Abstract
Purpose:
The overall objective of this study was to examine the hypothesis that victimization exposure among sexual and gender minority (SGM) youth would result in different latent classes and that victimization exposure class membership would relate to demographic, SGM-specific risk and protective factors, and health variables.
Methods:
Between April 2017 and December 2017, SGM youth (N = 17,112) aged 13–17 years completed self-report online surveys as part of the LGBTQ National Teen Survey. Data were analyzed between August 2020 and November 2020.
Results:
Three classes emerged: (1) no victimization exposure, (2) sexual harassment and bullying, and (3) poly-victimization (sexual victimization, sexual harassment, SGM-based bullying, and non-SGM bullying). The results demonstrated that victimization experiences co-occur disproportionately in vulnerable subpopulations of SGM youth, including those who identify as transgender or other gender minority, who are experiencing stigma-related stress and family rejection, and who had disclosed their sexual orientation to family members/parents. SGM youth who reported a diversity of strengths seemed to be protected against victimization. Finally, SGM youth in the sexual harassment and bullying class and the poly-victimization class were more likely to report depressive symptoms, self-perceived stress, and substance use than were SGM youth in the no victimization class, regardless of sex assigned at birth.
Conclusion:
These findings underscore the urgency with which affirmative prevention and intervention initiatives are needed for SGM youth to reduce risk factors for and correlates of victimization experiences. The data also underscore the importance of addressing SGM-specific risk and protective factors as part of comprehensive violence-related initiatives.
Keywords: latent class analysis, mental health, sexual and gender minority, substance use, victimization, youth
Introduction
Sexual and gender minority (SGM) youth experience disproportionately higher rates of victimization (e.g., bullying, dating violence) than heterosexual cisgender adolescents.1–3 However, it remains unknown how victimization types experienced by SGM youth intersect to create typologies. In the few studies that have used latent class analysis (LCA) to understand victimization typologies among adolescents, sexual minority adolescents were more likely than heterosexual adolescents to be in classes characterized by poly-victimization, which were related to more depressive symptoms and alcohol use.4,5 We expand on this application of LCA among a large sample of SGM youth to provide new information about key population subgroups who might experience unique victimization types.
Beyond identifying how victimization experiences intersect among SGM youth, research is needed to identify factors that differentiate victimization class membership. For example, victimized SGM youth report higher levels of alcohol use and more depressive symptoms than nonvictimized SGM youth.3,6–11 Research with SGM youth and adults has found that stigma-related stress increases victimization risk among SGM individuals,12–15 although these studies have primarily used variable-centered approaches versus person-centered approaches (e.g., LCA). This study aims to examine how victimization experiences cluster among a large national sample of SGM youth, assess whether demographic characteristics and SGM-specific risk variables predict class membership in this population, and use class membership to predict mental and behavioral health outcomes. Advancing knowledge of victimization experiences that are more likely to be associated with health risks could inform targeted intervention efforts for SGM youth most at risk and help providers to accurately attribute SGM youth's victimization-related outcomes.16
Protective factors, including mastery and sense of community, reduce the occurrence of victimization among general populations and among SGM youth.17,18 Although protective factors can include a diversity of strengths across various socio-ecological domains (i.e., poly-strengths),19–21 prior studies have primarily examined a single domain and few studies measure protective factors across multiple domains.22 Protective factors relevant to SGM youth, such as self-esteem, access to lesbian, gay, bisexual, transgender, and queer- (LGBTQ-) related support and resources, and presence of a gay–straight alliance (GSA) or similar club,21 may protect against SGM youth's risk of victimization, particularly victimization occurring in dating relationships.17,23 Moreover, there is evidence that violence prevention programs do not work as well for SGM youth compared with heterosexual cisgender youth.24,25 Thus, there is a need to identify modifiable SGM-specific protective factors for victimization experiences to inform SGM-affirming prevention efforts for SGM youth.
The purpose of this study was to (1) identify victimization typologies in a large national sample of SGM youth; (2) examine whether demographic factors and SGM-specific risk and protective factors are associated with victimization class membership; and (3) assess whether various combinations of victimization experiences differentially relate to depressive symptoms, self-perceived stress, and alcohol, marijuana, and cigarette use among SGM youth.
Methods
Participants and procedures
SGM youth aged 13–17 years living in the United States and able to read English (N = 17,112) participated in the cross-sectional LGBTQ Teen Survey,26 a nonprobability national web-based survey. Between April 2017 and December 2017, and in partnership with the Human Rights Campaign (HRC), youth were recruited through social media platforms (e.g., Twitter, Facebook, Instagram) and were offered HRC wristbands and entry into a $50 gift card drawing. Participants provided informed assent; parental permission was waived. Study procedures for the original LGBTQ Teen Survey study were approved by the University of Connecticut's Institutional Review Board (IRB). The University of Connecticut's IRB determined that IRB review and approval was not required for secondary analyses conducted in this article.
Measures
Demographic characteristics
Participants indicated their age, U.S. region of residence, and race/ethnicity (response options included: American Indian or Alaska Native, Asian or Pacific Islander, Biracial or Multiracial, Black or African American, Hispanic/Latinx, Middle Eastern/Arab American, White, and “something else”). Participants were asked, “What sex were you assigned at birth?” Response options were male and female. Participants were asked, “How do you describe your sexual identity?” Response options were gay or lesbian; bisexual; straight, that is, not gay; or “something else,” with a write-in response. Participants were asked, “What is your current gender identity?” Response options were male, female, trans male/boy, trans female/girl, nonbinary, genderqueer/gender nonconforming, and different identity, with a write-in response. From these options we created six mutually exclusive groups of adolescents—cisgender boy, cisgender girl, transgender male/boy, transgender female/girl, gender minority assigned male at birth, and gender minority assigned female at birth.
Youth with discordant sex assigned at birth and gender identity who exclusively chose the “trans male/boy” or “trans female/girl” options were recoded as transgender male/boy and transgender female/girl, respectively. Youth with discordant sex assigned at birth and gender identity who chose only the “male” or “female” gender identity options were recoded as transgender male/boy and transgender female/girl, respectively. Youth who chose only female sex assigned at birth and both “male” and “trans male/boy” were recoded as transgender male/boy. Youth who chose only male sex assigned at birth and both “female” and “trans female/girl” were recoded as transgender female/girl.
Youth who were assigned female sex at birth who checked nonbinary and/or genderqueer/nonconforming (even if they also selected binary identities) were recoded as “gender minority assigned female at birth”; those assigned male sex at birth who checked nonbinary and/or genderqueer/nonconforming were recoded as “gender minority assigned male at birth.” Racial/ethnic groups were collapsed into White and people of color. Sexual orientation groups were collapsed into monosexual (e.g., gay, lesbian), non-monosexual (e.g., bisexual), and something else (e.g., questioning). Gender identity groups were collapsed into cisgender (i.e., cisgender boy and cisgender girl) and transgender and other gender minority (i.e., transgender male/boy, transgender female/girl, gender minority assigned female at birth, and gender minority assigned male at birth).
SGM-specific risk variables
Stigma-related stress was assessed with the mean of the 10-item LGBT Stress scale27 (Cronbach's α = 0.87). Family rejection was assessed by an adapted 4-item family rejection scale, which was originally developed for SGM youth.28,29 Responses from this SGM-specific family rejection scale were mean centered using all available data with higher values indicating more family rejection. The scale demonstrated good internal consistency in the sample (α = 0.89). Sexual orientation disclosure was assessed by a single item; a dichotomous variable was created to indicate individuals who had disclosed their sexual orientation to at least a few family members/parents (46.5%, n = 7958) versus those who had not disclosed their sexual orientation to any family members/parents (53.5%, n = 9154).
SGM-specific protective variables
A poly-strengths composite was created from the following eight scales: self-esteem,30 sense of mastery,31 sense of control,32 positive feelings about being LGBTQ,32 perception of LGBTQ-supportive family members,33 presence of a GSA or similar club, perception of LGBTQ-affirmative teachers and school personnel,2 and access to LGBTQ-related support and resources. Specifically, z-scores were created for each indicator. Each z-score was then dichotomized as “thriving” (≥0.5) versus “not thriving” (<0.5), consistent with prior research.21 A composite variable was created to indicate individuals' total poly-strengths score (Cronbach's α = 0.89).
Mental health variables
Past-week depressive symptoms were measured by the 10-item Kutcher Adolescent Depression Scale.34 We used the highest tertile to calculate depressive symptoms: high (≥1.7) vs. low (<1.7) (Cronbach's α = 0.90). Self-perceived stress was assessed with a single item adapted from the Perceived Stress Scale: “On a scale from 1 to 10, with 1 being not stressed at all and 10 being very stressed, please mark the appropriate number corresponding with your average level of stress”;35 we used the highest tertile to calculate SGM youth most at-risk of stress: high (≥8.0) vs. low (<8) (Cronbach's α = 0.90).
Substance use variables
Participants were asked about their substance use with questions from the 2015 National Youth Risk Behavior Survey (YRBS).36 SGM youth were asked about lifetime use of alcohol, marijuana, and cigarettes. The highest tertile was used to calculate the presence of any alcohol use (0 = ≤2 days, 1 = ≥3 days). We created binary variables indicating any lifetime use (0 = none, 1 = any) of marijuana and cigarettes.
Victimization variables
Participants were asked about their past-year victimization experiences with questions from the 2015 YRBS.36 Participants rated the frequency of experiencing sexual victimization (0 = 0 times, 4 = ≥6 times), sexual victimization and physical victimization in dating relationships (both coded as 0 = 0 times/I did not go out with anyone during the past year, 5 = ≥6 times), and five past-year sexual harassment behaviors.37 Participants also rated the frequency of experiencing six forms of past-year SGM-based bullying (0 = never, 3 = ≥3 times), in addition to non-SGM forms of bullying, including past-year bullying on school property36 (0 = no, 1 = yes), off school property36 (0 = no, 1 = yes), and cyberbullying36 (0 = no, 1 = yes). Across all forms of victimization, the response “0 times,” “never,” or “I did not go out with anyone during the past year” were coded as 0. Any positive endorsement of victimization was coded as 1.
Statistical analysis
We used the three-step latent class analytic approach.38 We fit models with 1–8 classes with our six victimization indicators and specified a priori the following criteria to identify the most optimally fitting LCA model: relative fit, including low log likelihood, Akaike information criterion (AIC), Bayesian information criterion (BIC), and sample size-adjusted BIC (aBIC); entropy above 0.70; class size;39 and interpretability.39 Average posterior probabilities of class membership were used to examine class homogeneity.40
The local independence assumption was assessed by examining bivariate residuals (BVRs) among all pairs of items.41 For LCA models that violated the local independence assumption, direct effects were included to allow for dependent pairs of indicators with the highest residuals. We employed multinomial logistic regressions to model associations between demographic and SGM-specific risk and protective factors with victimization classes identified by the best fitting LCA model, while accounting for classification error in class assignment.38 Then, we used a multinomial logistic regression model regressing health risks onto class membership. For all regression models, we utilized the bias-adjusted maximum likelihood approach.
Little's missing completely at random test was significant (χ2 = 8584.80, df = 7281, p < 0.001), suggesting that the data were not missing completely at random. However, nearly all missingness was attributable to early survey termination rather than the skipping of sensitive items. Furthermore, differences across sexual orientation, gender identity, and race/ethnicity in missingness were relatively minor or negligible. As such, missing data were handled using Latent GOLD's multiple imputation, which provides reliable estimates.41 Descriptive statistics were conducted in SPSS version 24 (IBM Corp., Armonk, NY). LCA and the bias-adjusted three-step latent class analytic approach38 were implemented in Latent GOLD version 5.1.41
Results
Sample characteristics
Descriptive statistics are provided in Table 1.
Table 1.
Frequencies of Study Variables Among Sexual and Gender Minority Youth (N = 17,112)
| Total sample (N = 17,112) | Youth assigned male sex at birth, n = 4739 (27.7%) | Youth assigned female sex at birth, n = 12,373 (72.3%) | |
|---|---|---|---|
| Mean (SD) | |||
| Demographic characteristics | |||
| Age (range: 13–17; median = 15), years | 15.57 (1.27) | 15.84 (1.14) | 15.47 (1.30) |
| n (%) | |||
| Race/ethnicity | |||
| American Indian or Alaska Native | 96 (0.6) | 33 (0.7) | 63 (0.5) |
| Asian or Pacific Islander | 696 (4.1) | 216 (4.6) | 480 (3.9) |
| Biracial or Multiracial | 2508 (14.7) | 654 (13.8) | 1854 (15.0) |
| Black or African American | 959 (5.6) | 310 (6.5) | 649 (5.2) |
| Hispanic/Latinx | 1877 (11.0) | 618 (13.0) | 1259 (10.2) |
| Middle Eastern/Arab American | 53 (0.3) | 16 (0.3) | 37 (0.3) |
| Something else | 87 (0.5) | 27 (0.6) | 60 (0.5) |
| White | 10,836 (63.3) | 2865 (60.5) | 7971 (64.4) |
| Sexual orientation | |||
| Asexual | 725 (4.2) | 52 (1.1) | 673 (5.4) |
| Bisexual | 5970 (34.9) | 1177 (24.8) | 4793 (38.7) |
| Gay or lesbian | 6401 (37.4) | 3127 (66.0) | 3274 (26.5) |
| Straight | 279 (1.6) | 48 (1.0) | 231 (1.9) |
| Something else | 358 (2.1) | 40 (0.8) | 318 (2.6) |
| Pansexual | 2256 (13.2) | 186 (3.9) | 2070 (16.7) |
| Queer | 699 (4.1) | 60 (1.3) | 639 (5.2) |
| Questioning | 424 (2.5) | 49 (1.0) | 375 (3.0) |
| Gender identity | |||
| Cisgender boy | 4079 (23.8) | 4079 (86.1) | 0 (0.0) |
| Cisgender girl | 7396 (43.2) | 0 (0.0) | 7396 (59.8) |
| Transgender male/boy | 1404 (8.2) | 0 (0.0) | 1404 (11.3) |
| Transgender female/girl | 185 (1.1) | 185 (3.9) | 0 (0.0) |
| Gender minority assigned female at birth | 3573 (20.9) | 0 (0.0) | 3573 (28.9) |
| Gender minority assigned male at birth | 475 (2.8) | 475 (10.0) | 0 (0.0) |
| U.S. region of residence | |||
| Northeast | 3081 (18.0) | 878 (18.5) | 2203 (17.8) |
| Midwest | 3889 (22.7) | 1071 (22.6) | 2818 (22.8) |
| South | 6343 (37.1) | 1681 (35.5) | 4662 (37.7) |
| West | 3799 (22.2) | 1109 (23.4) | 2690 (21.7) |
| Presence of past-year sexual or physical victimization | |||
| Sexual victimization | 2321 (13.6) | 384 (8.1) | 1937 (15.7) |
| Sexual victimization in dating relationships | 1472 (8.6) | 227 (4.8) | 1245 (10.1) |
| Physical victimization in dating relationships | 821 (4.8) | 169 (3.6) | 652 (5.3) |
| Sexual harassment | 8565 (50.1) | 1986 (41.9) | 6579 (53.2) |
| Presence of past-year bullying | |||
| Non-SGM-based bullyinga | 6777 (39.6) | 1653 (34.9) | 5124 (41.4) |
| SGM-based bullying | 8269 (48.3) | 2217 (46.8) | 6052 (48.9) |
| SGM-specific risk and protective factors | |||
| Stigma-related stress (range: 0–4; median = 2.30) | 2.27 (1.03) | 2.21 (1.09) | 2.29 (1.04) |
| Family rejection (range: 0–3; median = 1) | 1.11 (0.90) | 0.94 (0.96) | 1.11 (0.96) |
| Sexual orientation disclosure to family members/parents | 7958 (46.5) | 2104 (44.4) | 5854 (47.3) |
| Poly-strengthsb (range: 0–8; median = 4) | 3.86 (1.67) | 4.11 (1.74) | 3.78 (1.64) |
| Presence of mental health problems | |||
| Depressive symptomsc | 3802 (22.2) | 553 (11.7) | 3249 (26.3) |
| Self-perceived stressd | 3355 (19.6) | 652 (13.8) | 2703 (21.8) |
| Presence of lifetime substance use | |||
| Alcohol usee | 4275 (25.0) | 1125 (23.7) | 3150 (25.5) |
| Marijuana use | 3030 (17.7) | 817 (17.2) | 2213 (17.9) |
| Cigarette use | 2474 (14.5) | 648 (13.7) | 1826 (14.8) |
Those who identified as straight also identified as transgender and other gender minority and so were retained in the analyses. Range, mean, median, and SD for age, stigma-related stress, family rejection, and poly-strengths are reported.
Non-SGM bullying (i.e., a composite index of bullying on school property, bullying off school property, and cyberbullying).
Poly-strengths (i.e., a composite index of self-esteem, sense of mastery, sense of control, positive feelings about being LGBTQ, perception of LGBTQ-supportive family members, presence of a gay–straight alliance or similar club, perception of LGBTQ-affirmative teachers and school personnel, and access to LGBT-related support and resources).
Highest tertile was used to calculate the presence of self-reported depressive symptoms.
Highest tertile was used to calculate the presence of self-perceived stress.
Highest tertile was used to calculate the presence of any alcohol use (i.e., 3 days or more).
LGBTQ, lesbian, gay, bisexual, transgender, and queer; SD, standard deviation; SGM, sexual and gender minority.
Model fit assessment and model comparisons of victimization classes
Table 2 indicates model fit indices and LCA model comparison statistics separately for youth assigned male sex at birth and youth assigned female sex at birth. Regardless of SGM youth's sex assigned at birth, the five-class solution had the lowest AIC, the four-class solution had the lowest BIC, the three-class solution had the lowest aBIC, and entropy was higher in the three-class solution, indicating relatively high class separation, suggesting better model fit than the four- through eight-class solutions.38 Furthermore, the AIC, BIC, and aBIC continued to decrease in the one- through three-class solutions, indicating relatively high class separation in the three-class solution. As such, the three-class solution was deemed optimal.40
Table 2.
Model Fit Indices and Model Comparison Statistics for Mixture Modeling of Victimization Experiences
| Number of classes | Log likelihood | Akaike information criterion | Bayesian information criterion | Sample size-adjusted Bayesian information criterion | Entropy | Number of free parameters |
|---|---|---|---|---|---|---|
| Youth assigned male sex at birth (n = 4739) | ||||||
| 1 | −8703.56 | 17,419.12 | 17,457.91 | 17,438.84 | N/A | 6 |
| 2 | −8098.93 | 16,223.87 | 16,307.90 | 16,266.59 | 0.87 | 13 |
| 2DE | −7890.55 | 15,813.11 | 15,916.53 | 15,865.69 | 0.82 | 16 |
| 3a | −7879.36 | 15,798.72 | 15,927.99 | 15,864.44 | 0.79 | 20 |
| 3DE | −7850.29 | 15,746.59 | 15,895.25 | 15,822.16 | 0.64 | 22 |
| 4 | −7855.98 | 15,765.97 | 15,940.49 | 15,854.69 | 0.65 | 27 |
| 5 | −7837.91 | 15,743.82 | 15,963.58 | 15,855.54 | 0.62 | 34 |
| 6 | −7835.18 | 15,752.36 | 16,017.37 | 15,887.08 | 0.67 | 41 |
| 7 | −7834.30 | 15,764.61 | 16,074.86 | 15,922.34 | 0.62 | 48 |
| 8 | −7833.50 | 15,777.01 | 16,132.50 | 15,957.73 | 0.58 | 55 |
| Youth assigned female sex at birth (n = 12,373) | ||||||
| 1 | −27,918.49 | 55,848.97 | 55,893.51 | 55,874.45 | N/A | 6 |
| 2 | −25,480.43 | 50,986.86 | 51,083.36 | 51,042.05 | 0.92 | 13 |
| 3 | −24,796.07 | 49,632.13 | 49,780.59 | 49,717.04 | 0.79 | 20 |
| 3DEa | −24,719.74 | 49,483.47 | 49,646.79 | 49,576.87 | 0.79 | 22 |
| 4 | −24,722.01 | 49,498.02 | 49,698.45 | 49,612.65 | 0.66 | 27 |
| 5 | −24,658.34 | 49,384.69 | 49,637.08 | 49,529.03 | 0.67 | 34 |
| 6 | −24,656.15 | 49,394.31 | 49,698.66 | 49,568.37 | 0.61 | 41 |
| 7 | −24,644.13 | 49,384.26 | 49,740.57 | 49,588.03 | 0.68 | 48 |
| 8 | −24,642.01 | 49,394.02 | 49,802.30 | 49,627.51 | 0.64 | 55 |
Model selected as providing the best fit, as demonstrated by the relatively small Akaike information criterion, Bayesian information criterion, relatively high entropy, and relatively few number of parameters. DE, direct effects (i.e., addition of residual associations [local dependencies]). Each criterion is based on the log likelihood.
Examination of BVRs in the three-class solution revealed a local independence violation. This resulting three-class model allowing for local dependencies showed the best fit for SGM youth assigned female sex at birth. Class 1 was characterized by low probabilities across victimization experiences (“No Victimization Class”; n = 2666, 56.3%; n = 6021, 48.7% for SGM youth assigned male sex at birth and female sex at birth, respectively). Class 2 was characterized by high probabilities of sexual harassment and bullying (“Sexual Harassment, SGM-based Bullying, and non-SGM-based Bullying Class”; n = 1838, 38.8% for SGM youth assigned male sex at birth; “Sexual Harassment and SGM-based Bullying Class”; n = 4232, 34.2% for SGM youth assigned female sex at birth). Finally, Class 3 was characterized by a high probability of sexual victimization, sexual harassment, SGM-based bullying, and non-SGM bullying (“Poly-victimization Class”; n = 235, 5.0%; n = 2120, 17.1% for SGM youth assigned male sex at birth and female sex at birth, respectively).
Correlates of victimization class membership
Regardless of sex assigned at birth, demographic characteristics and SGM-specific risk variables were associated with class membership, suggesting that victimization experiences co-occur disproportionately in vulnerable subpopulations of SGM youth, such as those who were younger, who identify as transgender and other gender minority, who live in the Midwest or South, who experienced stigma-related stress and family rejection, or who had disclosed their sexual orientation to family members/parents (Table 3). In addition, poly-strengths distinguished class membership such that SGM youth who reported a diversity of strengths were more likely to be in the “No Victimization Class” relative to the other classes.
Table 3.
Multinomial Logistic Regression Model of Correlates of Victimization Exposure Latent Classes Among Sexual and Gender Minority Youth
| |
Youth assigned male sex at birth, n = 4739 (27.7%) |
Youth assigned female sex at birth, n = 12,373 (72.3%) |
||||||
|---|---|---|---|---|---|---|---|---|
| Class 2 (“Sexual Harassment, SGM-based Bullying, and non-SGM-based Bullying Class”; n = 1838, 38.8%) |
Class 3 (“Poly-victimization Class”; n = 235, 5.0%) |
Class 2 (“Sexual Harassment and SGM-based Bullying Class”; n = 4232, 34.2%) |
Class 3 (“Poly-victimization Class”; n = 2120, 17.1%) |
|||||
| AOR (95% CI) | p | AOR (95% CI) | p | AOR (95% CI) | p | AOR (95% CI) | p | |
| Demographic characteristics | ||||||||
| Race/ethnicity | ||||||||
| White | ref | ref | ref | ref | ||||
| People of color | 0.80 (0.63–1.03) | 0.040 | 0.73 (0.49–1.08) | 0.060 | 0.70 (0.61–0.82) | <0.001 | 0.76 (0.65–0.88) | <0.001 |
| Sexual orientation | ||||||||
| Monosexuala | ref | ref | ref | ref | ||||
| Non-monosexualb | 0.73 (0.56–0.94) | 0.008 | 1.26 (0.88–1.79) | 0.104 | 0.79 (0.68–0.93) | <0.001 | 1.08 (0.92–1.27) | 0.171 |
| Something elsec | 0.84 (0.37–1.88) | 0.333 | 0.91 (0.29–2.82) | 0.433 | 0.83 (0.63–1.08) | 0.079 | 0.89 (0.67–1.17) | 0.195 |
| Age, years | 0.81 (0.74–0.90) | <0.001 | 0.84 (0.71–0.98) | 0.014 | 0.81 (0.77–0.86) | <0.001 | 0.88 (0.83–0.93) | <0.001 |
| Gender identity | ||||||||
| Cisgender | ref | ref | ref | ref | ||||
| Transgender and other gender minority | 1.40 (1.02–1.92) | 0.019 | 1.97 (1.24–3.12) | 0.002 | 1.78 (1.53–2.08) | <0.001 | 2.02 (1.73–2.35) | <0.001 |
| U.S. region of residence | ||||||||
| Northeast | ref | ref | ref | ref | ||||
| Midwest | 1.64 (1.15–2.35) | 0.003 | 1.59 (0.96–2.62) | 0.034 | 1.57 (1.25–1.98) | <0.001 | 1.62 (1.29–2.04) | <0.001 |
| South | 1.48 (1.05–2.08) | 0.012 | 1.19 (0.73–1.93) | 0.245 | 1.39 (1.12–1.71) | <0.001 | 1.49 (1.21–1.83) | <0.001 |
| West | 1.39 (0.94–2.03) | 0.048 | 1.12 (0.63–1.98) | 0.352 | 1.14 (0.90–1.44) | 0.138 | 1.16 (0.92–1.46) | 0.106 |
| SGM-specific risk variables | ||||||||
| Stigma-related stress | 1.38 (1.66–1.99) | <0.001 | 1.79 (1.42–2.27) | <0.001 | 1.79 (1.59–2.00) | <0.001 | 1.93 (1.72–2.16) | <0.001 |
| Family rejection | 2.22 (1.74–2.84) | <0.001 | 3.07 (2.32–4.07) | <0.001 | 2.26 (1.98–2.58) | <0.001 | 2.81 (2.47–3.20) | <0.001 |
| Sexual orientation disclosure to family members/parents | ||||||||
| Not disclosed sexual orientation to family members/parents | ref | ref | ref | ref | ||||
| Disclosed sexual orientation to at least a few family members/parents | 2.46 (1.77–3.43) | <0.001 | 2.23 (1.36–3.64) | <0.001 | 7.68 (6.33–9.33) | <0.001 | 8.18 (6.74–9.92) | <0.001 |
| SGM-specific protective variables | ||||||||
| Poly-strengthsd | ||||||||
| Thriving | 0.82 (0.73–0.92) | <0.001 | 0.74 (0.88–0.63) | <0.001 | 0.94 (0.87–1.02) | 0.076 | 0.89 (0.82–0.97) | <0.010 |
Significant AOR is indicated in bold. Omitted (ref) category is Class 1 (“No Victimization Class”) for latent classes of victimization exposure (n = 2666, 56.3% for youth assigned male at birth and n = 6021, 48.7% for youth assigned female at birth). All models utilized the bias-adjusted maximum likelihood approach.
Monosexual includes sexual and gender minority youth who identified as gay, lesbian, or straight.
Non-monosexual includes sexual and gender minority youth who identified as bisexual, queer, or pansexual.
Something else includes sexual and gender minority youth who identified as asexual, questioning, or something else.
Poly-strengths (i.e., a composite index of self-esteem, sense of mastery, sense of control, positive feelings about being LGBTQ, perception of LGBTQ-supportive family members, presence of a gay–straight alliance or similar club, perception of LGBTQ-affirmative teachers and school personnel, and access to LGBTQ-related support and resources).
AOR, adjusted odds ratio; CI, confidence interval; ref, reference group.
Victimization class membership as a health correlate
After adjusting for demographic characteristics and SGM-specific risk and protective factors, SGM youth assigned male sex at birth and in the “Sexual Harassment, SGM-based Bullying, and non-SGM-based Bullying Class” and the “Poly-victimization Class” were more likely to report health risks compared with SGM youth assigned male sex at birth in the “No Victimization Class” (Table 4). Similarly, SGM youth assigned female sex at birth and in the “Sexual Harassment and SGM-based Bullying Class” and the “Poly-victimization Class” were more likely to report health risks compared with SGM youth assigned female sex at birth in the “No Victimization Class” (Table 5).
Table 4.
Multinomial Logistic Regression Models Demonstrating Latent Class Correlates of Mental and Behavioral Health Risks Among Youth Assigned Male Sex at Birth (n = 4739)
| Depressive symptomsa |
Self-perceived stressb |
Lifetime alcohol usec |
Lifetime marijuana use |
Lifetime cigarette use |
||||||
|---|---|---|---|---|---|---|---|---|---|---|
| AOR (95% CI) | p | AOR (95% CI) | p | AOR (95% CI) | p | AOR (95% CI) | p | AOR (95% CI) | p | |
| Victimization exposure class | ||||||||||
| Class 1 (“No Victimization Class”; n = 2666, 56.3%) | ref | ref | ref | ref | ref | |||||
| Class 2 (“Sexual Harassment, SGM-based Bullying, and non-SGM-based Bullying Class”; n = 1838, 38.8%) | 10.84 (4.18–28.11) | <0.001 | 2.60 (1.60–4.25) | <0.001 | 3.72 (2.53–5.48) | <0.001 | 4.00 (2.50–6.39) | <0.001 | 5.26 (2.89–9.56) | <0.001 |
| Class 3 (“Poly-victimization Class”; n = 235, 5.0%) | 13.22 (5.11–34.27) | <0.001 | 2.84 (1.56–5.14) | <0.001 | 6.36 (3.89–10.40) | <0.001 | 6.85 (4.04–11.62) | <0.001 | 8.23 (4.37–15.49) | <0.001 |
| Demographic characteristics | ||||||||||
| Race/ethnicity | ||||||||||
| White | ref | ref | ref | ref | ref | |||||
| People of color | 1.22 (0.98–1.53) | 0.039 | 1.05 (0.86–1.28) | 0.322 | 0.79 (0.67–0.93) | 0.028 | 0.98 (0.82–1.18) | 0.428 | 0.88 (0.72–1.06) | 0.093 |
| Sexual orientation | ||||||||||
| Monosexual | ref | ref | ref | ref | ref | |||||
| Non-monosexual | 1.10 (0.87–1.40) | 0.206 | 0.79 (0.64–0.98) | 0.016 | 1.28 (1.08–1.52) | 0.002 | 1.36 (1.13–1.64) | <0.001 | 1.27 (1.03–1.55) | 0.010 |
| Something else | 1.45 (0.81–2.59) | 0.105 | 0.92 (0.55–1.55) | 0.374 | 0.76 (0.44–1.28) | 0.152 | 1.09 (0.65–1.84) | 0.374 | 1.20 (0.69–2.11) | 0.261 |
| Age, years | 1.02 (0.93–1.12) | 0.301 | 1.02 (0.94–1.11) | 0.281 | 1.40 (1.30–1.51) | <0.001 | 1.35 (1.25–1.46) | <0.001 | 1.18 (1.08–1.27) | <0.001 |
| Gender identity | ||||||||||
| Cisgender | ref | ref | ref | ref | ref | |||||
| Transgender and other gender minority | 2.10 (1.60–2.76) | <0.001 | 1.89 (1.47–2.43) | <0.001 | 0.74 (0.58–0.93) | 0.005 | 1.06 (0.84–1.35) | 0.312 | 0.92 (0.71–1.20) | 0.274 |
| U.S. region of residence | ||||||||||
| Northeast | ref | ref | ref | ref | ref | |||||
| Midwest | 0.87 (0.62–1.21) | 0.203 | 0.85 (0.64–1.15) | 0.147 | 0.85 (0.67–1.08) | 0.086 | 1.00 (0.77–1.30) | 0.496 | 1.54 (1.13–2.08) | 0.002 |
| South | 0.89 (0.66–1.21) | 0.229 | 0.94 (0.72–1.23) | 0.322 | 1.01 (0.82–1.26) | 0.456 | 1.03 (0.81–1.31) | 0.401 | 1.74 (1.31–2.30) | <0.001 |
| West | 1.01 (0.72–1.40) | 0.488 | 1.13 (0.84–1.51) | 0.211 | 0.87 (0.68–1.10) | 0.119 | 1.04 (0.81–1.36) | 0.359 | 1.40 (1.03–1.91) | 0.017 |
| SGM-specific risk variables | ||||||||||
| Stigma-related stress | 1.23 (1.08–1.39) | <0.001 | 1.24 (1.12–1.37) | <0.001 | 0.99 (0.91–1.08) | 0.397 | 0.91 (0.83–1.00) | 0.028 | 0.87 (0.79–0.97) | 0.007 |
| Family rejection | 1.50 (1.32–1.71) | <0.001 | 1.20 (1.08–1.35) | <0.001 | 1.09 (0.98–1.21) | 0.049 | 1.18 (1.06–1.32) | <0.001 | 1.20 (1.07–1.36) | <0.001 |
| Sexual orientation disclosure to family members/parents | ||||||||||
| Not disclosed sexual orientation to family members/parents | ref | ref | ref | ref | ref | |||||
| Disclosed sexual orientation to at least a few family members/parents | 2.32 (1.86–2.88) | <0.001 | 1.04 (0.86–1.27) | 0.334 | 3.37 (2.88–3.95) | <0.001 | 2.86 (2.41–3.40) | <0.001 | 2.68 (2.21–3.23) | <0.001 |
| SGM-specific protective variables | ||||||||||
| Poly-strengthsd | ||||||||||
| Thriving | 0.58 (0.53–0.64) | <0.001 | 0.83 (0.78–0.89) | <0.001 | 1.03 (0.97–1.10) | 0.166 | 1.00 (0.94–1.07) | 0.492 | 0.89 (0.82–0.95) | <0.001 |
All models adjusted for demographic characteristics and SGM-specific risk and protective variables. Significant AOR is indicated in bold. All models utilized the bias-adjusted maximum likelihood approach.
Highest tertile was used to calculate the presence of self-reported depressive symptoms.
Highest tertile was used to calculate the presence of self-perceived stress.
Highest tertile was used to calculate the presence of any alcohol use (i.e., 3 days or more).
Poly-strengths (i.e., a composite index of self-esteem, sense of mastery, sense of control, positive feelings about being LGBTQ, perception of LGBTQ-supportive family members, presence of a gay–straight alliance or similar club, perception of LGBTQ-affirmative teachers and school personnel, and access to LGBTQ-related support and resources).
Table 5.
Multinomial Logistic Regression Models Demonstrating Latent Class Correlates of Mental and Behavioral Health Risks Among Youth Assigned Female Sex at Birth (n = 12,373)
| Depressive symptomsa |
Self-perceived stressb |
Lifetime alcohol usec |
Lifetime marijuana use |
Lifetime cigarette use |
||||||
|---|---|---|---|---|---|---|---|---|---|---|
| AOR (95% CI) | p | AOR (95% CI) | p | AOR (95% CI) | p | AOR (95% CI) | p | AOR (95% CI) | p | |
| Victimization exposure class | ||||||||||
| Class 1 (“No Victimization Class”; n = 6021; 48.7%) | ref | ref | ref | ref | ref | |||||
| Class 2 (“Sexual Harassment and SGM-based Bullying Class”; n = 4232; 34.2%) | 6.06 (4.13–8.89) | <0.001 | 1.99 (1.49–2.64) | <0.001 | 3.71 (2.74–5.01) | <0.001 | 4.50 (3.00–6.76) | <0.001 | 5.14 (3.06–8.62) | <0.001 |
| Class 3 (“Poly-victimization Class”; n = 2120; 17.1%) | 8.19 (5.85–11.48) | <0.001 | 2.41 (1.89–3.07) | <0.001 | 6.70 (5.15–8.70) | <0.001 | 8.86 (6.19–12.66) | <0.001 | 9.46 (5.97–14.99) | <0.001 |
| Demographic characteristics | ||||||||||
| Race/ethnicity | ||||||||||
| White | ref | ref | ref | ref | ref | |||||
| People of color | 1.08 (0.98–1.20) | 0.059 | 1.07 (0.97–1.18) | 0.102 | 0.90 (0.82–0.99) | 0.018 | 1.21 (1.08–1.35) | 0.007 | 1.02 (0.91–1.15) | 0.356 |
| Sexual orientation | ||||||||||
| Monosexual | ref | ref | ref | ref | ref | |||||
| Non-monosexual | 1.10 (1.00–1.23) | 0.046 | 1.22 (0.98–1.22) | 0.051 | 0.99 (0.89–1.09) | 0.389 | 0.82 (0.73–0.92) | <0.001 | 0.99 (0.87–1.12) | 0.425 |
| Something else | 1.36 (1.15–1.62) | <0.001 | 1.21 (1.03–1.43) | 0.011 | 0.61 (0.51–0.73) | <0.001 | 0.54 (0.44–0.66) | <0.001 | 0.59 (0.48–0.74) | <0.001 |
| Age, years | 0.99 (0.96–1.03) | 0.484 | 1.02 (0.98–1.05) | 0.181 | 1.46 (1.40–1.52) | <0.001 | 1.63 (1.55–1.70) | <0.001 | 1.34 (1.28–1.41) | <0.001 |
| Gender identity | ||||||||||
| Cisgender | ref | ref | ref | ref | ref | |||||
| Transgender and other gender minority | 1.57 (1.42–1.73) | <0.001 | 1.23 (1.11–1.36) | <0.001 | 0.97 (0.88–1.06) | 0.245 | 1.09 (0.98–1.21) | 0.063 | 1.27 (1.14–1.43) | <0.001 |
| U.S. region of residence | ||||||||||
| Northeast | ref | ref | ref | ref | ref | |||||
| Midwest | 1.05 (0.89–1.22) | 0.277 | 0.81 (0.70–0.94) | 0.002 | 0.89 (0.77–1.03) | 0.655 | 1.12 (0.95–1.32) | 0.226 | 1.32 (1.11–1.58) | <0.001 |
| South | 0.89 (0.77–1.03) | 0.058 | 0.90 (0.78–1.03) | 0.063 | 0.96 (0.84–1.10) | 0.352 | 1.06 (0.91–1.24) | 0.227 | 1.30 (1.10–1.53) | <0.001 |
| West | 0.99 (0.85–1.16) | 0.456 | 0.89 (0.77–1.04) | 0.074 | 1.10 (0.95–1.27) | 0.254 | 1.27 (1.07–1.50) | 0.003 | 1.22 (1.11–1.47) | 0.015 |
| SGM-specific risk variables | ||||||||||
| Stigma-related stress | 1.26 (1.19–1.33) | <0.001 | 1.21 (1.15–1.27) | <0.001 | 1.02 (0.96–1.08) | 0.254 | 0.90 (0.85–0.96) | <0.001 | 0.93 (0.88–0.99) | 0.019 |
| Family rejection | 1.55 (1.45–1.65) | <0.001 | 1.23 (1.16–1.30) | <0.001 | 1.25 (1.18–1.33) | <0.001 | 1.28 (1.20–1.37) | <0.001 | 1.36 (1.27–1.46) | <0.001 |
| Sexual orientation disclosure to family members/parents | ||||||||||
| Not disclosed sexual orientation to family members/parents | ref | ref | ref | ref | ref | |||||
| Disclosed sexual orientation to at least a few family members/parents | 2.84 (2.57–3.15) | <0.001 | 1.055 (0.95–1.16) | 0.192 | 2.53 (2.30–2.78) | <0.001 | 2.91 (2.61–3.25) | <0.001 | 2.55 (2.27–2.87) | <0.001 |
| SGM-specific protective variables | ||||||||||
| Poly-strengthsd | ||||||||||
| Thriving | 0.65 (0.62–0.68) | <0.001 | 1.23 (1.16–1.30) | <0.001 | 0.97 (0.93–1.01) | 0.052 | 0.99 (0.95–1.03) | 0.274 | 0.91 (0.87–0.95) | <0.001 |
All models adjusted for demographic characteristics and SGM-specific risk and protective variables. Significant AOR is indicated in bold. All models utilized the bias-adjusted maximum likelihood approach.
Highest tertile was used to calculate the presence of self-reported depressive symptoms.
Highest tertile was used to calculate the presence of self-perceived stress.
Highest tertile was used to calculate the presence of any alcohol use (i.e., 3 days or more).
Poly-strengths (i.e., a composite index of self-esteem, sense of mastery, sense of control, positive feelings about being LGBTQ, perception of LGBTQ-supportive family members, presence of a gay–straight alliance or similar club, perception of LGBTQ-affirmative teachers and school personnel, and access to LGBTQ-related support and resources).
Discussion
Extending previous research, the current study highlights heterogeneity in victimization classes among a large national sample of SGM youth. Although a no victimization class emerged, more than 50% of SGM youth in our sample reported sexual harassment, and more than 40% reported SGM-based bullying, among other victimization experiences. These findings are consistent with extant research on SGM victimization42 and underscore the pervasiveness of victimization among SGM youth across the United States and the urgency with which prevention and intervention efforts are needed for this population.
The finding that transgender and other gender minority youth, regardless of sex assigned at birth, were more likely to be in classes characterized by poly-victimization adds to a growing body of literature highlighting the social determinants of poor health facing this vulnerable population.11,14,43 Indeed, transgender and other gender minority youth might be more vulnerable to victimization rooted in both heterosexism and cissexism.44 These findings also indicate that SGM youth in the Midwestern and Southern regions of the United States greatly need affirming prevention initiatives. Notably, younger SGM youth were more likely to report sexual harassment, SGM-based bullying, and poly-victimization compared with older SGM youth, consistent with prior research among sexual minority youth in general.45 Documenting similarities and differences in the manifestation and impact of victimization across SGM youth subgroups is critical to advancing intervention and prevention efforts for SGM youth.
Although previous research has documented that family rejection predicts psychosocial difficulties, including suicide, depression, and substance use among SGM youth,29 this is the first study to demonstrate associations between family rejection and victimization experiences in this population. Moreover, family rejection among SGM youth is also associated with adverse experiences, including homelessness,46–48 and homelessness is predictive of subsequent victimization via risk-taking behaviors.49 This is also the first study to document that stigma-related stress is related to victimization class membership among SGM youth, suggesting that SGM youth presenting with stigma-related stress may benefit from connection to affirming, trauma-related programming. Furthermore, these findings underscore the importance of prevention efforts for SGM youth to reduce stigma-related stress and connect SGM youth with positive adult role models to help buffer against the deleterious impacts of family rejection.
In addition, the findings highlight SGM youth's resilience and indicate that programs seeking to enhance the accumulation of strengths, rather than focusing on specific types, may help to reduce the likelihood of victimization. Future research should extend these findings by investigating potential moderators (e.g., emotion-oriented coping) of the association between poly-strengths and class membership. Future research should also consider utilizing a heterosexual cisgender comparison group to examine whether there are general and SGM-specific differences in victimization class membership and class correlates among youth. Finally, youth in classes characterized by sexual harassment, SGM-based bullying, and poly-victimization were more likely to report depressive symptoms, stress, and substance use, which coincides with the literature from largely heterosexual cisgender youth demonstrating that exposure to victimization is associated with more health problems2 and highlights the need for affirming interventions for SGM youth.
Limitations
First, the data were cross-sectional and utilized inconsistent time frames (e.g., past-year victimization, past-week depressive symptoms), which limit our ability to address temporal sequencing among variables. Nevertheless, this study provides critical foundational knowledge that can be replicated by future longitudinal research. Several measures were single-item indicators, which could have limited variability in these predictors of class membership and health risks or caused misclassification in our latent classes. In addition, there was no measure of perpetration, and details (e.g., SGM status) about perpetrators were not assessed.
This study also utilized retrospective self-report measures, thereby introducing the possibility of biased reporting. However, recent evidence suggests concordant results across self-report and interviewer-rated retrospective reports of early victimization exposure.50 As noted previously,51 data are not representative of all SGM youth in the United States given that only SGM youth who had access to the internet participated in this study. As such, our results cannot be generalized to youth who do not have access to online networks where HRC advertised the study.26 Nevertheless, research suggests that 95% of youth have access to smartphones.52
This study recruited from social media platforms, including Twitter, Facebook, and Instagram, consistent with prior research with similar populations.53,54 Given that social media platforms have different purposes, audiences, and norms (e.g., sharing life events vs. social blogging), whether our study's findings generalize to SGM youth who use other social media platforms remains unknown.55 Average time spent on social media and connection speed also likely impacted potentially eligible participants' likelihood of inclusion in this study.
Due to sample size limitations, we were not able to fully disaggregate our models across all sexual orientation and gender identity subgroups. Consistent with prior research that has utilized LCA to uncover clusters of SGM youth reporting victimization experiences and associated risk and protective factors,56 and to provide insight into potential variation (or lack thereof) in patterns of victimization within SGM youth populations, this study's analyses were stratified by participants' sex assigned at birth. Disaggregating by sex assigned at birth may inadvertently reinforce essentialist notions of sex and gender identity. As such, future research should consider sampling this population in a more targeted manner to assess potential differences across sex assigned at birth, gender identity, and sexual orientation.
Prevention and intervention implications
These data highlight the need for providers to assess for SGM-specific risk factors as part of comprehensive violence-related initiatives among SGM youth regardless of sex assigned at birth. Given that previous research suggests that violence prevention programs do not work as well for SGM youth compared with heterosexual cisgender youth,24,25 this further demonstrates the need for programs to address SGM-specific risk factors. Furthermore, because of the critical role of important adults in promoting resilience among youth, and our findings that stigma-related stress, family rejection, and identity disclosure to family members were associated with membership in classes characterized by sexual harassment, SGM-based bullying, and non-SGM-based bullying as well as poly-victimization, efforts are needed to connect SGM youth with little familial support to other SGM youth and adults. Also, evidence-based cyberbullying prevention and intervention efforts are needed to increase awareness of behaviors that constitute SGM-based cyberbullying and to increase empathy for SGM youth who experience cyberbullying.57
Programs aimed to help foster SGM youth's strengths (e.g., GSAs) may help to reduce victimization risk in this vulnerable population by increasing the presence of SGM youth in schools and encouraging SGM youth and their cisgender heterosexual allies to engage in activist roles to promote a supportive school climate.18,58 These programs would also likely lead to other health benefits (e.g., reduced alcohol use). Finally, person-centered analytic approaches, such as LCA,6,56 could enhance (1) screening of SGM youth, (2) connecting SGM youth to support services, and (3) developing affirming prevention and response efforts for this population.
Conclusions
Our findings underscore the need for comprehensive assessment of victimization experiences among SGM youth in practice-based settings as well as the importance of affirmative prevention and intervention initiatives for SGM youth to reduce risk factors and enhance protective factors for victimization. These initiatives may be important for younger SGM youth, transgender and other gender minority youth, those who have disclosed their sexual orientation to family members/parents, and those experiencing stigma-related stress and family rejection, as these youth were more likely to be in classes characterized by multiple forms of victimization, which were associated with greater health risks.
Acknowledgments
We acknowledge the efforts of Ellen Kahn, Gabe Murchison, Liam Miranda, and the Harvard T. H. Chan Population Health Sciences Social Epidemiology Working Group for their work in supporting, conceptualizing, and managing the LGBTQ Teen Study.
Authors' Contributions
J.R.S. provided input on the study design, conducted data analysis, drafted the initial article, and provided revisions to the article. K.M.E. provided input on the study design and analyses, drafted the initial article, and provided revisions to the article. E.C.H. provided input on the study design and analyses and reviewed the initial and revised article. R.J.W. conceptualized and designed the study, coordinated and supervised data collection, and reviewed the initial and revised article. All authors approved the final article as submitted and agree to be accountable for all aspects of the work.
Disclaimer
This research uses secondary data from the LGBTQ Teen Study, designed by Ryan J. Watson and Rebecca M. Puhl in collaboration with the Human Rights Campaign Foundation, and supported by the University of Connecticut's Office for Vice President of Research. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Author Disclosure Statement
No competing financial interests exist.
Funding Information
J.R.S. acknowledges support from the National Institute on Alcohol Abuse and Alcoholism grant under K01AA028239. R.J.W. acknowledges support from the National Institute on Drug Abuse grant under K01DA047918.
References
- 1. Earnshaw VA, Bogart LM, Poteat VP, et al. : Bullying among lesbian, gay, bisexual, and transgender youth. Pediatr Clin North Am 2016;63:999–1010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Basile KC, Clayton HB, DeGue S, et al. : Interpersonal violence victimization among high school students—Youth Risk Behavior Survey, United States, 2019. MMWR Suppl 2020;69:28–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Edwards KM: Incidence and outcomes of dating violence victimization among high school youth: The role of gender and sexual orientation. J Interpers Violence 2018;33:1472–1490. [DOI] [PubMed] [Google Scholar]
- 4. Siller L, Edwards KM, Banyard VL: Violence typologies among youth: A latent class analysis of middle and high school youth. J Interpers Violence 2020 [Epub ahead of print]; DOI: 10.1177/0886260520922362. [DOI] [PubMed] [Google Scholar]
- 5. Sessarego SN, Siller L, Edwards KM: Patterns of violence victimization and perpetration among adolescents using latent class analysis. J Interpers Violence 2019 [Epub ahead of print]; DOI: 10.1177/0886260519862272. [DOI] [PubMed] [Google Scholar]
- 6. Scheer JR, Antebi-Gruszka N, Sullivan TP: Physical and sexual victimization class membership and alcohol use correlates among sexual minority and heterosexual female youth. Psychol Violence 2021;11:434–444. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Edwards KM, Waterman EA, Banyard VL: Do depression and binge drinking explain the increased risk of sexual violence among sexual minority middle and high school girls? J Interpers Violence 2020 [Epub ahead of print]; DOI: 10.1177/0886260520938506. [DOI] [PubMed] [Google Scholar]
- 8. Langenderfer-Magruder L, Walls N, Whitfield DL, et al. : Partner violence victimization among lesbian, gay, bisexual, transgender, and queer youth: Associations among risk factors. Child Adolesc Soc Work J 2016;33:55–68. [Google Scholar]
- 9. Paceley MS, Fish JN, Thomas MMC, Goffnett J: The impact of community size, community climate, and victimization on the physical and mental health of SGM youth. Youth Soc 2020;52:427–448. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Scheer JR, McConocha E, Behari K, Pachankis JE: Sexual violence as a mediator of sexual orientation disparities in alcohol use, suicidality, and sexual-risk behaviour among female youth. Psychol Sex 2021;12:37–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Scheer JR, Mereish EH: Intimate partner violence and illicit substance use among sexual and gender minority youth: The protective role of cognitive reappraisal. J Interpers Violence 2019 [Epub ahead of print]; DOI: 10.1177/0886260519881001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Edwards KM, Sylaska KM: The perpetration of intimate partner violence among LGBTQ college youth: The role of minority stress. J Youth Adolesc 2013;42:1721–1731. [DOI] [PubMed] [Google Scholar]
- 13. Edwards KM, Sylaska KM, Neal AM: Intimate partner violence among sexual minority populations: A critical review of the literature and agenda for future research. Psychol Violence 2015;5:112–121. [Google Scholar]
- 14. Martin-Storey A, Pollitt AM, Baams L: Profiles and predictors of dating violence among sexual and gender minority adolescents. J Adolesc Health 2021;68:1155–1161. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Watson RJ, Fish JN, Poteat VP, Rathus T: Sexual and gender minority youth alcohol use: Within-group differences in associations with internalized stigma and victimization. J Youth Adolesc 2019;48:2403–2417. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Lowe SR, James P, Arcaya MC, et al. : Do levels of posttraumatic growth vary by type of traumatic event experienced? An analysis of the Nurses' Health Study II. Psychol Trauma 2020 [Epub ahead of print]; DOI: 10.1037/tra0000554. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Sabina C, Banyard VL: Moving toward well-being: The role of protective factors in violence research. Psychol Violence 2015;5:337–342. [Google Scholar]
- 18. Marx RA, Kettrey HH: Gay-straight alliances are associated with lower levels of school-based victimization of LGBTQ+ youth: A systematic review and meta-analysis. J Youth Adolesc 2016;45:1269–1282. [DOI] [PubMed] [Google Scholar]
- 19. Hatchel T, Merrin GJ, Espelage D: Peer victimization and suicidality among LGBTQ youth: The roles of school belonging, self-compassion, and parental support. J LGBT Youth 2019;16:134–156. [Google Scholar]
- 20. Hamby S, Taylor E, Mitchell K, et al. : Poly-victimization, trauma, and resilience: Exploring strengths that promote thriving after adversity. J Trauma Dissociation 2020;21:376–395. [DOI] [PubMed] [Google Scholar]
- 21. Banyard V, Edwards KM, Jones L, Mitchell K: Poly-strengths and peer violence perpetration: What strengths can add to risk factor analyses. J Youth Adolesc 2020;49:735–746. [DOI] [PubMed] [Google Scholar]
- 22. Hamby S, Taylor E, Mitchell K, et al. : Health-related quality of life among adolescents as a function of victimization, other adversities, and strengths. J Pediatr Nurs 2020;50:46–53. [DOI] [PubMed] [Google Scholar]
- 23. Lösel F, Farrington DP: Direct protective and buffering protective factors in the development of youth violence. Am J Prev Med 2012;43(2 Suppl 1):S8–S23. [DOI] [PubMed] [Google Scholar]
- 24. Coker AL, Bush HM, Clear ER, et al. : Bystander program effectiveness to reduce violence and violence acceptance within sexual minority male and female high school students using a cluster RCT. Prev Sci 2020;21:434–444. [DOI] [PubMed] [Google Scholar]
- 25. Waterman EA, Edwards KM, Beaulieu AE, Banyard VL: Who sees opportunity to help? A prospective study on adolescents' detection of intervention opportunities in situations of sexual and dating violence. J Soc Pers Relatsh 2020;37:2843–2862. [Google Scholar]
- 26. Watson RJ, Wheldon CW, Puhl RM: Evidence of diverse identities in a large national sample of sexual and gender minority youth. J Res Adolesc 2020;30 (Suppl 2):431–442. [DOI] [PubMed] [Google Scholar]
- 27. Rosario M, Rotheram-Borus MJ, Reid H: Gay-related stress and its correlates among gay and bisexual male adolescents of predominantly Black and Hispanic background. J Community Psychol 1996;24:136–159. [Google Scholar]
- 28. Ryan C, Huebner D, Diaz RM, Sanchez J: Family rejection as a predictor of negative health outcomes in white and Latino lesbian, gay, and bisexual young adults. Pediatrics 2009;123:346–352. [DOI] [PubMed] [Google Scholar]
- 29. Ryan C, Russell ST, Huebner D, et al. : Family acceptance in adolescence and the health of LGBT young adults. J Child Adolesc Psychiatr Nurs 2010;23:205–213. [DOI] [PubMed] [Google Scholar]
- 30. Rosenberg M: Society and the Adolescent Self-Image. Revised Ed. Middletown, CT: Wesleyan University Press, 1989. [Google Scholar]
- 31. Folkman S, Lazarus RS, Gruen RJ, DeLongis A: Appraisal, coping, health status, and psychological symptoms. J Pers Soc Psychol 1986;50:571–579. [DOI] [PubMed] [Google Scholar]
- 32. Shidlo A: Internalized homophobia: Conceptual and empirical issues in measurement. In: Psychological Perspectives on Lesbian and Gay Issues, Vol. 1. Lesbian and Gay Psychology: Theory, Research, and Clinical Applications. Edited by Greene B, Herek GM. Thousand Oaks, CA: Sage Publications, Inc., 1994, pp 176–205. [Google Scholar]
- 33. Procidano ME, Heller K: Measures of perceived social support from friends and from family: Three validation studies. Am J Community Psychol 1983;11:1–24. [DOI] [PubMed] [Google Scholar]
- 34. LeBlanc JC, Almudevar A, Brooks SJ, Kutcher S: Screening for adolescent depression: Comparison of the Kutcher adolescent depression scale with the Beck depression inventory. J Child Adolesc Psychopharmacol 2002;12:113–126. [DOI] [PubMed] [Google Scholar]
- 35. Cohen S, Kamarck T, Mermelstein R: A global measure of perceived stress. J Health Soc Behav 1983;24:385–396. [PubMed] [Google Scholar]
- 36. Kann L, McManus T, Harris W, et al. : Youth Risk Behavior Surveillance - United States, 2015. MMWR Surveill Summ 2016;10:1–174. [DOI] [PubMed] [Google Scholar]
- 37. Hill C, Kearl H: Crossing the Line: Sexual Harassment at School. Washington, DC: American Association of University Women, 2011. [Google Scholar]
- 38. Bakk Z, Tekle FB, Vermunt JK: Estimating the association between latent class membership and external variables using bias-adjusted three-step approaches. Sociol Methodol 2013;43:272–311. [Google Scholar]
- 39. Lanza ST, Rhoades BL: Latent class analysis: An alternative perspective on subgroup analysis in prevention and treatment. Prev Sci 2013;14:157–168. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Nylund KL, Asparouhov T, Muthén BO: Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Struct Equ Modeling 2007;14:535–569. [Google Scholar]
- 41. Vermunt JK, Magidson J: Technical Guide for Latent GOLD 5.1: Basic, Advanced, and Syntax. Belmont, MA: Statistical Innovations, Inc., 2016. [Google Scholar]
- 42. Walls NE, Atteberry-Ash B, Kattari SK, et al. : Gender identity, sexual orientation, mental health, and bullying as predictors of partner violence in a representative sample of youth. J Adolesc Health 2019;64:86–92. [DOI] [PubMed] [Google Scholar]
- 43. Johns MM, Lowry R, Andrzejewski J, et al. : Transgender identity and experiences of violence victimization, substance use, suicide risk, and sexual risk behaviors among high school students—19 states and large urban school districts, 2017. Morb Mortal Wkly Rep 2019;68:67–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Norris AL, Orchowski LM: Peer victimization of sexual minority and transgender youth: A cross-sectional study of high school students. Psychol Violence 2020;10:201–211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Russell ST, Fish JN: Sexual minority youth, social change, and health: A developmental collision. Res Hum Dev 2019;16:5–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Shelton J, Bond L: “It just never worked out”: How transgender and gender expansive youth understand their pathways into homelessness. Fam Soc 2017;98:284–291. [Google Scholar]
- 47. Robinson BA: Conditional families and lesbian, gay, bisexual, transgender, and queer youth homelessness: Gender, sexuality, family instability, and rejection. J Marriage Fam 2018;80:383–396. [Google Scholar]
- 48. McCann E, Brown M: Homelessness among youth who identify as LGBTQ+: A systematic review. J Clin Nurs 2019;28:2061–2072. [DOI] [PubMed] [Google Scholar]
- 49. Tyler KA, Schmitz RM: A comparison of risk factors for various forms of trauma in the lives of lesbian, gay, bisexual and heterosexual homeless youth. J Trauma Dissociation 2018;19:431–443. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50. Gayer-Anderson C, Reininghaus U, Paetzold I, et al. : A comparison between self-report and interviewer-rated retrospective reports of childhood abuse among individuals with first-episode psychosis and population-based controls. J Psychiatr Res 2020;123:145–150. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51. Watson RJ, Fish JN, Poteat VP, et al. : Teacher support, victimization, and alcohol use among sexual and gender minority youth: Considering ethnoracial identity. Prev Sci 2021;22:590–601. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52. Anderson M, Jiang J: Teens, social media & technology 2018. Washington, DC: Pew Research Center, 2018. Available at www.pewresearch.org/internet/2018/05/31/teens-social-media-technology-2018 Accessed July 1, 2021.
- 53. Prescott TL, Phillips II G, DuBois LZ, et al. : Reaching adolescent gay, bisexual, and queer men online: Development and refinement of a national recruitment strategy. J Med Internet Res 2016;18:e200. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54. Guillory J, Wiant KF, Farrelly M, et al. : Recruiting hard-to-reach populations for survey research: Using Facebook and Instagram advertisements and in-person intercept in LGBT bars and nightclubs to recruit LGBT young adults. J Med Internet Res 2018;20:e197. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55. Ilakkuvan V, Johnson A, Villanti AC, et al. : Patterns of social media use and their relationship to health risks among young adults. J Adolesc Health 2019;64:158–164. [DOI] [PubMed] [Google Scholar]
- 56. Baams L: Disparities for LGBTQ and gender nonconforming adolescents. Pediatrics 2018;141:e20173004. [DOI] [PubMed] [Google Scholar]
- 57. Abreu RL, Kenny MC: Cyberbullying and LGBTQ youth: A systematic literature review and recommendations for prevention and intervention. J Child Adolesc Trauma 2018;11:81–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 58. Poteat VP, Scheer JR, Marx RA, et al. : Gay-straight alliances vary on dimensions of youth socializing and advocacy: Factors accounting for individual and setting-level differences. Am J Community Psychol 2015;55:422–432. [DOI] [PMC free article] [PubMed] [Google Scholar]