Emotional Changes Following Discrimination Induction in Gender- and Sexuality-Diverse Adolescents

Abstract

Sexual and gender minority (SGM)–identifying adolescents are particularly vulnerable to negative psychological outcomes, including engagement in nonsuicidal self-injury (NSSI). However, little is known about why these relationships exist. We used experimental methods to test the psychological mediation framework in an online sample of 328 adolescents who reported female sex at birth and a range of sexual and gender identities. Participants reported on depressive symptoms, self-criticism (both self-report and implicit), NSSI, and discrimination. They also completed a discrimination-based mood induction to test emotional reactivity. At baseline, SGM participants reported higher levels of implicit and self-reported self-criticism, depressive symptoms, discrimination, and higher rates of NSSI compared with cisgender, heterosexual participants (ps < .03). Following the discrimination induction, SGM-identifying participants exhibited larger emotional reactivity compared with cisgender heterosexual participants, as measured by change in negative mood, F(1, 326) = 7.33, p = .01, ${\eta }_{\text{p}}^2=.02$, and state self-criticism, F(1, 326) = 4.67, p = .03, ${\eta }_{\text{p}}^2=.014$, but not implicit affect toward the self. This effect was associated with baseline depressive symptoms, self-criticism, NSSI history, and discrimination. Post hoc analyses revealed that participants who tended to reframe experiences of discrimination as opportunities for growth exhibited attenuated emotional reactivity to the induction; findings remained significant after adjusting for SGM status and event severity (ps < .001). Results indicate that adolescents identifying as SGM may experience elevated psychological distress compared with their cisgender heterosexual peers and that stigma-related stressors may increase emotion dysregulation and maladaptive cognitive styles, paralleling previously proposed psychological mediation models.

Individuals identifying as a sexual minority (SM; e.g., lesbian, gay, bisexual, queer) and gender minority (GM; i.e., people who identify with a gender other than their sex assigned at birth) tend to be at higher risk for a wide range of physical and mental health problems compared with their heterosexual counterparts (Cochran, 2001; Smalley, Warren, & Barefoot, 2016; Steever, Francis, Gordon, & Lee, 2014). Among adolescents and young adults, sexual and gender minority (SGM)-identifying youth report higher rates of internalizing (Riley, Kirsch, Shapiro, & Conley, 2016) and externalizing symptoms (Hatzenbuehler, Corbin, & Fromme, 2008; Seil, Desai, & Smith, 2014; Ziyadeh et al., 2007). Perhaps even more alarmingly, these youth tend to show disproportionate experiences with self-injurious thoughts and behaviors (SITBs; Batejan, Jarvi, & Swenson, 2015; King et al., 2008; Marshal et al., 2011), including nonsuicidal self-injury (NSSI; Fox, Choukas-Bradley, Salk, Marshal, & Thoma, 2020; Walls, Laser, Nickels, & Wisneski, 2010) or intentional self-harm enacted without suicidal intent (Nock, 2010).

Less research has been conducted to understand why SGM-identifying youth report higher rates of these negative health outcomes (Mustanski, 2015). Increasingly, however, these disparities are understood in the context of the minority stress model (Meyer, 2003). According to this model, SGM-specific stressors contribute to psychological distress that then predisposes SGM-identifying individuals to negative psychological outcomes. Providing support for this model, sexual minority (SM) adolescents report higher rates of bullying and victimization than their peers (Coulter, Herrick, Friedman, & Stall, 2016). Moreover, experiences of victimization among SM adolescents are associated with higher rates of past year suicidal ideation, plans, and attempts (Bouris, Everett, Heath, Elsaesser, & Neilands, 2016); development of severe mental illness (Kidd, Howison, Pilling, Ross, & McKenzie, 2016); and higher levels of psychological distress in adulthood (Birkett, Newcomb, & Mustanski, 2015). Taken together, this research suggests that minority stress may relate to SITB rates among SGMs.

Although useful, most research on this topic to date has been correlational. Moreover, the minority stress model is agnostic to the question of how minority stressors increase risk. In an effort to shed light on this topic, Hatzenbuehler (2009) proposed a psychological mediation framework in which minority stressors lead to increased emotion dysregulation and maladaptive cognitive styles, which then increase an individual’s risk for psychopathology. Providing initial support for this model, a recent study found that discrimination-based stressors are associated with greater use of maladaptive emotion regulation strategies (e.g., suppression, rumination) and, moreover, that ruminating (vs. distracting) after a discrimination-based induction is associated with greater distress (Hatzenbuehler, Dovidio, Nolen-Hoeksema, & Phills, 2009). Similarly, another study found that rumination mediated the relationship between microaggression and depression symptoms in a sample of SM youth (Kaufman, Baams, & Dubas, 2017). Currently lacking, however, are experimental investigations of this model in SGM adolescents.

The present study sought to test key assumptions of the psychological mediation framework in an online sample of SGM-identifying youth to better understand the elevated risk of NSSI observed in this population. We specifically aimed to examine adolescents aged 14–15 years because SGM identity (Calzo, Antonucci, Mays, & Cochran, 2011) and SITB engagement (Nock et al., 2013) begin to develop during this period. In addition to comparing rates and severity of discrimination, psychopathology, maladaptive cognitive styles (i.e., self-criticism), and NSSI, we aimed to address the following questions:

• Question 1: Do SGM-identifying youth demonstrate greater emotional reactivity and self-criticism in response to a discrimination-based mood induction compared with their cisgender, heterosexual peers?

The psychological mediation framework posits that those identifying as SGM are at higher risk for mental health problems because minority stressors lead to emotion dysregulation and negative cognitive styles (Hatzenbuehler, 2009). We were interested in testing experimentally induced emotion reactivity and self-criticism as measures of emotion dysregulation and maladaptive coping styles, respectively. We focused on these constructs because there is substantial research linking self-criticism (e.g., Fox, et al., 2018; Gilbert et al., 2010; Glassman, Weierich, Hooley, Deliberto, & Nock, 2007; St Germain & Hooley, 2012) and self-reported emotion reactivity (Franklin et al., 2013; Glenn, Blumenthal, Klonsky, & Hajcak, 2011; Nock, Wedig, Holmberg, & Hooley, 2008) with NSSI. As such, these constructs may be particularly useful in understanding high rates of NSSI observed among SGM youth.

If SGM-identifying youth demonstrate greater emotional reactivity (i.e., increase in negative affect) and reactive self-criticism (i.e., increase in negative self-thoughts) in response to the discrimination-based mood induction, results would support the psychological mediation framework. Alternatively, if SGM participants exhibit similar or less emotional reactivity and self-criticism, results might suggest that minority status is unrelated or leads to habituation to discrimination experiences. This is a possibility because habituation has been observed in some studies of racial and ethnic minorities (e.g., Gaylord-Harden, Cunningham, & Zelencik, 2011; Gevonden et al., 2016) and could extend to SGM populations.

• Question 2: Are experiences of social stress (e.g., discrimination) associated with greater emotional reactivity, self-criticism, and depressive symptoms?

Regardless of overall group differences, the present study aimed to assess relationships among discrimination experiences, experimentally induced emotional reactivity, both trait and reactive self-criticism, and depressive symptoms. If more frequent and severe experiences of minority-related victimization are associated with higher levels of emotion reactivity and state and trait self-criticism, results would support the psychological mediation framework. However, if higher levels of discrimination are associated with an attenuated emotional response, results would again support habituation to these experiences.

Method

Participants

Participants were 328 adolescents (mean age = 14.66 years; SD = 0.51) recruited online from social media websites related to the lesbian, gay, bisexual, transgender, and questioning (or queer) + community. Inclusion criteria required participants to be 14–15 years of age, report female sex assigned at birth, and be fluent in English. In addition, participants had to identify with one or more of the following identities: heterosexual female (n = 91), sexual minority female (including homosexual, bisexual, or questioning/other; n = 179), or gender minority (i.e., reporting a gender identity that was incongruous with biological sex at birth; n = 58). Of note, only two of the gender minority participants identified as heterosexual.¹ Participants primarily identified as White (n = 217), with the remaining participants identifying as Latinx (n = 20), Black (n = 18), Asian (n = 36), or another race/ethnicity (n = 38). Sample size was determined by feasibility of recruiting an online sample of adolescents meeting these criteria rather than by a power calculation.

We chose to include only those participants who were assigned female sex at birth for several reasons. Regarding those identifying as SGM, prior research highlights that there are differences in the ways that females and males can be affected by sexual orientation discrimination and harassment (Mitchell, Ybarra, & Korchmaros, 2014). Additionally, among those identifying as gender minority (GM), there is evidence those assigned female at birth show the highest risk for psychopathology and SITB outcomes (Eisenberg et al., 2017; Petersen & Hyde, 2011; Thoma et al., 2019; Toomey, Syvertsen, & Shramko, 2018). Among those identifying as cisgender, people assigned female at birth tend to show higher risk for nonfatal SITBs (Fox, Millner, Mukerji, & Nock, 2018).

Measures

Demographic information.

A variety of questions were used to assess basic demographic information (e.g., age, sex, date of birth, sexual orientation). Categorical self-report items were used to assess biological sex (female; male; other; prefer not to say), gender identity (female; male; female → male; male → female; other; prefer not to say), and sexual orientation (straight/heterosexual; gay/lesbian/homosexual; bisexual; unsure/questioning/other; prefer not to say). However, we also used three continuous self-report items adapted from the Klein Sexual Orientation Grid (Klein, Sepekoff, & Wolf, 1985) to assess current sexual/romantic attraction, current romantic/sexual behavior, and past romantic/sexual behavior toward both boys/masculine people and girls/feminine people. Qualitative review of the current sample indicated that responses to the single-item measure of sexual orientation were similar to the dimensional responses to the Klein Sexual Orientation Grid, and sexual orientation groups were fairly consistent (e.g., participants who indicated cisgender heterosexual female identities also indicated strong attraction to males and little or no attraction to females). Because responses were largely consistent within individual participants and because the hypotheses driving this study are most closely related to identity, we chose to categorize participants based on the identities they reported in the categorical item. Gender minority participants were classified as those who reported a gender identity that was not female (because all participants were required to have female sex at birth).

Children’s Depression Inventory 2 (CDI-2; Kovacs, 2010).

The CDI-2 is a 28-item, self-report measure of depressive symptoms in children over the past 2 weeks. Each item consists of three statements (e.g., “I am sad once in a while,” “I am sad many times,” “I am sad all the time”). Participants are asked to select the response that best describes their feelings over the past 2 weeks. The CDI-2 has good internal consistency with Cronbach’s alpha ranging from 0.67 to 0.91 for all subscales tested in participants aged 7–17 years (Bae, 2012). The present study removed the item assessing suicidal ideation.

Abbreviated version of the SITBI (Nock, Holmberg, Photos, & Michel, 2007).

This 14-item self-report measure assessed participants’ lifetime history with self-injurious thoughts and behaviors, including NSSI. Some items assessed any NSSI engagement (e.g., “Have you ever purposely hurt yourself without wanting to die?”), and others assess lifetime prevalence (e.g., “How many times in your life have you purposely hurt yourself without wanting to die?”). We also included additional language to clarify the types of behaviors we were assessing. Specifically, prior to asking about NSSI engagement, we stated, “In answering these questions, please do not include minor ways of hurting yourself, like picking at wounds, biting your lips or nails, or getting tattoos or piercings. Also, do not include indirect ways of hurting yourself, such as starving yourself or overdosing without wanting to die.” The abbreviated SITBI has been validated in an adolescent sample (Nock et al., 2007).

Self-Rating Scale (Hooley, Ho, Slater, & Lockshin, 2010).

The self-rating scale is an eight-item scale assessing participants’ feelings of self-criticism and negative self-worth. All items are rated on a Likert scale from 1 (strongly disagree) to 7 (strongly agree). Sample items include “Sometimes I feel completely worthless” and “If others criticize me, they must be right.” The Self-Rating Scale has demonstrated strong internal consistency (α = .73) in previous research (Hooley et al., 2010).

Expanded Everyday Discrimination Scale (Williams et al., 2008).

The Everyday Discrimination Scale measures the extent to which participants experience chronic, every day, relatively minor discrimination (e.g., “You receive poorer service than other people at restaurants or stores”). Participants are asked to rate each experience from 1 (never) to 5 (almost every day). The expanded version of the scale adds a tenth item (“You are followed around in stores”) to the original nine-item measure to capture a wider range of discrimination experiences. The original scale has been validated in an adolescent sample (Clark, Coleman, & Novak, 2004).

Mood and self-criticism Visual Analog Scales (VAS).

Participants were shown VASs to assess mood and self-criticism. Each VAS consisted of a horizontal line from 1 to 100. Participants were asked to indicate on the line how negatively they are feeling right now (to assess negative mood), and how negatively they are feeling about themselves right now (to assess self-criticism). This measure was administered before and after the mood induction (T1 and T2) to assess changes in mood and feelings about the self.

Self-criticism Affect Misattribution Procedure (AMP; Franklin, Lee, Puzia, & Prinstein, 2014).

The AMP is a brief computer-based task that measures implicit affect. On each trial of the AMP, the program flashes an emotional picture for 75 ms, a blank screen for 125 ms, an ambiguous Chinese symbol for 100 ms, and finally a gray screen that remains until the participant presses a key. Participants were instructed to press the p key if they felt that the Chinese symbol was more pleasant compared with the average Chinese symbol and the q key if they felt the symbol was less pleasant compared with the average Chinese symbol. Importantly, participants were told to ignore the emotional pictures and instead to rate the Chinese symbols only. Several studies have indicated that participants’ evaluations of the Chinese characters are influenced by the nature of the picture that flashes before it, with more pleasant pictures generating more pleasant evaluations of subsequent Chinese symbols (see Payne, Cheng, Govorun, & Stewart, 2005). This is believed to occur because of misattributing the feelings evoked from the picture to the more ambiguous Chinese symbol. Potentially because of this misattribution, the AMP measures implicit affective reactions to the emotional pictures that are flashed at the beginning of each trial. Of note, participants who reported familiarity with Chinese characters were excluded from analysis.

The stimuli in the present study included words related to the self (i.e., me, my, I, myself, and mine). These stimuli have been used in prior online work as an implicit measure of self-criticism (Fox, et al., 2018; Franklin et al., 2016). As with the VAS, participants completed the full measure once at baseline (T1) and again after completing the discrimination manipulation (T2).

Procedure

The present analysis was part of a larger project that has been reported elsewhere (Smith, Wang, Carter, Fox, & Hooley, 2020). Briefly, we used a variety of online and offline recruitment techniques in an effort to reach a large number of participants within the target demographic. We joined online social media websites (e.g., Facebook, Tumblr, online forums) and posted study advertisements in online communities related to SGM populations. We also ran advertisements on Facebook and Instagram, targeting user profiles that had shown interest in terms related to SGM identities. Additionally, we distributed flyers in and around high schools with a link to the online study. All recruitment materials described the study as an online study about the ways that adolescents feel about themselves in an effort to avoid any priming effects by mentioning SGM identities or NSSI engagement.

Interested members of these communities completed a brief screening questionnaire to assess inclusion criteria (i.e., age, English fluency, computer access, female sex assigned at birth, gender identity, and sexual orientation). Those who qualified were e-mailed a link to the full assessment. After completing the study, participants were paid $5 in online gift cards to either Starbucks or Amazon. We recommended that participants use an e-mail address that did not contain identifying information such as name or date of birth to maintain anonymity.

We provided all participants with online and phone resources, both within and outside the United States. Additionally, when participants indicated risk of future self-injury, we e-mailed them directly. In this e-mail, we mentioned that it seemed like they were going through a difficult time and encouraged them to use the provided resources at any point, especially if they did not feel safe. All study materials, measures, methods, and procedures were approved by the Harvard University Committee on the Use of Human Subjects.

The study followed a quasiexperimental design. First, participants completed the VAS measures of negative mood and self-criticism and rated AMP stimuli at baseline (T1). Next, they completed the battery of questionnaires, detailed above. Participants completed these questionnaires after the AMP to avoid potential priming effects, in which the questionnaires might invoke feelings similar to stereotype threat.

Discrimination Induction

The next part of the study was intended to induce feelings of discrimination. Adapting a previously used manipulation (Bastian, Jetten, & Fasoli, 2011), participants were asked to “think about all of the times in which you experienced discrimination, bullying, or a negative outcome because of who you are/think of the event that had the biggest negative impact on you.” We included one item to assess perceived severity of the event: “How bad was this event, on a scale of 1–10?” After responding to this item, participants were then instructed to write about the event for 5 min, including how they felt about themselves after the event occurred and all of the consequences of that event. If participants ran out of things to write, they were instructed to read over their writing until 5 min had passed. This discrimination induction procedure follows a model commonly used in psychological research to elicit negative moods in children as young as 4 years (Masters, Barden, & Ford, 1979; Nasby & Yando, 1982).

After the discrimination manipulation, participants completed the VAS ratings and the AMP an additional time (T2) to allow for examination of the effects of the induction on these variables. Finally, at the end of the study, all participants experienced a positive mood induction to counteract any potential negative effects from previous parts of the study.

Results

Descriptive Statistics and Preliminary Analyses

Across the whole sample, symptoms of psychopathology were quite high; approximately half the sample (n = 167, 50.9%) reported lifetime NSSI engagement, even though this was not a factor in participant recruitment. Moreover, mental health variables were moderately correlated with each other and with levels of discrimination (see Figure 1 for means, SDs, and Pearson correlations across depressive symptoms, trait self-criticism and implicit affect toward the self, discrimination, and NSSI history).

Figure 1.

Depressive symptoms measured by Childhood Depression Inventory (M = 21.51; SD = 10.47); self-criticism measured by Self-Rating Scale (M = 35.44, SD = 10.96); implicit affect toward the self measured by the Affect Misattribution Procedure self-score was 64.09 (SD = 22.59). Discrimination was measured through the Everyday Discrimination Scale (M = 16.66, SD = 9.92). p < .05 was used for statistical significance, however, a conservative Bonferroni correction for these analyses would place significance at p < .008.

Consistent with hypotheses, independent-sample t tests revealed that SGM-identifying participants showed significantly higher levels of self-criticism, depressive symptoms, and discrimination and significantly lower levels of implicit affect toward the self (i.e., AMP self) compared with cisgender heterosexual participants (see Figure 2).² A larger proportion of SGM participants identified as White (73.3%) compared with cisgender heterosexual participants (48.35%, χ²[5] = 24.25, p < .001; see Supplemental Figure 1). Race was not significantly associated with the results reported in the remainder of the article. In addition, SGM participants reported higher event severity for the discrimination exercise (mean = 6.32, SD = 2.57) compared with cisgender heterosexual participants (mean = 5.63, SD = 2.65, t[156] = 2.10, p = .04).

Figure 2.

SM only = participants identifying as sexual minority and cisgender; SM + GM = participants identifying as gender minority (almost all identified as sexual minority as well). p < .05 was used for statistical significance, however, a conservative Bonferroni correction for these analyses would place significance at p < .02 for each panel.

Discrimination-Based Negative Mood Induction

We conducted three 2 (occasion: pre-, post-) × 2 (group: SGM vs. cisgender heterosexual) repeated-measures ANOVAs to examine changes in negative mood, state self-criticism, and implicit affect toward the self before and after the discrimination induction, and to examine how group membership impacted those changes.

Results indicated a main effect of occasion on negative mood, F(1, 326) = 20.80, p < .001, ${\eta }_{\text{p}}^2=.06$, and on state self-criticism, F(1, 326) = 4.93, p = .03, ${\eta }_{\text{p}}^2=.02$, but not implicit affect toward the self, F(1, 291) = 0.62, p = .43, ${\eta }_{\text{p}}^2=.002$. In other words, negative mood and state self-criticism, but not implicit affect toward the self, increased across the discrimination-based mood induction.

Additionally, there was a significant Occasion × Group interaction for negative mood, F(1, 326) = 7.33, p = .01, ${\eta }_{\text{p}}^2=.02$, and state self-criticism, F(1, 326) = 4.67, p = .03, ${\eta }_{\text{p}}^2=.014$. No Occasion × Group interaction emerged for implicit affect toward the self, F(1, 291) = 0.01, p = .91, ${\eta }_{\text{p}}^2=.00$. Estimated marginal means indicated that participants identifying as SGM showed larger increases in state negative mood and self-criticism as a function of the mood induction compared with those participants identifying as cisgender heterosexual (see Figure 3).³

Figure 3.

Error bars represent standard error. SGM = sexual and gender minority.

Factors Associated With Changes in Mood After the Discrimination Induction

In light of baseline differences in mental health and discrimination variables between SGM and cisgender heterosexual participants and differences in the severity of events reported, we did not include these variables as covariates in the repeated-measures ANOVA (Miller & Chapman, 2001). Instead, we used Pearson correlations to test the possibility that changes in negative mood and state self-criticism across the discrimination induction (i.e., a variable created by subtracting negative mood and state self-criticism prediscrimination induction from negative mood and state self-criticism postdiscrimination induction) would be impacted by mental health variables, severity of the event reported, and self-reported discrimination (see Figure 4). As expected, changes in negative mood were significantly and moderately associated with baseline symptoms of depression, trait self-criticism, NSSI history, severity of the specific event, and discrimination. Changes in state self-criticism were not significantly associated with baseline symptoms of depression, severity of the event, or NSSI history, although they did show a small association with discrimination and trait self-criticism. In other words, greater mood reactivity to the discrimination induction was associated with baseline mental health and discrimination variables, with higher levels of those variables linked to small but significantly greater mood reactivity. These effects were smaller and less consistent when considering self-criticism reactivity.

Figure 4.

Neg = negative. p < .05 was used for statistical significance, however, a conservative Bonferroni correction for these analyses would place significance at p < .007.

Of the 226 SGM participants who completed the writing exercise, 84 mentioned their SGM identity within their response. Of the 85 cisgender heterosexual participants who completed the writing exercise, none wrote about SGM identity. Writing about SGM identity was not associated with a greater self-reported event severity. There were no significant differences in change in negative mood between SGM participants who did versus did not mention SGM identity (p = .33).

To test the possible confounding effect of event severity on the results of the discrimination mood induction, we next conducted a multivariate linear model using event severity and SGM status as predictors of change in negative mood. This yielded a significant regression model, F(2, 315) = 8.12, p < .001, with R² = .05. In this model, the effect of event severity was significant (B = 0.94, p < .01), and SGM identity was significantly associated with a larger increase in negative mood (B = 4.58, p < .02). Results indicate that SGM identity remained associated with change in negative mood after adjusting for self-reported event severity.

Post Hoc Tests of Resilience

In reviewing participants’ qualitative responses to the negative mood induction, we noticed that many participants ended their written response describing how they grew, or positively changed, as a result of their experience. We were interested in testing whether these positive, cognitive reframes buffered responses to the discrimination induction. To test this post hoc hypothesis, responses were coded for the presence (1) or absence (0) of these positive reframes. Cisgender heterosexual (33 of 85; 38.8%) participants were significantly more likely than SGM participants (52 of 225; 23%) to spontaneously use cognitive reframing, χ²(1) = 7.00, p < .01. Using one-sample t tests, we tested whether the use of this reframe strategy was associated with blunted changes in negative mood and self-criticism across the induction. Results showed that use of this strategy resulted in significantly less reactivity, both in terms of changes in negative mood and self-criticism (see Supplemental Figure 2). Adjusting for self-reported event severity yielded similar results, as did adjusting for SGM status. In other words, spontaneous use of this cognitive reframing strategy was associated with reduced reactivity to the discrimination-based mood induction.

Discussion

The psychological mediation framework proposes that people identifying as SGM experience minority stress, that these stigma-related stressors increase emotion dysregulation and maladaptive cognitive styles, and that these maladaptive processes serve to increase risk for psychopathology (Hatzenbuehler, 2009). Results of the present study supported the first two components of this framework using both self-report and experimental measures of emotion dysregulation and maladaptive cognitive styles in an online sample of adolescents identifying as SGM and cisgender heterosexual.

Paralleling prior research (e.g., Brown & Jones, 2016), SGM-identifying participants reported significantly, although modestly, higher levels of discrimination, self-criticism, experimentally induced emotion reactivity, depression, and NSSI compared with those identifying as cisgender heterosexual. Consistent with the psychological mediation framework, levels of discrimination were significantly and positively associated with each of these outcome variables. In addition, the relationship between SGM status and emotional reactivity remained significant after adjusting for self-reported event severity, indicating that the difference in reactivity is only partially explained by differences in severity between SGM and cisgender heterosexual participants. Results are consistent with prior research linking minority identity-based stressors with psychological distress in previous studies of SM populations (Kelleher, 2009) and racial and ethnic minorities (e.g., Rucker, Neblett Jr, & Anyiwo, 2014).

Higher self-reported depressive symptoms, event severity, self-criticism, NSSI history, and discrimination frequency were each associated with a greater response to the discrimination induction. These relationships indicate a possible pathway through which negative psychological processes (which were higher in SGM participants) lead to greater emotional responding to discrimination experiences. Results are consistent with prior research demonstrating that SM-specific victimization mediated the effect of SM status on depressive symptoms and suicidality in adolescents (Burton, Marshal, Chisolm, Sucato, & Friedman, 2013). Studies of social stressors in undergraduates have also found subjective social status to moderate cortisol response to a social evaluative stressor (Gruenewald, Kemeny, & Aziz, 2006).

In addition to demonstrating a pathway of risk, the present study also sheds light on a potential resilience factor: cognitive restructuring. Indeed, similar types of reappraisal have been noted in qualitative studies of SM youth (Erhard & Ben-Ami, 2016). Moreover, in studies of racial and ethnic minorities, greater use of cognitive reframing has been associated with better psychological functioning in the face of discrimination (Juang et al., 2016; Soto et al., 2012). In the current study, participants who spontaneously reframed their experience of discrimination as an event through which they grew (e.g., “[This experience] made me much stronger as a person ….”) exhibited attenuated emotional reactivity, in terms of negative mood and self-criticism, to the discrimination induction. This restructuring was associated with significantly lower levels of depressive symptoms (but not NSSI). These exploratory results highlight that such restructuring could be a protective factor to consider in understanding both risk and resilience for negative mental health outcomes. Specifically, findings support the value of cognitive restructuring as a target for clinical interventions, consistent with previous research investigating cognitive-behavioral therapy and cognitive restructuring in adolescents (e.g., Cândea et al., 2018; Spirito, Esposito-Smythers, Wolff, & Uhl, 2011). It is also possible, however, that emotionally healthier participants were more able to reframe.

The present study should be considered in light of its limitations. First, we did not include participants assigned male sex at birth. Given previous research showing that sex assigned at birth is related to key differences in psychopathology, both in and outside the context of SGM identity (e.g., Mitchell et al., 2014; Thoma et al., 2019), the present results are likely not generalizable to those assigned male sex at birth. Second, results from the present investigation are both cross-sectional and experimental, and were limited because of a difference in event severity between groups during the writing exercise. Therefore, it is challenging to assess causality within the examined relationships, both in terms of self-reported discrimination (because perhaps of memory biases) and emotional reactivity/cognitive biases. It is also possible that the order of study measures (i.e., completing baseline measures prior to the mood induction) could have led to a priming effect that may have influenced the efficacy of the mood induction. Future longitudinal investigations of this framework should include three or more time points to more thoroughly investigate this model and particularly to test whether emotional reactivity and cognitive biases mediate relationships among minority stress and psychopathology. Future studies could additionally use tasks that are not directly related to discrimination (e.g., the Trier social stress task; Kirschbaum, Pirke, & Hellhammer, 1993) to evaluate emotion reactivity in response to unrelated stressors. Third, because participants were recruited from online communities related to lesbian, gay, bisexual, transgender (LGBT) interests (including many cisgender heterosexual participants), it is possible that sample bias may have affected the results (Meyer & Wilson, 2009). In particular, because many of the cisgender heterosexual participants were recruited from LGBT environments, it is possible that they experience more SGM-related stressors than the general cisgender heterosexual population (e.g., by spending time with SGM peers or exposing themselves to SGM-related discrimination). Although none of the cisgender heterosexual participants who completed the writing exercise mentioned SGM identity, it is possible that these experiences may exist and have an effect on emotional reactivity. However, this bias would lead to a smaller effect size when comparing between SGM and cisgender heterosexual populations, meaning that the true differences may be larger than those found in the present study. Finally, the measurement of sexual and gender identity is a rapidly evolving field, and the use of a categorical variable to classify sexual and gender identity may restrict the conclusions that can be drawn from such a heterogeneous group of identities.

Overall, this investigation sheds light on the factors that underpin the relationship between SGM status and negative outcomes. As such, it has important implications for this at-risk population. In particular, this is the first experimental investigation of discrimination experiences in SGM-identifying youth, and results indicate that SGM status may be related to negative psychological processes that can exacerbate the effects of discrimination. Given the relationship between stigma and health inequality (Hatzenbuehler, Phelan, & Link, 2013), this study presents important implications for future research and for our clinical understanding of the experiences of SGM adolescents.