and all students within these classes were eligible to participate. We limited our sample to youth who participated in the YRBS in 2019 (n = 247,381). We elected to use the 2019 YRBS instead of the 2021 YRBS because fewer local jurisdictions elected to query about perceived gender nonconformity in 2021 (1 local jurisdiction) than in 2019 (23 local jurisdictions). We further excluded respondents who had missing data on any major demographic variables of interest, resulting in a final sample of 70,047 high school-aged youth. Further information on the YRBS and its associated questionnaire can be found elsewhere [34]. Because this study uses deidentified secondary data, it was deemed exempt by the Institutional Review Board of < redacted for review >. This study was not preregistered.

Measures

Perceived Gender Nonconformity (PGNC)

Participants were asked, “A person’s appearance, style, dress, or the way they walk, or talk may affect how people describe them. How do you think people at school would describe you?” Response options were “very feminine,” “mostly feminine,” “somewhat feminine,” “equally feminine and masculine,” “somewhat masculine,” “mostly masculine,” and “very masculine.” Based on a youth’s response to this and to the question “What is your sex?” (Response options were “female” or “male”), a 7-point PGNC scale was created. Youth were categorized from most gender conforming (1, indicating very feminine female students and very masculine male students) to most gender nonconforming (7, indicating very masculine female students and very feminine male students).

Bullying

Three forms of victimization were measured: bullying at school, cyberbullying, and engagement in a physical fight. Each bullying indicator was assessed with a single item. that had the response options of “yes” or “no” (yes = 1, no = 0). The wording for the first two items were as follows: if they had ever been bullied on school property during the past 12 months (i.e. bullying at school), and if they had been electronically bullied during the past 12 months (i.e. cyberbullying). The response options for these items were “yes” or “no” (yes = 1, no = 0). Youth then reported how many times they had engaged in a physical fight in the past 12 months. Reponses options to this question were then dichotomized to 0 (no incidents) and 1 (any incidents). All three of these dichotomous indicators were summed to create a bullying index with a range of 0–3, wherein higher scores were indicative of more exposure to bullying in the past 12 months.

Depression

Participants were asked, “During the past 12 months, did you ever feel so sad or hopeless almost every day for two weeks or more in a row that you stopped doing some usual activities?” Youth who reported one or more times within the last year were coded as having experienced depressive symptoms (yes = 1, no = 0).

Suicidality

Three indicators of suicidality were measured: suicidal ideation, suicidal plan, and suicidal attempts. Each suicidality indicator was assessed with a single item that had the response options of “yes” or “no” (yes = 1, no = 0). Suicidal ideation was evaluated by asking youth whether they had seriously thought about killing themselves in the past 12 months. Suicide planning was assessed was evaluated by asking youth if they had ever planned about how they would end their lives in the last 12 months. Suicide attempts were evaluated by asking youth if they had ever attempted to end their lives in the previous 12 months. All three of these dichotomous indicators were summed to create a suicidality index with a range of 0–3, wherein higher scores were indicative of greater degrees of suicidality in the past 12 months.

Intersectional social location

Three axes of social location were used to categorize participants (sex, race/ethnicity, and sexual identity). Youth reported their sex as either 1 (male) or 2 (female). Race/ethnicity was reported as 1 (white), 2 (Black), 3 (Hispanic), or 4 (Not Specified Race). Sexual identity was assessed with the question “Which of the following best describes you?” Response options were “heterosexual (straight),” “lesbian or gay,” “bisexual,” and “not sure sexual identity.” These variables were combined to create 32 different intersectional identity categories.

Age

Age was measured using an interval age variable (i.e., How old are you?).

Forced sex

History of forced sexual intercourse was measured based on response to the question “Have you ever been physically forced to have sexual intercourse when you did not want to?” Adolescents who answered “yes” were coded 1 whereas “no” were coded 0.

Alcohol use

Alcohol use was measured based on the question “During your life, on how many days have you had at least one drink of alcohol?” Adolescents who reported drinking alcohol at least once were coded 1; otherwise, they were coded 0.

Cannabis use

Cannabis use was measured based on the question “During your life, how many times have you used marijuana?” Adolescents who reported using cannabis at least once were coded 1; otherwise, they were coded 0.

Sleep

Sleep duration was measured using a single item in which youth reported how many hours of sleep they get on an average school night. Due to the robust association between sleep and suicidality identified in prior YRBS studies, we have included sleep as a discrete covariate in all analyses.

Statistical analysis

Descriptive statistics and correlations were calculated using SPSS. Bivariate associations among study variables were assessed using Pearson’s r coefficients for continuous-to-continuous variable associations, point-biserial coefficients for dichotomous-to-continuous variable associations, and phi coefficients for dichotomous-to-dichotomous variable associations [35]. Study hypotheses were tested using a multigroup serial multiple mediation model in the structural equation modeling (SEM) framework in Mplus 8.1 [36]. To account for the complex sampling structure in the YRBS, we specified stratification by study site and clustering students with sites (using TYPE = COMPLEX) and utilized the maximum likelihood estimation with robust standard errors (MLR) in all models to yield unbiased estimates. To evaluate model fitness, we used Confirmatory Fit Index (CFI) values greater than or equal to 0.90, Root Mean Square Error of Approximation (RMSEA) values less than or equal to 0.08, and Standardized Root Mean Square Residual (SRMR) values less than or equal to 0.08 as indicators of well-fitting models [37].

Hayes posits that a variable may play a mediatory or indirect role between two variables, which do not initially appear associated [38]. He accounts for the possibility of an indirect effect by observing that the total effect is a sum of many different paths and may include two or more indirect paths working in opposite directions [38]. In accordance, we tested our first three hypotheses by constructing an initial serial multiple mediation model that included all participants to examine whether the association of perceived gender nonconformity with suicidality among youth was mediated by bullying and/or depression. The goal of this model was to investigate the total and direct effects, reflected by the standardized regression coefficient and significance among the independent and dependent variables, and to assess 3 indirect effects, which that showed a change in suicidality for every 1-unit change in perceived gender nonconformity that was mediated by the potential mediator:

• IE₁: perceived gender nonconformity → bullying → suicidality

• IE₂: perceived gender nonconformity → depression → suicidality

• IE₃: perceived gender nonconformity → bullying → depression → suicidality.

We then conducted a multigroup analysis, using youth’s intersectional identities as a group variable, to examine our fourth hypothesis. We opted to conduct a multigroup analysis, over other techniques, because it allowed us to (a) test predefined (also known as a priori) data groups, in this case intersectional identity groups, to determine the existence of significant differences across group-specific parameter estimates (e.g., outer weights, outer loadings, and path coefficients) and (b) test for variations between different groups in identical models in order to assess the performance of these models [39]. To conduct our multigroup analysis, we first constructed a model where all parameters were constrained to be equal across all intersectional social location groups (e.g., White Straight Female, Black Bisexual Male). Next, a multigroup model that allowed all parameters to vary across groups was tested. Within all models, we considered several confounding variables to strengthen analytic conclusions, including age, and sleep capacity. When comparing the performance of constrained versus unconstrained multigroup models, the log-likelihood difference test was used. Concerning this test, a p-value of 0.05 or below would be indicative of the nested model being a better fitting model than the alternative. We also utilized measures of change in fit indices to evaluate model performance. Following the recommendation by Chen, the criteria used to evaluate change in fit indices was a change in CFI of 0.01, supplemented by a change in SRMR of 0.01 between the nested model and the alternative model [40].

Results

Preliminary analyses

Table 1 shows the intersectional social location sample sizes, demographic information, descriptive statistics of the study sample. Table 2 provides a summary of means, standard deviations, and correlations for each of the study variables. As seen in Table 2, each of the variables evaluated were significantly correlated.

Table 1.

Sample demographics

	Unweighted		Weighted %
	n	%
Perceived Gender Nonconformity (PGNC)
1—Perceived as very gender conforming	4038	26.2	25.8
2	5935	38.5	39.0
3	2413	15.6	14.9
4	2139	13.9	13.3
5	400	2.6	2.7
6	264	1.7	1.4
7—Perceived as very gender nonconforming	234	1.5	2.8
Suicidality
0	45588	76.7	76.9
1	6157	10.4	9.9
2	4229	7.1	7.4
3	3428	5.8	5.8
Bullying
0	39562	64.3	63.8
1	14289	23.2	23.6
2	5825	9.5	9.7
3	1890	3.1	3.0
Depression
No	45017	64.6	62.8
Yes	24651	35.4	37.2
Intersectional Identity
White Straight Female	13310	19.0	18.3
White Lesbian Female	475	0.7	0.5
White Bisexual Female	2311	3.3	3.0
White Not Sure Female	833	1.2	1.1
Black Straight Female	4451	6.4	5.0
Black Lesbian Female	234	0.3	0.2
Black Bisexual Female	805	1.1	1.0
Black Not Sure Female	292	0.4	0.3
Hispanic Straight Female	7447	10.6	12.5
Hispanic Lesbian Female	260	0.4	0.4
Hispanic Bisexual Female	1497	2.1	2.1
Hispanic Not Sure Female	596	0.9	0.9
Not Specified Straight Female	3124	4.5	3.8
Not Specified Lesbian Female	149	0.2	0.1
Not Specified Bisexual Female	690	1.0	0.8
Not Specified Not Sure Female	322	0.5	0.4
White Straight Male	14756	21.1	21.3
White Gay Male	377	0.5	0.5
White Bisexual Male	690	1.0	1.0
White Not Sure Male	443	0.6	0.6
Black Straight Male	4292	6.1	5.6
Black Gay Male	142	0.2	0.2
Black Bisexual Male	147	0.2	0.2
Black Not Sure Male	128	0.2	0.2
Hispanic Straight Male	7674	11.0	13.6
Hispanic Gay Male	234	0.3	0.5
Hispanic Bisexual Male	309	0.4	0.5
Hispanic Not Sure Male	278	0.4	0.5
Not Specified Straight Male	3345	4.8	4.4
Not Specified Gay Male	95	0.1	0.1
Not Specified Bisexual Male	161	0.2	0.2
Not Specified Not Sure Male	180	0.3	0.2
Age
1. 12 years old or younger	125	0.2	0.2
2. 13 years old	127	0.2	0.2
3. 14 years old	9174	13.1	12.2
4. 15 years old	18855	26.9	24.5
5. 16 years old	18224	26.0	25.8
6. 17 years old	15647	22.3	23.8
7. 18 years old or older	7895	11.3	13.3
Sleep
1. 4 or less hours	7248	10.3	10.3
2. 5 h	10541	15.0	15.7
3. 6 h	17191	24.5	24.4
4. 7 h	19104	27.3	27.3
5. 8 h	12173	17.4	17.2
6. 9 h	2769	4.0	3.8
7. 10 or more hours	1021	1.5	1.4
Alcohol Use
No	34502	49.3	47.6
Yes	35545	50.7	52.4
Cannabis Use
No	44542	65.6	66.3
Yes	23408	34.4	33.7
Forced Sex
No	64300	91.8	91.5
Yes	5747	8.2	8.5

Table 2.

Bivariate correlations among study variables, along with Means and SDs

		1	2	3	4	5	6	7	8	9
1	PGNC	1
2	Bullying	.12**	1
3	Depression*	.14**	.29**	1
4	Suicidality	.21**	.33**	.45**	1
5	Age	-.02**	-.07**	.02**	-.01**	1
6	Sleep	-.10**	-.13**	-.21**	-.19**	-.11**	1
7	Alcohol Use*	-.05**	.16**	.19**	.16**	.17**	-.11**	1
8	Cannabis Use*	-.02**	.16**	.19**	.18**	.17**	-.11**	.50**	1
9	Forced Sex*	.06**	.20**	.19**	.27**	.06**	.09**	.11**	.15**	1
M		2.44	.52	.37	.42	5.00	3.42	.52	.34	1.92
SD		1.39	.79	.48	.86	1.24	1.39	.50	.47	.28

All correlations are two-tailed and weighted, M Mean, SD Standard Deviation

^**p <.001; *denotes point-biserial correlation or phi coefficients

Initial serial mediation model

The next step in our analyses was to test the fit of our hypothesized model to the full sample of youth (see Fig. 1). As a result of the exploratory and cross-sectional nature of this study all covariates were regressed on all independent and dependent variables to isolate our specific effects of interest, which resulted in perfect model fit: χ²(0) = 0.00, p < 0.001, scaling correction factor = 1.00; RMSEA = 0.00, CFI = 1.00, TLI = 1.00, and SRMR = 0.00.

Fig. 1.

Visual depiction of the standardized serial multiple mediation model showing the association between perceived gender nonconformity and suicidality with bullying and depression as mediators, adjusting for age, exposure to forced sex, alcohol use, cannabis use, and sleep. Standardized regression coefficients are presented. *** p <.001

Perceived gender nonconformity was positively associated with bullying (ß = 0.05, p < 0.001, 95% CI 0.03, 0.08), depression (ß = 0.03, p < 0.001, 95% CI 0.02, 0.04), and suicidality (ß = 0.09, p < 0.001, 95% CI 0.07, 0.11). Bullying was positively associated with depression (ß = 0.13, p < 0.001, 95% CI 0.12, 0.14), and suicidality (ß = 0.18, p < 0.001, 95% CI 0.16, 0.21). Depression was positively associated with suicidality (ß = 0.59, p < 0.001, 95% CI 0.55, 0.62).

The association between perceived gender nonconformity and suicidality (B_T = 0.12, p < 0.001, 95% CI 0.10, 0.15) was partially mediated by bullying and depression with remaining unexplained direct effect, after accounting for the serial mediation, being 0.033 (p < 0.001, 95% CI 0.02, 0.04). The total IEs included 3 significant pathways: (1) IE₁: 0.009, p < 0.001, 95% CI 0.005, 0.014; (2) IE₂: 0.020, p < 0.001, 95% CI 0.014, 0.026; (3) IE₃: 0.004, p < 0.001, 95% CI 0.002, 0.006). IE₁, IE₂, and IE₃ accounted for 27.27%, 60.60%, and 12.12% of the total effect, respectively.

Intersectional multigroup model

We next investigated potential differences in the model based on youth’s intersectional identities. We began by constructing a model where all parameters were held constant across all groups. This model fit the data as follows: χ²(806) = −1965.931, p < 0.001, scaling correction factor = 3.61; Loglikelihood (282) = −214764.677, scaling correction factor = 4.90, RMSEA = 0.03, CFI = 0.88, TLI = 0.88, and SRMR = 0.07. We then compared the fit indices of this model to one in which all the parameters were held constant across all groups. This model fit the data as follows: χ²(0) = 0.183, p < 0.001, scaling correction factor = 1.00; Loglikelihood (1088) = −211219.189, scaling correction factor = 3.94; RMSEA = 0.00, CFI = 1.00, TLI = 1.00, and SRMR = 0.003. We found that while both models had good fit statistics, the unconstrained model performed better based on the RMSEA, CFI, TLI, and SRMR statistics and produced a significant log-likelihood difference value (D (282) = 87313.85, p > 0.001), leading to the conclusion that the unconstrained model, where all paths were free to vary across groups, was a better fit. This also informed us that a singular serial mediation model could not be assumed across all groups.

Group-specific serial mediation model parameters are presented in Table 3 (unstandardized main effects) and Table 4 (unstandardized indirect effects). We discuss significant effects below and include standardized main and indirect coefficients in text.

Table 3.

Unstandardized main effects of the serial mediation model

	Perceived Gender Nonconformity → Suicidality					Bullying → Suicidality					Depression → Suicidality					Bullying → Depression					Perceived Gender Nonconformity → Depression					Perceived Gender Nonconformity → Bullying
				95% CI					95% CI					95% CI					95% CI					95% CI					95% CI
	B	S.E	P-Value	Lower	Upper	B	S.E	P-Value	Lower	Upper	B	S.E	P-Value	Lower	Upper	B	S.E	P-Value	Lower	Upper	B	S.E	P-Value	Lower	Upper	B	S.E	P-Value	Lower	Upper
White Straight Female	.07	.03	.032	.006	.126	.20	.02	.000	.154	.245	.45	.03	.000	.390	.516	.14	.01	.000	.115	.159	.01	.02	.555	-.021	.038	-.01	.03	.690	-.062	.041
White Lesbian Female	.12	.17	.459	-.205	.453	.32	.14	.018	.056	.586	.66	.21	.001	.257	1.070	.19	.06	.003	.063	.308	.09	.06	.117	-.023	.209	-.18	.08	.018	-.324	-.031
White Bisexual Female	.13	.08	.078	-.015	.278	.22	.06	.000	.096	.341	.76	.10	.000	.558	.952	.06	.02	.000	.030	.094	.03	.03	.227	-.021	.090	-.02	.09	.842	-.189	.154
White Not Sure Female	.27	.20	.179	-.125	.667	.23	.11	.035	.016	.450	.43	.16	.008	.113	.741	.18	.04	.000	.109	.256	.07	.05	.162	-.028	.165	.03	.11	.773	-.189	.254
Black Straight Female	.04	.04	.358	-.044	.123	.19	.05	.000	.097	.281	.47	.05	.000	.368	.567	.13	.02	.000	.089	.167	.02	.02	.334	-.022	.065	-.02	.04	.655	-.099	.063
Black Lesbian Female	-.42	.11	.000	-.629	-.217	.46	.17	.007	.122	.791	.87	.26	.001	.353	1.390	-.03	.07	.626	-.168	.101	.08	.06	.128	-.024	.191	.26	.09	.004	.082	.427
Black Bisexual Female	.39	.13	.003	.133	.642	.31	.09	.000	.139	.490	.83	.16	.000	.518	1.140	.15	.04	.000	.070	.220	.08	.05	.097	-.014	.167	-.17	.08	.041	-.332	-.007
Black Not Sure Female	.10	.12	.427	-.142	.335	.46	.10	.000	.270	.659	.60	.18	.001	.252	.944	.16	.06	.008	.041	.279	-.15	.08	.062	-.299	.007	-.05	.11	.629	-.267	.161
Hispanic Straight Female	.09	.04	.015	.018	.167	.31	.05	.000	.206	.409	.46	.05	.000	.360	.565	.16	.02	.000	.127	.196	-.02	.02	.274	-.055	.016	-.04	.03	.186	-.096	.019
Hispanic Lesbian Female	.35	.32	.282	-.284	.976	.45	.22	.040	.021	.875	.91	.21	.000	.492	1.331	.17	.05	.001	.071	.274	.06	.07	.370	-.073	.195	-.23	.09	.012	-.403	-.050
Hispanic Bisexual Female	.12	.10	.207	-.066	.306	.28	.09	.001	.114	.450	.84	.13	.000	.586	1.094	.08	.03	.010	.018	.133	-.02	.04	.700	-.090	.061	-.16	.10	.120	-.351	.041
Hispanic Not Sure Female	.31	.11	.006	.088	.524	.42	.13	.001	.173	.670	.42	.31	.165	-.174	1.022	.16	.06	.011	.037	.283	.11	.04	.008	.028	.181	-.08	.11	.470	-.292	.135
Not Specified Straight Female	-.02	.06	.705	-.133	.090	.07	.06	.208	-.038	.176	.54	.08	.000	.369	.700	.14	.03	.000	.090	.193	-.02	.03	.508	-.081	.040	-.04	.06	.520	-.144	.073
Not Specified Lesbian Female	-.55	.05	.000	-.647	-.446	.74	.13	.000	.475	1.002	1.24	.12	.000	1.010	1.476	.02	.08	.773	-.128	.173	.03	.04	.441	-.049	.112	.18	.11	.097	-.032	.389
Not Specified Bisexual Female	.13	.20	.499	-.251	.514	.34	.13	.008	.090	.590	.90	.13	.000	.638	1.163	.19	.05	.000	.087	.297	-.05	.06	.444	-.168	.073	.25	.14	.071	-.021	.514
Not Specified Not Sure Female	-.33	.28	.233	-.868	.211	.30	.13	.023	.040	.552	.65	.22	.003	.224	1.072	.18	.05	.000	.088	.266	.09	.09	.282	-.077	.266	-.05	.08	.527	-.203	.104
White Straight Male	.08	.03	.007	.021	.135	.07	.02	.000	.035	.111	.59	.04	.000	.516	.659	.11	.01	.000	.088	.130	.03	.02	.066	-.002	.056	.05	.03	.134	-.015	.111
White Gay Male	-.06	.14	.697	-.334	.223	.39	.13	.002	.147	.641	.75	.14	.000	.486	1.013	.07	.06	.241	-.049	.195	-.02	.07	.784	-.163	.123	.18	.10	.061	-.008	.364
White Bisexual Male	.12	.12	.307	-.110	.348	.31	.10	.003	.102	.509	.79	.16	.000	.475	1.106	.01	.05	.919	-.091	.101	.12	.06	.038	.006	.226	.33	.08	.000	.171	.485
White Not Sure Male	-.19	.09	.038	-.371	-.011	.32	.10	.001	.125	.519	.26	.22	.229	-.165	.687	.18	.04	.000	.107	.257	-.09	.05	.082	-.183	.011	-.01	.13	.928	-.262	.239
Black Straight Male	-.02	.03	.517	-.089	.045	.08	.04	.028	.008	.153	.48	.08	.000	.327	.639	.13	.02	.000	.087	.167	.01	.01	.386	-.014	.037	-.01	.03	.804	-.057	.044
Black Gay Male	.43	.10	.000	.239	.620	.23	.24	.338	-.235	.686	.17	.32	.594	-.454	.793	.14	.08	.099	-.026	.304	.05	.06	.426	-.073	.173	.12	.11	.267	-.092	.330
Black Bisexual Male	1.04	.25	.000	.559	1.528	.59	.31	.060	-.025	1.195	.82	.38	.031	.077	1.567	.24	.10	.015	.046	.426	.13	.10	.189	-.065	.332	-.49	.18	.006	-.840	-.141
Black Not Sure Male	.28	.18	.122	-.076	.643	-.28	.30	.357	-.877	.316	.18	.45	.681	-.693	1.061	-.59	.19	.002	-.956	-.225	.36	.06	.000	.230	.479	.36	.07	.000	.222	.494
Hispanic Straight Male	.08	.03	.003	.028	.134	.19	.04	.000	.108	.264	.50	.06	.000	.388	.609	.16	.02	.000	.117	.196	.05	.01	.000	.028	.074	.15	.02	.000	.107	.184
Hispanic Gay Male	.26	.13	.045	.006	.522	.20	.19	.299	-.175	.569	.50	.23	.029	.050	.946	-.01	.09	.865	-.181	.152	.03	.06	.632	-.085	.139	.27	.10	.005	.081	.456
Hispanic Bisexual Male	-.41	.09	.000	-.573	-.240	.22	.18	.214	-.127	.565	1.05	.19	.000	.681	1.413	.18	.05	.000	.092	.276	.04	.05	.415	-.054	.130	.09	.13	.489	-.168	.351
Hispanic Not Sure Male	-.36	.12	.003	-.590	-.122	.21	.11	.046	.004	.422	-.37	.25	.133	-.855	.113	.18	.07	.007	.050	.314	-.16	.05	.001	-.247	-.064	.10	.08	.211	-.054	.243
Not Specified Straight Male	.10	.06	.075	-.010	.217	.09	.06	.098	-.017	.206	.64	.09	.000	.465	.807	.12	.03	.000	.064	.177	.05	.03	.071	-.004	.094	-.04	.05	.446	-.136	.060
Not Specified Gay Male	.44	.37	.227	-.274	1.157	.06	.41	.888	-.747	.863	.00	.70	.995	−1.365	1.374	.05	.11	.638	-.170	.278	.20	.09	.020	.031	.369	-.15	.16	.335	-.455	.155
Not Specified Bisexual Male	.48	.19	.012	.103	.851	.00	.23	.999	-.447	.446	1.37	.41	.001	.566	2.178	.24	.09	.010	.056	.425	-.15	.08	.058	-.311	.005	.43	.10	.000	.242	.614
Not Specified Not Sure Male	.17	.18	.355	-.187	.521	.33	.11	.003	.111	.540	.68	.33	.040	.031	1.333	.08	.08	.340	-.079	.229	-.03	.06	.606	-.141	.082	.05	.12	.658	-.185	.294

Significant effects are bolded

SE standard error, 95% CI 95% Confidence Interval

Table 4.

Unstandardized indirect effects of the serial mediation model

	Total Indirect Effects					Perceived Gender Nonconformity → Bullying → Suicidality					Perceived Gender Nonconformity → Depression → Suicidality					Perceived Gender Nonconformity → Bullying → Depression → Suicidality
				95% CI					95% CI					95% CI					95% CI
	Total	S.E	P-Value	Lower	Upper	IE1	S.E	P-Value	Lower	Upper	IE2	S.E	P-Value	Lower	Upper	IE3	S.E	P-Value	Lower	Upper
White Straight Female	.001	.010	.898	-.018	.021	-.002	.005	.693	-.012	.008	.004	.007	.554	-.009	.017	-.001	.002	.692	-.004	.003
White Lesbian Female	-.017	.048	.724	-.112	.078	-.057	.030	.062	-.117	.003	.062	.041	.129	-.018	.141	-.022	.017	.188	-.054	.011
White Bisexual Female	.021	.033	.520	-.043	.086	-.004	.020	.846	-.042	.035	.026	.021	.218	-.015	.067	-.001	.004	.846	-.009	.007
White Not Sure Female	.039	.041	.341	-.042	.121	.008	.025	.763	-.042	.057	.029	.022	.189	-.014	.073	.003	.009	.773	-.015	.020
Black Straight Female	.005	.015	.711	-.023	.034	-.003	.008	.649	-.018	.012	.010	.010	.333	-.010	.030	-.001	.003	.664	-.006	.004
Black Lesbian Female	.182	.074	.015	.036	.327	.116	.078	.134	-.036	.268	.073	.064	.259	-.054	.199	-.007	.017	.660	-.040	.026
Black Bisexual Female	-.010	.051	.844	-.110	.090	-.053	.033	.102	-.117	.011	.063	.033	.052	-.001	.128	-.020	.013	.126	-.046	.006
Black Not Sure Female	-.117	.079	.140	-.272	.038	-.025	.051	.632	-.125	.076	-.087	.057	.128	-.199	.025	-.005	.010	.626	-.025	.015
Hispanic Straight Female	-.024	.014	.090	-.052	.004	-.012	.010	.223	-.031	.007	-.009	.008	.281	-.026	.007	-.003	.002	.204	-.007	.002
Hispanic Lesbian Female	-.082	.105	.435	-.286	.123	-.102	.070	.146	-.239	.036	.056	.061	.359	-.063	.175	-.036	.020	.070	-.074	.003
Hispanic Bisexual Female	-.066	.053	.213	-.170	.038	-.044	.034	.198	-.111	.023	-.012	.033	.704	-.077	.052	-.010	.007	.176	-.024	.004
Hispanic Not Sure Female	.006	.054	.914	-.100	.112	-.033	.049	.498	-.129	.063	.044	.029	.122	-.012	.101	-.005	.010	.606	-.026	.015
Not Specified Straight Female	-.016	.019	.398	-.053	.021	-.002	.004	.575	-.011	.006	-.011	.017	.507	-.043	.021	-.003	.004	.524	-.011	.006
Not Specified Lesbian Female	.176	.112	.117	-.044	.396	.132	.097	.175	-.059	.322	.039	.052	.450	-.063	.141	.005	.018	.782	-.030	.040
Not Specified Bisexual Female	.084	.092	.362	-.097	.265	.084	.062	.173	-.037	.205	-.042	.057	.453	-.153	.068	.043	.031	.165	-.018	.103
Not Specified Not Sure Female	.041	.069	.557	-.095	.176	-.015	.023	.520	-.059	.030	.061	.063	.330	-.062	.184	-.006	.010	.567	-.025	.014
White Straight Male	.023	.009	.011	.005	.040	.004	.002	.121	-.001	.008	.016	.008	.060	-.001	.033	.003	.002	.126	-.001	.007
White Gay Male	.065	.088	.460	-.107	.237	.070	.052	.176	-.032	.172	-.015	.055	.784	-.122	.092	.010	.010	.320	-.009	.029
White Bisexual Male	.193	.059	.001	.078	.309	.100	.039	.009	.025	.176	.092	.049	.059	-.003	.187	.001	.013	.918	-.023	.026
White Not Sure Male	-.027	.050	.593	-.125	.072	-.004	.041	.927	-.083	.076	-.023	.016	.156	-.054	.009	-.001	.006	.928	-.013	.011
Black Straight Male	.005	.007	.541	-.010	.019	-.001	.002	.807	-.005	.004	.005	.006	.383	-.007	.018	.000	.002	.805	-.003	.003
Black Gay Male	.038	.026	.137	-.012	.088	.027	.025	.282	-.022	.076	.008	.013	.516	-.017	.034	.003	.006	.634	-.009	.014
Black Bisexual Male	-.272	.290	.347	-.841	.296	-.287	.233	.219	-.745	.171	.110	.068	.107	-.024	.243	-.095	.075	.205	-.242	.052
Black Not Sure Male	-.074	.155	.632	-.377	.229	-.100	.107	.349	-.310	.110	.065	.158	.680	-.245	.375	-.039	.097	.690	-.230	.152
Hispanic Straight Male	.064	.010	.000	.045	.083	.027	.007	.000	.014	.040	.026	.006	.000	.013	.038	.011	.003	.000	.006	.016
Hispanic Gay Male	.065	.055	.243	-.044	.173	.053	.052	.309	-.049	.155	.014	.028	.628	-.041	.069	-.002	.011	.861	-.024	.020
Hispanic Bisexual Male	.078	.096	.418	-.110	.266	.020	.041	.627	-.061	.101	.040	.053	.447	-.063	.143	.018	.027	.506	-.034	.070
Hispanic Not Sure Male	.071	.042	.088	-.011	.154	.020	.019	.293	-.017	.058	.058	.042	.167	-.024	.139	-.006	.007	.385	-.021	.008
Not Specified Straight Male	.022	.021	.287	-.019	.063	-.004	.006	.536	-.015	.008	.029	.016	.073	-.003	.060	-.003	.004	.465	-.011	.005
Not Specified Gay Male	-.008	.197	.968	-.395	.379	-.009	.067	.896	-.139	.122	.001	.140	.995	-.274	.276	.000	.006	.995	-.011	.011
Not Specified Bisexual Male	-.069	.145	.636	-.352	.215	.000	.098	.999	-.191	.191	-.210	.152	.168	-.508	.089	.141	.077	.065	-.009	.292
Not Specified Not Sure Male	.000	.064	.996	-.125	.126	.018	.041	.666	-.062	.098	-.020	.046	.661	-.110	.070	.003	.008	.733	-.013	.019

Significant effects are bolded

SE standard error, 95% CI 95% Confidence Interval

• White Straight Female (n = 13,310). Perceived gender nonconformity was positively associated with suicidality (ß = 0.09, p < 0.05, 95% CI 0.01, 0.17). No significant indirect association between perceived gender nonconformity and suicidality emerged.

• Black Lesbian Female (n = 234). Perceived gender nonconformity was negatively associated with suicidality (ß = −0.73, p < 0.001, 95% CI −1.02, −0.43). No significant indirect association between perceived gender nonconformity and suicidality emerged.

• Black Bisexual Female (n = 805). Perceived gender nonconformity was positively associated with suicidality (ß = 0.37, p < 0.01, 95% CI 0.12, 0.61).

• Hispanic Straight Female (n = 7,447). Perceived gender nonconformity was positively associated with suicidality (ß = 0.12, p < 0.05, 95% CI 0.02, 0.21).

• Hispanic Not Sure Female (n = 596). Perceived gender nonconformity was positively associated with suicidality (ß = 0.38, p < 0.05, 95% CI 0.09, 0.67).

• Not Specified Lesbian Female (n = 149). Perceived gender nonconformity was negatively associated with suicidality (ß = −0.93, p < 0.001, 95% CI −1.18, −0.68).

• White Straight Male (n = 14,756). Perceived gender nonconformity was positively associated with suicidality (ß = 0.13, p < 0.01, 95% CI 0.03, 0.23).

• White Bisexual Male (n = 690). A partial mediation whereby the indirect effect of perceived gender nonconformity on suicidality occurred through bullying emerged (B_Ind = 0.100, p < 0.05, 95% CI 0.025, 0.176).

• White Not Sure Male (n = 443). Perceived gender nonconformity was negatively associated with suicidality (ß = −0.38, p < 0.05, 95% CI −0.77, 0.00).

• Black Gay Male (n = 142). Perceived gender nonconformity was positively associated with suicidality (ß = 0.99, p < 0.001, 95% CI 0.58, 1.40).

• Black Bisexual Male (n = 147). Perceived gender nonconformity was positively associated with suicidality (ß = 1.77, p < 0.001, 95% CI 0.73, 2.81).

• Hispanic Straight Male (n = 7,674). Perceived gender nonconformity was positively associated with suicidality (ß = 0.20, p < 0.01, 95% CI 0.07, 0.33). The total IEs included 3 significant pathways: (1) IE₁: perceived gender nonconformity—bullying—suicidality (a₁*b₁ = 0.027, p < 0.001, 95% CI 0.014, 0.040); (2) IE₂: perceived gender nonconformity—depression—suicidality (a₂*b₂ = 0.026, p < 0.001, 95% CI 0.013, 0.038); (3) IE₃: perceived gender nonconformity—bullying—depression—suicidality (a₁*d₂₁*b₂ = 0.011, p < 0.001, 95% CI 0.006, 0.016). IE₁, IE₂, and IE₃ accounted for 42.19%, 40.63%, and 17.19% of the total effect, respectively.

• Hispanic Gay Male (n = 234). Perceived gender nonconformity was positively associated with suicidality (ß = 0.43, p < 0.05, 95% CI 0.00, 0.86).

• Hispanic Bisexual Male (n = 309). Perceived gender nonconformity was negatively associated with suicidality (ß = −0.55, p < 0.001, 95% CI −0.80, −0.31).

• Hispanic Not Sure Male (n = 278). Perceived gender nonconformity was negatively associated with suicidality (ß = −0.63, p < 0.01, 95% CI −1.10, −0.16).

• Not Specified Bisexual Male (n = 161). Perceived gender nonconformity was positively associated with suicidality (ß = 0.55, p < 0.05, 95% CI 0.12, 0.97).

Discussion

Consistent with our central hypothesis, we found that bullying and depression, respectively, serially mediated the association between perceived gender nonconformity and suicidality. Although ample research has identified the significant associations between bullying, depression, and suicidality among high school-aged youth, there has been a dearth of research that has sought to contextualize these associations with perceived gender nonconformity to understand differential risk experienced by certain groups of adolescents. Most of the limited research on perceived gender nonconformity and suicidality examines this relationship in consideration of either bullying or depression but often not in consideration of both; furthermore, much of this available research samples adults, not adolescents [41–43]. Our findings expand scholarly understanding in this area by evidencing both bullying and depression in adolescence as potential mechanisms of mediation in the association between perceived gender nonconformity and suicidality.

Our finding that perceived gender nonconformity is positively associated with bullying can be attributed to concurrent adolescent developmental processes. First, as youth move through adolescents they often turn towards peers, and away from nuclear family members, for external validation and esteem [44]. Second, adolescents may become more sensitive to societal expectations of gender presentation as they begin prepping for the new found independence and autonomy associated with emerging adulthood [44]. These concurrent processes often follow nonlinear trajectories, particularly in regard to youth’s relationship with gender presentation. For example, prior research suggests that the rigid following of gender presentation norms is initial high in early childhood, decreases in middle childhood, and increases in early adolescence [45]. These data coupled with our findings indicate that gender exploration that is tolerated in middle childhood—which is often minimized as childhood play—becomes maligned throughout adolescence as youth become more sensitive to the gender dynamics of their environment. Consequently, bullying may play a major role in this process by operating as a peer-focused form of gender policing, which refers to the enforcement of societal expectations concerning gender through acts of criticism, harassment, or even violence towards individuals who deviate from these expectations. For instance, perceived gender nonconformity during adolescence is associated with overt and relational peer victimization and aggression [46]. Further research suggest that this process of gender policing may also be influenced by assigned sex at birth. Toomey et al. found that perceived gender nonconformity is associated with problematic peer relations particularly among female adolescents [46]. Moreover, male youth with higher perceived gender nonconformity are at increased risk for bullying and cyberbullying, whereas female youth with higher perceived gender nonconformity are less likely to report cyberbullying [47]. In accordance with prior studies, our findings exhibit the importance of studying the proximal and distal effects of perceived gender nonconformity during adolescence, as it is a sensitive period of gender identity development and policing.

The direct association between perceived gender nonconformity and suicidality varied greatly by youth’s intersectional identity. This association was positive among (1) White Straight Females, (2) Black Bisexual Females, (3) Hispanic Straight Females, (4) Hispanic Not Sure Females, (5) White Straight Males, (6) Black Gay Males, (7) Black Bisexual Males, (8) Hispanic Straight Males, (9) Hispanic Gay Males, and (10) Not Specific Bisexual Males. Notably, the direct association between perceived gender nonconformity and suicidality was negative among (1) Black Lesbian Female, (2) Not Specific Bisexual Females, (3) White Not Sure Females, (4) Hispanic Bisexual Males, and (5) Hispanic Not Sure Males. The positive association between perceived gender nonconformity and suicidality may partially be explained by negative beliefs towards gender nonconformity that are deeply woven into the culture of the United States of America. Historically, gender nonconformity has often been a central trigger for antigay discrimination and violence across different time periods. During the twentieth century, individuals perceived as gender nonconforming were frequently targeted for discrimination, violent attacks, and police brutality [48, 49]. For instance, in early 20th-century New York, sexual minorities were widely recognized as distinct social types in the streets of working-class neighborhoods and were often targeted by youth gangs who dominated much of the criminal street activity [48]. Unlike same-gender attraction, which can often be concealed through a process known as passing—a strategy with its own mental health consequences—gender nonconformity is typically visible and thus a more easily targeted stigma [50–52]. Importantly, this stigma is based on perceived gender nonconformity, which may bear little relationship to a person’s actual gender identity or sexual orientation. Although same-gender attraction and relationships have become more socially acceptable in some contexts—such as the decriminalization of sodomy by the U.S. Supreme Court in 2003—gender-nonconforming appearance or behavior continues to be widely stigmatized and remains a frequent target of violence [53]. Research also shows that conformity to gender norms influences social acceptance, as gay men and lesbians are often “less liked” when perceived as unmasculine or unfeminine, respectively [53]. Simultaneously, models such as the cry of pain/defeat and entrapment model of suicidal behavior, suggest that suicidality is caused by feelings of defeat (loss of social status or failure) and entrapment (a sense of being unable to escape a difficult situation) [54]. When youth feel defeated and entrapped, they may activate a"helplessness script"that involves giving-up behaviors, potentially leading to suicidal thoughts and actions [54]. Taken together, these data indicate that it is reasonable to believe that youth who are perceived as being gender nonconforming would experience greater levels of stress due to that perception and may be more vulnerable to suicidality.

Our negative association may be related to the ways in which racial/ethnic identity and sexual identity synergize to influence societal expectations of gender expression. For instance, hypermasculinity is stereotypically associated with the Black Lesbian community by way of the Stud identity group. Studs are Black Lesbians who wear masculine-coded clothing, adopt masculine mannerisms, and perform traditionally male relationship scripts (e.g., being primary financial providers) without desiring to be any other gender. These Studs challenge the hegemonic idea that biological maleness is necessary for masculinity as they are often integrated into social networks saturated by Black Straight Males. Lane-Steele argues that Studs strategically construct and perform their masculinity in ways that shield them from sexism, racism, and homophobia both in and out of their Black community [55]. By adopting a form of masculinity common among their Black Male peers, Studs can gain access to some levels of male privilege and power which, in turn, can act as useful defense mechanisms against multiple types of discrimination and oppression [55]. If this is the case, then the negative association between perceived gender nonconformity and suicidality observed among certain youth groups may be indicative of distinctive sociocultural resilience processes that warrant further study.

Moreover, it is essential to extend this resilience-focused inquiry into other culturally grounded youth communities. For example, among African American adolescents, a strong ethnic–racial identity—encompassing awareness, pride, and centrality—has been repeatedly associated with reduced internalizing symptoms and substance use, while parental ethnic-racial socialization practices (e.g., cultural socialization, preparation for bias, and spirituality) function as buffers against discrimination-related distress [56]. Similarly, among Latinx and Asian American adolescents, commitment to cultural identity has been demonstrated to mitigate the negative psychological effects of racism and fosters self-esteem and academic engagement [57, 58]. Within LGBTQ + youth more broadly, engagement with affirming institutions—such as Gay–Straight Alliances and inclusive school policies—alongside family and peer support, significantly enhance resilience and reduce mental health symptoms [59]. Notably, among Asian American LGBTQ + teens, strong cultural connections halved suicide attempt risk compared to peers less anchored in their heritage [60]. Collectively, these examples illustrate that culturally specific identity labels, communal rituals, family-based teachings, spiritual traditions, and institutional supports may operate as within-group protective factors. Incorporating these frameworks via a socioecological minority-stress lens would deepen insights into how group-specific cultural resources foster adolescent resilience and inform interventions tailored to diverse intersectional identities.

The hypothesized serial mediation model that was demonstrated in the initial model was only replicable among Hispanic Straight Males. The prominence of these indirect associations may be related to traditional forms of machismo, a Hispanic-specific value system centered around the standards of self-reliance, strength, and compulsory leadership [61]. While pressures to conform to societal expectations of gender are not exclusive to Hispanic communities, the sociocultural processes that are associated with being a Hispanic male in the U.S. may exacerbate these pressures as these youth are often expected to adhere to cultural values, beliefs, and practices [62]. As compulsory protectors of their culture, Hispanic Straight Males may be rigidly policed and regulated by their peers to demonstrate prescribed gender expectations [61]. Should they fail to meet these expectations, Hispanic Straight Males may be exposed to greater levels of bullying which may then be associated with increased feelings of depression and higher rates of suicidality.

Implications

Taken together, our findings carry significant implications for the various systems that shape adolescents'daily experiences. First, given that adolescents spend a substantial portion of their time in educational settings, the school environment warrants particular attention. School systems are often organized around cisnormative assumptions, which marginalize students whose gender expression deviates from these normative expectations [63]. As a result, youth who are perceived as being gender nonconforming are frequently subjected to persistent bullying and physical harassment [64]. Considering our findings, we recommend that schools implement explicit policies that protect students who transgress traditional gender norms. Generic antibullying policies that omit references to sexual orientation and gender identity have been shown to offer limited protection for sexual and gender minority youth and fail to adequately address the unique forms of discrimination they face [65]. Moreover, drawing on our intersectional analysis, we emphasize the importance of policies that also address race- and ethnicity-based discrimination. Such policies should confront both structural forms of oppression and individual-level manifestations, including bullying, and must be actively enforced by educators and administrators [66]. Nevertheless, further research is needed to assess the extent to which these policies effectively disrupt the potential mechanisms identified in our serial mediation model.

Second, consistent with prior research, sexual and gender minority youth who have engaged in self-harm, attempted suicide, or experienced identity-based abuse are more likely to seek mental health services [67] It is therefore reasonable to assume that many adolescents in our sample may interact with mental health care systems at some point. We recommend that mental health providers conduct comprehensive risk assessments that explicitly include questions about peer relationships, bullying, and suicidality. Importantly, such assessments should be informed by an intersectional lens, recognizing how multiple dimensions of identity—such as race, gender identity, and sexual orientation—interact to shape adolescents’ risk profiles. Our findings suggest that risk pathways are not uniform across groups; thus, mental health providers should avoid generalizations based on limited demographic markers. For example, it would be inappropriate to assume that Black lesbian females are at heightened risk for suicidality solely based on patterns observed among Black bisexual females or Black gay males. Instead, clinicians should approach each adolescent's identity and lived experience as distinct and contextually nuanced.

Limitations

The results of this study should be interpreted in view of its limitations. First, the use of mediation models with cross-sectional YRBS data carries several limitations. Most notably, the lack of temporal sequencing prevents definitive conclusions about causality or the directionality of effects. Additionally, associations identified may be influenced by unmeasured confounding variables, and the reliance on self-reported data introduces potential bias. As such, findings should be interpreted as exploratory and hypothesis-generating rather than confirmatory. Secondly, there is the potential for recall bias, given the retrospective nature of the YRBS. Third, YRBS does not assess gender identity, thus we were unable to unambiguously ascertain the number of transgender or gender diverse youth who may be present with the current dataset. This limitation prevented us from examining the interaction between actual gender identity and perceived gender identity. Future research should be dedicated to examining this key association. Additionally, the “not sure” category for sexual identity does not distinguish between individuals who are questioning their sexual orientation and individuals who do not know what the question is asking—this presents challenges when interpreting the data obtained from this question. We experience similar challenges when examining the “Not Specified Race/Ethnicity” classification. Due to the robust levels of heterogeneity represented within these groups, we were unable to draw any reasonable or substantive interpretations of findings associated with these groups. More nuanced and granular assessments are warranted in future investigations. Fourth, while the bullying index offers a concise measure of cumulative victimization, several limitations should be noted. First, each indicator is assessed using a single binary item, which may limit the depth and nuance of the data by not capturing frequency, severity, or context of the experiences. Second, summing diverse forms of victimization assumes equal weighting and ignores potential differences in impact across bullying types. Finally, the index does not distinguish between victimization and perpetration in the case of physical fighting, potentially conflating different roles in peer aggression.

Conclusion

Overall, and notwithstanding these limitations, the main strength of our study is the rigorous and intersectional assessment of the four-way association between perceived gender nonconformity, bullying, depression, and suicidality using a diverse sample of adolescents. Results of the present study highlight the need to routinely conduct sociocultural specific assessments of identity related stressors as similar factors may confer additional risk for some while reducing risk among others. Future work may address this by creating mental health screeners that distinguish certain culturally specific stress and resilience factors, such as Stud or machismo. In conclusion, our results stress the need to examine perceived social statuses/identities, particularly related to gender, as a risk factor of suicidality.

Authors’ contributions

MGC and MM conceptualized the study. MGC analyzed and interpreted study results. J.B., Y.B.F., S. D., N.A. J., L.B. & G. P. II provided substantial contributions and supported the analysis, review, and editing of the article.

Funding

This research was supported by the National Institutes of Health and the National Institute on Alcohol Abuse and Alcoholism under Grant Award Numbers R01 AA024409 and R01 AA029044. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The content is solely the authors'responsibility and does not necessarily represent the official views of the National Institutes of Health, or the National Institute on Alcohol Abuse and Alcoholism.

Data availability

The data that support the findings of this study are available from Centers for Disease Control and Prevention, but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of Centers for Disease Control and Prevention.

Declarations

Ethics approval and consent to participate

Youth participation was anonymous and voluntary. Youth actively consented to participate in the YRBS. Parental active written consent or passive consent is obtained based on local procedures prior to youths’ participation by the Centers for Disease Control and Prevention. Study activities that were conducted in conjunction with this study were classified as not human subjects research by the IRB at Northwestern University in accordance with the Declaration of Helsinki. All YRBS data collection activities are completed by the Centers for Disease Control and Prevention outside of our team's scope.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.