The association of sexual minority status and bullying victimization is modified by sex and grade: findings from a nationally representative sample

This study discovered that the relationship between sexual minority status and experiencing onsite or cyberbullying in the past 12 months was significantly influenced by biological sex assigned at birth in a nationally representative sample. Among heterosexual individuals, females exhibited a higher likelihood of experiencing bullying compared to males. However, among sexual minorities, males faced a higher likelihood of being bullied than females. This effect modification by biological sex assigned at birth was significant on both multiplicative and additive scales. Additionally, we observed a negative association between bullying victimization and school grade for both heterosexual and sexual minority adolescents, with a formal test of effect modification by grade proving significant.

Beginning in 2015, the YRBSS introduced questions about students’ sexual orientation in the national high school questionnaire. This study incorporated all available nationally representative data and was the first to explore the association between sexual minority status and onsite or cyberbullying victimization in this sample.

We found that 7.1% (95% CI: 6.5–7.8) of females and 18.7% (95% CI: 17.2–20.0) of males self-reported being a sexual minority. Several factors could elucidate our results. Culturally and socially, males who express non-heteronormative orientations or identities often face heightened stigma, which may paradoxically enhance their self-recognition as part of a minority and prompt more frequent self-reporting [31, 32]. Additionally, the distinct socialization patterns for male and female students—where males may perceive their divergence from gender norms as particularly pronounced—could intensify their identification with the sexual and gender minority (SGM) community [33]. Furthermore, the more prominent representation of male SGM figures in the media and public narratives may also play a role, fostering self-identification and the propensity to disclose such an identity [34, 35].

An increased likelihood of bullying victimization among heterosexual females compared to heterosexual males was also observed in a study involving 1,402 Brazilian students [36]. It is possible that female adolescents were more likely to express femininity in Brazilian cultural norms and thus appeared more physically fragile than males. Previous research also indicated that most bullying among girls occurred within friendship or acquaintanceship groups, suggesting that heightened bullying victimization among female heterosexuals may stem from emotions triggered by their friendships, such as jealousy, suspicion, disappointment, and anger [37]. Qualitative studies revealed that girls consider their friendships extremely important and nominated the breaking of a friendship as the most anxiety-provoking aspect of school life. Disputes and conflicts related to friendship among girls can be recognized as bullying incidents with clearly identified victims [37].

Our study was in line with previous studies. Researchers found that when asked about why they thought they were bullied in the past 3 months, male sexual minority students were more likely to list their sexual orientation or what others think about their sexual orientation and how they expressed their gender as reasons, compared to female sexual minority students [38]. Similar results were reported in a Rhode Island sample where sexual minority boys reported the higher odds of being recently bullied than heterosexual boys [39].

Several factors could explain the difference in bullying victimization odds between heterosexual and sexual minority students, including homophobia and heterosexism, gender role non-conformity, social acceptability of anti-gay pejoratives, and invisibility of sexual minorities [40‐43]. Specifically, with heterosexuality dominating most schools, the gender expressions of sexual minority students, including those identifying as non-binary, deviated from their sex assigned at birth, triggering homophobia and anti-gay sentiments among students. It is also possible that sexual minority students assigned male at birth had limited participation in group activities. As group participation is a symbol of peer community belonging among high school students, limited engagement in group activities may cause male sexual minority students to appear distant from heterosexual peers, increasing the likelihood of being bullied [44].

Furthermore, the discrepancy in bullying victimization between male and female sexual minority students may be explained from a feminist perspective: male sexual minority students were often stereotyped as effeminate, and under the possible influence of misogyny, they were more vulnerable to becoming victims of bullying behavior [45]. Future studies could further evaluate the effects of these factors, as our study did not explore their possible roles in the observed outcomes.

In interpreting our findings, it was important to consider the baseline rates of bullying victimization. While our results indicate that sexual minority males have higher odds of being bullied compared to their heterosexual counterparts, this does not necessarily imply a higher absolute risk of bullying compared to sexual minority females. In fact, despite the lower odds ratio observed among sexual minority females compared to heterosexual females, the actual prevalence of bullying may be similar for sexual minority females and males. This was due to the higher baseline rate of bullying among males in general. Therefore, while sexual minority status increases the relative risk of bullying for both males and females, the absolute risk of bullying for sexual minority females and males might be more alike than the odds ratios alone suggest. This underscored the importance of considering both relative and absolute measures of risk in understanding the impact of sexual minority status on bullying victimization. Future research should further investigate these nuances to provide a more comprehensive understanding of bullying dynamics among adolescents.

We also found that although sexual minority students were more likely to be bullied, the situation improved as school grade increased. From a developmental psychology perspective, older individuals may be more psychologically mature than younger people and more likely to accept social norms, including showing greater respect for others, being less aggressive, and paying more attention to interpersonal relationships, which may contribute to a general decrease in bullying [46]. However, this finding was derived from an imputation method and when we analyzed the full data set, the interaction on the multiplicative scale disappeared. This indicated that the observed trend may not be exclusive to sexual minority students, but rather in line with a general trend in the school environment.

Our findings emphasize the importance of considering biological sex assigned at birth when developing strategies to prevent bullying among sexual minority adolescents. In a previous review, 44 interventions aimed at reducing stress among sexual minorities were systematically examined. Although all of these interventions were designed to promote bullying behavior change at multiple levels, few considered biological sex assigned at birth as a factor when designing the intervention [47]. Our study suggests that considering heterogeneity and underlying mechanisms between male and female students could potentially increase the efficacy and effectiveness of bullying behavior interventions.

This study has some limitations. First, other variables associated with both sexual minority status and bullying victimization, such as childhood adverse experiences [26, 48, 49], were not available in the dataset. Comparable constraints were present in the sex variable, as YRBS inquired about “sex” rather than explicitly focusing on “sex assigned at birth.” Consequently, we are unable to pinpoint young individuals who do not conform to the binary classification. Future research should utilize a more all-encompassing variable to encompass this data.

Besides, students selecting ‘Not sure’ in the sexual orientation question may have been uncertain regarding the wording of the question rather than their sexual orientation. YRBSS lacks an alternative for expressing ambiguity about the question’s phrasing, such as ‘I am unsure about the question.’ Consequently, we presumed that students opting for ‘Not sure’ were ambiguous about their sexual orientation. This assumption could result in misclassification of the ‘exposure’ variable, in this case, sexual orientation, potentially leading to a bias in our results towards the null hypothesis.

In addition, in our effort to streamline analyses and align with public health intervention strategies that typically address broader groups, we categorized “Gay or Lesbian” and “Bisexual” identities into a single “sexual minority” variable. While this facilitated a clear comparative analysis, we recognize it may not fully capture the distinct experiences within each subgroup. Consequently, our study further implemented stratification analysis across different subgroups within sexual minorities. This analysis revealed a significant sex-based effect modification on the multiplicative scale for both gay/lesbian and bisexual students. In contrast, on the additive scale, this sex-based effect modification was significant only in the comparison between gay/lesbian and heterosexual students, while it was not significant when comparing bisexual and heterosexual students. Furthermore, when assessing the effect modification by grade level, our findings indicated a significant grade-based effect modification on the multiplicative scale for both gay/lesbian and bisexual students. However, this grade-based effect modification was not significant on the additive scale when comparing gay/lesbian and bisexual students to their heterosexual counterparts. Comprehensive details of these findings are provided in the supplementary material.

We also recognize that treating sex and sexual identity as independent variables may oversimplify the complex interplay between these identities. Intersectionality theory stipulates that these aspects of identity are interconnected and cannot be fully understood in isolation [7, 8]. Aligned with recommendations in intersectionality literature, we interpret the interaction between sex and sexual identity on the additive scale to assess whether intersecting elements of identity are greater than their sum and thus not independent [9].

Second, due to the self-reporting nature of the YRBSS questionnaire, this study was subject to recall bias. Furthermore, we only considered the victimization aspect of bullying behavior, even though bullying perpetration is also crucial in understanding and preventing bullying behavior [50]. However, acquiring such information is difficult using self-reported questionnaire data, as few students would admit to bullying others.

Moreover, our study primarily investigated sex and grade as effect modifiers to guide public health programs for students based on their sex assigned at birth and educational level. While race/ethnicity might also serve as an effect modifier, our focus was not on developing interventions for specific racial/ethnic groups, given the associated implementation complexities. Additionally, the YRBSS data used in our study does not comprehensively capture the nuances of race/ethnicity, which include both phenotypic traits and cultural identities, limiting our ability to fully explore its impact as a modifier.

It is important to acknowledge that the missingness in the sexual identity variable may not be random. This non-random nature of missing data is likely due to the reluctance or policy of many schools to include questions about sexual identity in their surveys, resulting in a systematic absence of this data. For sensitivity analysis, we conducted a complete case analysis. This additional analysis supports the robustness of our main conclusions, despite the noted potential bias in data collection.