A weekly diary study of minority stress, coping, and internalizing symptoms among gay men

Abstract

Objective:

Research has demonstrated that gay men are at increased risk for internalizing disorders compared to heterosexual men and that minority stressors are risk factors. However, the mechanisms underlying the associations between minority stressors and internalizing symptoms remain unclear. The current study examined coping strategies (active and disengaged coping) as mediators of the associations between minority stressors (internalized homonegativity and rejection sensitivity) and internalizing symptoms.

Method:

A sample of 147 gay men completed a baseline questionnaire and weekly questionnaires for seven consecutive weeks.

Results:

At the between-person level, higher internalized homonegativity and rejection sensitivity were associated with higher disengaged coping, but not active coping. In turn, higher disengaged coping was associated with higher internalizing symptoms. Disengaged coping mediated the between-person association between internalized homonegativity and internalizing symptoms. At the within-person level, higher internalized homonegativity and rejection sensitivity were associated with higher disengaged coping, which in turn was associated with higher internalizing symptoms. Higher rejection sensitivity was also associated with higher active coping. Disengaged coping mediated the within-person associations between both minority stressors and internalizing symptoms. Of note, some associations with internalized homonegativity became non-significant controlling for rejection sensitivity, suggesting that the latter has a stronger influence on coping and internalizing symptoms.

Conclusions:

Findings demonstrate that gay men’s negative thoughts and feelings about their sexual orientation and anxious expectations of rejection vary from week to week and this weekly fluctuation has an impact on mental health. Further, findings implicate disengaged coping as a mechanism through which minority stressors influence internalizing symptoms.

Epidemiological studies have demonstrated that gay men are at increased risk for mood and anxiety disorders compared to heterosexual men (Bostwick, Boyd, Hughes, & McCabe, 2010; Cochran & Mays, 2009; Cochran, Sullivan, & Mays, 2003; Gilman, Cochran, Mays, Hughes, Ostrow, & Kessler, 2001; Meyer, 2003; Sandfort, de Graaf, Bijl, & Schnabel, 2001). Minority stress theory proposes that sexual minorities experience stress associated with their stigmatized social status (referred to as minority stress), which accounts for their increased risk for psychological disorders (Meyer, 2003). Meyer (2003) placed minority stressors on a continuum of proximity to the individual, where distal minority stressors refer to external, objective experiences (e.g., discrimination, victimization) and proximal minority stressors refer to internal, subjective experiences (e.g., the internalization of negative societal attitudes toward non-heterosexuality, or internalized stigma; anxious expectations of rejection, or rejection sensitivity). Research has consistently demonstrated that minority stressors are associated with negative mental health outcomes for sexual minorities, particularly internalizing (depression and anxiety) symptoms (e.g., Feinstein et al., 2012; Hatzenbuehler, McLaughlin, Keyes, & Hasin, 2010; Herek, Gillis, Cogan, & Glunt, 1997; Herek et al., 1999; Huebner, Rebchook, & Kegeles, 2004; Mays & Cochran, 2001; Newcomb & Mustanski, 2010). However, the mechanisms underlying these associations are unclear.

There is reason to believe that how sexual minorities cope with distal minority stressors such as discrimination may be one mechanism underlying the associations between proximal minority stressors and internalizing symptoms. Specifically, negative thoughts/feelings about one’s sexual orientation and anxious expectations of rejection may lead sexual minorities to cope with discrimination in ways that maintain or exacerbate psychological distress. Building on minority stress theory, Hatzenbuehler’s (2009) psychological mediation framework proposed that minority stressors lead to disruptions in psychological processes, which then contribute to mental health problems. Coping strategies represent one type of psychological process that may mediate the associations between minority stressors and mental health problems. Unfortunately, coping has received surprisingly little attention in the literature on mental health among sexual minorities and it has been conceptualized in such diverse ways that it is difficult to draw conclusions about its role.

Several studies have demonstrated associations between coping styles and mental health problems among sexual minorities using traditional taxonomies of coping strategies. A common taxonomy is the engagement-disengagement distinction, which focuses on orientation toward or away from stress. Engagement coping (also referred to as approach or active coping) involves actively dealing with stressors or related emotions, whereas disengagement coping (also referred to as avoidance coping) involves escaping stressors or subsequent emotions (e.g., Moos & Schaefer 1993; Roth & Cohen 1986; Skinner et al. 2003). In general, engagement coping is associated with improved well-being and disengagement coping is associated with negative mental health outcomes (Compas, Connor-Smith, Osowiecki, & Welch, 1997; Compas, Connor-Smith, Saltzman, Thomsen, & Wadsworth, 2001). Similar findings have been demonstrated in sexual minority samples. For instance, higher active coping was associated with lower depressive symptoms among lesbians and gay men (Zea, Reisen, & Poppen, 1999), and higher avoidant coping was associated with worse mood among HIV-positive gay men (Nicholson & Long, 1990). Additionally, Lehavot (2012) found that higher maladaptive coping and lower adaptive coping were associated with worse mental health among sexual minority women.

A few studies also provide support for coping as a mechanism underlying the associations between minority stress and mental health problems. Bianchi and colleagues (2004) found that higher discrimination was associated with lower active coping among HIV-positive gay men, which was associated with poorer health behaviors (e.g., sleep, diet, exercise) (Bianchi, Zea, Poppen, Reisen, & Echeverry, 2004), but they did not test mediation. Similarly, Nicholson and Long (1990) found that higher internalized homophobia was associated with higher avoidance coping among HIV-positive gay men, which was associated with worse mood; again, they did not test mediation. In samples of sexual minority women, Szymanski and colleagues found that avoidance coping and social support mediate the association between internalized heterosexism and psychological distress (Szymanski & Kashubeck-West, 2008; Szymanski & Owens; 2008). Kaysen and colleagues (2014) found that maladaptive coping mediated the association between internalized homophobia and psychological distress also among sexual minority women. In sum, several studies provide support for coping strategies as mechanisms underlying the associations between minority stressors and mental health outcomes, but significant gaps in our understanding of these associations remain.

A major limitation of research on how sexual minorities cope with stress is that studies typically assess coping in response to stress in general rather than in response to discrimination. It is likely that there are differences between how sexual minorities cope with general life stress compared to minority stress, given that individuals make different attributions for stressors they perceive to be discriminatory versus non-discriminatory (Major, Kaiser, & McCoy, 2003; Schmitt & Branscombe, 2002; Schmitt et al., 2003). Additionally, the aforementioned studies on how sexual minorities cope with stress were cross-sectional. The conclusions that can be drawn from cross-sectional studies like these are limited to how a person’s general tendency to cope with stress in a certain way relates to their general well-being.

To address these limitations, the current study examined coping specifically in response to discrimination using an intensive repeated-measures design (a weekly diary). Diary designs have several advantages over cross-sectional designs. First, they decrease recall bias (e.g., inaccurate recollection of events) and bias toward the over-reporting of severe incidents. Second, they allow for tests of within-person effects (e.g., change over time for a given individual). This can tell us if gay men’s reports of minority stress, internalizing symptoms, and coping strategies are stable from week to week or if they vary. Finally, they can inform our understanding of how a gay man’s levels of internalized homonegativity and rejection sensitivity on a given week influence how they cope with discrimination and their levels of internalizing symptoms that same week. This provides a stronger test of these associations by capturing real-life events and coping responses as they are happening week-to-week in the lives of gay men. Further, if these associations are present at the within-person level, then it will highlight the need to consider the fluctuating nature of minority stress and coping strategies in order to fully understand what puts gay men at increased risk for mental health problems. In sum, examining the associations among minority stress, coping, and internalizing symptoms using a diary design has the potential to inform mental health interventions for gay men by elucidating treatment targets (e.g., coping strategies that contribute to worse mental health) and individual differences that influence the use of these specific coping strategies.

The Current Study

The current study examined coping strategies (active and disengaged coping) as mediators of the associations between minority stressors (internalized homonegativity and rejection sensitivity) and internalizing symptoms in a sample of gay men using a weekly diary design. First, we hypothesized that internalized homonegativity and rejection sensitivity would be negatively associated with active coping and positively associated with disengaged coping and internalizing symptoms. Second, we hypothesized that active coping would be negatively associated with internalizing symptoms and disengaged coping would be positively associated with internalizing symptoms. Third, we hypothesized that active and disengaged coping would mediate the associations between minority stressors and internalizing symptoms.

We focused on gay men rather than all sexual minority individuals for two reasons. First, it is particularly important to understand how gay men cope with discrimination, because they report more sexual orientation-related discrimination than other sexual minority groups (for a meta-analysis, see Katz-Wise & Hyde, 2012). Second, other sexual minority groups experience unique sexual orientation-related stressors (for sexual minority women, see Dyar, Feinstein, Eaton, & London, 2016; for bisexual individuals, see Brewster & Moradi, 2010; Feinstein & Dyar, 2017). Focusing on a single group can lead to a better understanding of that group’s experiences without assuming that their experiences are similar to those of other groups. It will be important for future research to examine whether findings from the current study extend to other sexual minority groups.

Method

Participants

Participants included 147 gay men who participate in an eight-week study focused on gay men’s life experiences. Eligibility criteria included: being at least 18 years old, identifying as a gay man, living in a borough of New York City, having Internet access, and being able to read and write in English. Although we did not require participants to identify as cisgender men to participate (i.e., transgender gay men were eligible), all of the participants identified as cisgender gay men. Retention was excellent with 89.1% of participants completing all 8 weeks. The mean age of the sample was 37.27 years (SD = 11.42; range = 18–65). Approximately two-thirds of the sample identified as White (68.5%) with the rest identifying as Black (11.0%), Asian (9.6%), Latino (5.5%), or multi-racial (3.4%). Income ranged from less than $10,000 to more than $150,000 (mode = $40,000-$49,999). All participants completed high school; 47.6% completed a four-year degree, 21.8% had a master’s degree, and 5.4% had a doctoral degree.

Procedures

Data collection occurred between November 2013 and May 2014. Participants were recruited in-person and online. In-person recruitment was conducted at gay men’s social groups and gay bars in New York City, whereas online recruitment was conducted through websites (e.g., groups for gay men in New York City on Facebook) and listservs. Individuals who were interested in the study and met eligibility criteria were emailed a link to a website to complete baseline questionnaires on the morning of the agreed upon start date. For the next seven consecutive weeks participants were emailed a link to a website to complete weekly questionnaires on the mornings of the agreed upon dates. Participants were required to complete each week’s questionnaires on the day they were instructed to do so. Participants were compensated up to $50 for completing questionnaires, which was pro-rated based on level of completion. To increase compliance, participants who completed questionnaires for at least six weeks were entered into a raffle to win an additional $500.

Measures

Demographics (age, race/ethnicity, sexual orientation, gender, education, and income) were assessed at baseline (week 1) and all other variables (discrimination, internalized homonegativity, rejection sensitivity, internalizing symptoms, and coping) were assessed on a weekly basis for the next seven consecutive weeks (weeks 2–8). Coping was only assessed on weeks when participants endorsed discrimination. The time frame referenced in the other weekly measures depended on whether or not participants endorsed discrimination during a given week. Participants who endorsed discrimination were instructed to base their responses on the time “since the event you described,” while participants who did not endorse discrimination were instructed to base their responses on the time “over the course of the past week.”

Discrimination.

Participants were provided with a list of eight stressful situations related to one’s sexual orientation and they were instructed to indicate which (if any) they had experienced over the past week. The situations were taken from the Heterosexist Harassment, Rejection, and Discrimination Scale (Szymanski, 2009). Examples items include: “Were you treated unfairly by anyone because of your sexual orientation?” and “Were you verbally insulted because of your sexual orientation or called an anti-gay name like faggot, sissy, or other names?” If they endorsed more than one discrimination experience over the past week, then they were asked to focus on the one that had the most negative impact on them when answering other questions. If they did not endorse any discrimination experiences over the past week, then they were not asked the subsequent questions about coping.

Internalized homonegativity (IH).

The three-item IH subscale from the Lesbian, Gay, and Bisexual Identity Scale (LGBIS; Mohr & Kendra, 2011) was used to assess negative thoughts and feelings about one’s sexual orientation. It was adapted for weekly administration by asking participants to rate the extent to which they agreed or disagreed with each statement since the discrimination event they described (if they endorsed discrimination) or during the past week (if they did not endorse discrimination). Items included: (1) If it were possible, I would choose to be straight; (2) I wish I were heterosexual; and (3) I believe it is unfair that I am attracted to people of the same-sex. Each item was rated on a 1–6 scale (1 = disagree strongly, 6 = agree strongly) and total scores were computed by averaging responses across items. Cronbach’s alpha ranged from .88 to .92 across weeks.

Rejection sensitivity (RS).

Given that existing measures of RS assess the construct as a trait rather than a fluctuating state, a measure was developed for the current study. The measure was modeled after the Gay-Related Rejection Sensitivity Scale (Pachankis et al., 2008) and a daily assessment of appearance-related RS (Park & Pinkus, 2009). Respondents were asked two questions: (1) “How worried or anxious were you about being rejected because of your sexual orientation since the event you described [or over the course of the past week for participants who did not endorse discrimination]? (2) “How likely was it that you would be rejected because of your sexual orientation since the event you described [or over the course of the past week for participants who did not endorse discrimination]?” Items were rated on a 1–5 scale (1 = not at all, 5 = extremely) and total scores were computed by multiplying responses to the items. Cronbach’s alpha ranged from .91 to .95 across weeks.

Internalizing symptoms.

Internalizing symptoms were assessed with the following question, which was asked twice (once referring to depression and once referring to anxiety): “How depressed [anxious] did you feel since the event you described [or over the course of the past week for participants who did not endorse discrimination] Items were rated on a 1–5 scale (1 = very slightly or not at all, 5 = extremely) and total scores were computed by summing responses. Cronbach’s alpha ranged from .75 to .88 across weeks.

Coping.

The Brief Cope (Carver, 1997) was used to assess coping and it was adapted to refer to how participants coped with the discrimination they experienced each week. As noted, coping was only assessed on weeks when participants endorsed discrimination. The Brief Cope consists of 28 items that assess 14 coping strategies. Respondents were asked to rate the extent to which they used each of the 28 items in response to the discrimination they experienced each week on a 1–4 scale (1 = I didn’t do this at all, 4 = I did this a lot). Subscale scores for the 14 coping strategies (each represented by 2 items) were calculated by averaging responses for the respective items. Given that coping was only assessed on weeks when discrimination was reported, we were not sufficiently powered to test the reliability of each subscale at each study time point. Estimates of Cronbach’s alpha are biased in samples with 30 or fewer participants (Yurdugul, 2008) and only 7 to 33 participants completed the Brief Cope on any given week. Therefore, reliabilities were tested across all study time points. To address non-independence of observations (i.e., some participants had Brief Cope data at more than one time point), reliabilities were tested using each participant’s first set of data for this measure, resulting in 74 independent observations. Cronbach’s alphas ranged from .74 to .96 for most subscales with two exceptions (.62 for acceptance and .57 for venting).

Data Analyses

Multilevel modeling (MLM) with robust maximum likelihood estimation was conducted using Mplus Version 7. MLM was used because data were nested within person (i.e., each person had repeated observations for each variable). Additionally, MLM can estimate between-person effects (i.e., differences between individuals) and within-person effects (i.e., change over time for a given individual). A total of 52 out of 1,127 weekly assessments (4.6%) were not completed. Given that coping was only assessed on weeks when participants endorsed discrimination, analyses that include coping only use data from those weeks (i.e., 140 weekly assessments from 76 participants; 12.4% of all weekly assessments from 51.7% of the sample). If a participant completed some, but not all, measures during a given week, then their data were included in analyses and the missing data (.2%) was handled using full information maximum likelihood (FIML).

Prior to testing our hypotheses, we conducted a multi-level confirmatory factor analysis (CFA) to test the factor structure of the Brief Cope. The developer of the measure recommends using one’s own data to determine the composition of higher-order factors, because different samples can exhibit different patterns of relations (Carver, 2007). We used the comparative fit index (CFI), the Tucker-Lewis index (TLI), and the root-mean-square error (RMSEA) to assess model fit. CFI and TLI greater than .90 and RMSEA less than .06 indicate acceptable fit (Bentler, 1992; Bentler & Bonett, 1980; Browne & Cudeck, 1993; Kline, 2005). We also used chi-square tests of model fit, but they can be overpowered in moderately sized samples and reject models that fit well. To compare alternative models, we used change in Bayesian Information Criterion (BIC) values. A difference of 10 points between BIC values is associated with 150 to 1 odds that the model with the lower BIC value is superior (p < .001), a difference of 6 points is associated with 20 to 1 odds that the model with the lower BIC value is superior (p value of < .05), and a difference of less than 2 points indicates little difference between the models (Raftery, 1995).

Next, intraclass correlations (ICCs) were computed for each variable using intercept only models (i.e., no predictors included). ICCs represent the amount of total variance accounted for by between-person effects. Then, we separated the between- and within-person components of each predictor variable by using person-mean centered variables at the within-person level and grand-mean centered person-means at the between-person level (see Enders & Tofighi, 2007). We used MLM to regress each outcome variable on the between- and within-person components of each predictor variable. For each analysis, we compared fixed- versus random-effects models. In the fixed-effects models, the between- and within-person associations were modeled as fixed effects. In the random-effects models, the between-person association was modeled as a fixed effect and the within-person association was modeled as a random effect. BIC values were used to determine whether the fixed- or random-effects models were superior and results from the better fitting models are reported. Fixed-effects models were superior in most cases and, as such, results are reported from fixed-effects models unless otherwise noted. Lagged associations (i.e., outcome variables at time n regressed on predictor variables at time n – 1) were not tested, because we were underpowered. Tests of lagged associations would have required participants to endorse discrimination at least two out of seven weeks and this was only the case for 29 participants.

Finally, the Monte Carlo Method for Assessing Mediation (MacKinnon, Lockwood, & Williams, 2004) was used to test the significance of the indirect effects in the mediation analyses. This method is similar to the parametric bootstrap approach (Efron & Tibshirani, 1986) and can be used to examine mediation in multilevel models (Bauer, Preacher, & Gil, 2006). Parameter estimates and standard errors are used to simulate random draws from the a and b distributions and the product of these values is computed. This was repeated 20,000 times and the resulting distribution was used to estimate 95% confidence intervals. Confidence intervals that do not include zero are significant at an alpha level of .05.

Results

Descriptives

Half of participants (51.7%; n = 76) endorsed discrimination during at least one week over the course of the seven weeks, including 32.0% (n = 47) during one week, 8.8% (n = 13) during two weeks, 6.1% (n = 9) during three weeks, 2.7% (n = 4) during four weeks, 0.7% (n = 1) during five weeks, and 1.4% (n = 2) during six weeks). Participants reported relatively low mean levels of minority stressors and internalizing symptoms (see Table 1). The ICCs were .53 for RS and .51 for internalizing symptoms, indicating similar between- and within-person variability. In contrast, the ICCs were .81 IH, .78 for active coping, and .68 for disengaged coping, indicating that they were more “trait-like” (i.e., most of the variance was due to individual differences rather than weekly fluctuation). Bivariate correlations are reported in Table 2.

Table 1.

Estimates of fixed effects in intercept only models.

Variable	Estimated Mean (SE)	Estimated Residual Variance (SE)	Estimated Variance of the Random Intercept (SE)
Internalized homonegativity	1.58 (.08)	.22 (.05)	.94 (.21)
Rejection sensitivity	3.00 (.28)	8.98 (1.66)	9.86 (3.74)
Internalizing symptoms	1.85 (.06)	.42 (.04)	.43 (.08)

Table 2.

Correlations among main variables of interest.

Variable	Internalized homonegativity	Rejection sensitivity	Internalizing symptoms	Active coping	Disengaged coping
Internalized homonegativity	-	.43**	.14	.15	.35**
Rejection sensitivity	.23**	-	.41**	.11	.34**
Internalizing symptoms	.13*	.19**	-	.14	.31**
Active coping	.09	.28**	.12	-	.47**
Disengaged coping	.29**	.40**	.30**	.86**	-

Note. Within-person correlations are reported below the diagonal; between-person correlations are reported above the diagonal;

^*p < .05;

^**p < .01

Confirmatory factor analysis of the Brief Cope

Based on previous research (Lehavot, 2012; Meyer, 2001), we began with a two-factor model (Active Coping and Disengaged Coping). In our hypothesized model, eight subscales represented Active Coping (active coping, planning, positive reframing, acceptance, humor, religion, emotional support, and instrumental support), six subscales represented Disengaged Coping (self-distraction, denial, venting, substance use, behavioral disengagement, and self-blame), and we allowed the two factors to correlate. Additionally, we hypothesized that the two-factor structure would be present at the within- and between-person levels. Although Lehavot did not conduct a factor analysis, she demonstrated good internal consistencies for these factors (which she referred to as adaptive and maladaptive coping, respectively; α = .74−.81). Meyer (2001) removed two subscales (substance use and self-distraction) due to poor fit and then demonstrated good internal consistency for adaptive coping (α = .81), but weaker internal consistency for maladaptive coping (α = .57). These studies provide preliminary support for our hypothesized model, although they suggest that our Disengaged Coping factor may require modifications. When we compared our hypothesized two-factor model to a single-factor model, change in BIC values indicated that the two-factor model (BIC = 4589.88) was preferred over a single-factor model (BIC = 4638.93; ΔBIC = 49.05). Although the RMSEA indicated good model fit, the CFI and TLI did not (χ²[152] = 284.75, RMSEA = .04, CFI = .67, TLI = .61).

To address this, we made several changes to the factors. These changes were guided by factor loadings, modification indices, and available theory and research. We removed subscales that did not load on the factors well, including: acceptance and humor at the within- and between-person levels (standardized factor loadings = −.04−.21, ps > .17), religion at the within-person level (standardized factor loading = .13, p = .46), and substance use at the within-person level (standardized factor loading = .29, p = .32). We also allowed the following residuals to correlate: planning and active coping at the within-person level (r = .50, 95% CI = .29, .71, SE = .11, p < .001) and the between-person level (r = .78, 95% CI = .64, .91, SE = .07, p < .001), and emotional support and reframing at the within-person level (r = .42, 95% CI = .17, .67, SE = .13, p = .001). The final model fit the data well (χ²[84] = 108.60, RMSEA = .02, CFI = .93, TLI = .91). Standardized factor loadings for the final model are presented in Table 3. These latent factors were used to represent Active Coping and Disengaged Coping in all subsequent analyses.

Table 3.

Results of the Confirmatory Factor Analysis of the Brief Cope

Factor	Indicator	Estimate	95% CI	SE	p
Within-person level
Active coping	Active	.47	.08, .87	.20	.018
	Plan	.67	.45, .89	.11	< .001
	Reframe	.42	.06, .78	.18	.023
	Emotional Support	.62	.35, .89	.14	< .001
	Instrumental Support	.69	.51, .86	.09	< .001
Disengaged coping	Distract	.59	.22, .96	.19	.002
	Denial	.30	−.008, .61	.16	.057
	Vent	.65	.46, .84	.10	< .001
	Disengage	.52	.18, .85	.17	.002
	Self-Blame	.54	.14, .93	.20	.007
Between-person level
Active coping	Active	.58	.36, .79	.11	< .001
	Plan	.54	.31, .77	.12	< .001
	Reframe	.53	.33, .74	.11	< .001
	Religion	.53	.28, .77	.12	< .001
	Emotional Support	.76	.59, .93	.08	< .001
	Instrumental Support	.94	.87, 1.01	.03	< .001
Disengaged coping	Distract	.41	.23, .60	.09	< .001
	Denial	.58	.32, .84	.13	< .001
	Vent	.37	.07, .67	.15	.014
	Substance Use	.74	.56, .92	.09	< .001
	Disengage	.65	.43, .87	.11	< .001
	Self-Blame	.71	.56, .87	.08	< .001

Note. The estimate of the correlation between the Active Coping and Disengaged Coping factors at the within-person level was .86 (95% CI = .29, .71; SE = .14, p < .001) and at the between-person level was .47 (95% CI = .18, .76; SE = .15; p = .001).

Associations between minority stressors, coping strategies, and internalizing symptoms

First, we tested the associations between minority stressors and coping strategies in four models (see Table 4). Given that coping was only assessed on weeks when participants endorsed discrimination, analyses that include coping only use data from those weeks (i.e., a total of 140 weekly surveys from 76 participants). For most of these analyses, the fixed-effects models were superior and, as such, results are reported from fixed-effects models. The one exception was that the random-effects model testing the association between RS and disengaged coping was superior, so results are reported from the random-effects model for this analysis. There was a significant within-person fixed effect of RS on active coping, indicating that active coping was utilized more on weeks when gay men reported higher RS than their average levels. None of the other effects for active coping were significant. There were significant between- and within-person fixed effects of IH and RS on disengaged coping. The between-person effects indicate that gay men who reported higher tendencies to internalize stigma and anxiously expect rejection also reported greater use of disengaged coping. The within-person fixed effects indicate that disengaged coping was utilized more on weeks when gay men reported higher IH and RS than their average levels. Although the random-effects model was superior for the association between RS and disengaged coping, the within-person association between RS and disengaged coping did not vary across participants (i.e., the random effect was not significant). When both minority stressors were included in the same model, the between-person effect of IH on disengaged coping became non-significant, but all other effects remained the same (see Table 5).

Table 4.

Associations between minority stressors and coping strategies in separate models.

Predictor	Estimate	95% CI	SE	P
IH predicting Active Coping
Between-person variability in IH	.15	−.06, .36	.11	.170
Within-person variability in IH	.23	−.28, .75	.26	.376
RS predicting Active Coping
Between-person variability in RS	.03	−.02, .09	.03	.221
Within-person variability in RS	.10	.03, .18	.04	.005
IH predicting Disengaged Coping
Between-person variability in IH	.38	.08, .69	.16	.014
Within-person variability in IH	.71	.28, 1.13	.22	.001
RS predicting Disengaged Coping
Between-person variability in RS	.11	.02, .19	.04	.012
Within-person variability in RS (fixed effect)	.22	.09, .36	.07	.001
Within-person variability in RS (random effect)	.06	−.02, .13	.04	.155

Notes. IH = internalized homonegativity; RS = rejection sensitivity; for the association between RS and Disengaged Coping, the random-effect model was preferred and, as such, both the fixed- and random-effect are reported; for all other models, the fixed-effect models were preferred.

Table 5.

Associations between minority stressors and active coping (in the same model) and disengaged coping (in the same model).

Predictor	Estimate	95% CI	SE	P
IH and RS predicting Active Coping
Between-person variability in IH	.11	−.11, .33	.11	.344
Within-person variability in IH	.10	−.48, .68	.30	.733
Between-person variability in RS	.02	−.03, .08	.03	.442
Within-person variability in RS	.10	.02, .17	.04	.007
IH and RS predicting Disengaged Coping
Between-person variability in IH	.24	−.07, .56	.16	.129
Within-person variability in IH	.58	.06, 1.10	.26	.028
Between-person variability in RS	.08	.001, .16	.04	.047
Within-person variability in RS	.14	.07, .20	.03	< .001

Note. IH = internalized homonegativity; RS = rejection sensitivity; CI = confidence interval; SE = standard error; N (participants) = 76; N (events) = 140; for all models, the fixed-effect models were preferred.

Second, we tested the associations between minority stressors and internalizing symptoms in four models (see Table 6). For these analyses, random-effects models were superior and, as such, results are reported from random-effects models. The between-person effect of IH on internalizing symptoms was not significant, but the within-person fixed effect of IH on internalizing symptoms and the associated random effect (i.e., the variance in the within-person effect across individuals) were significant. This indicates that higher internalizing symptoms were reported on weeks when gay men reported higher IH than their average levels, but this association varied across participants. Additionally, there was a significant between-person effect of RS on internalizing symptoms, indicating that gay men who reported higher tendencies to anxiously expect rejection also reported higher internalizing symptoms. The within-person fixed of RS on internalizing symptoms and the associated random effect were also significant. This indicates that higher internalizing symptoms were reported on weeks when gay men reported higher RS than their average levels, but this association varied across participants. When both minority stressors were included in the same model, the within-person fixed effect of IH on internalizing symptoms became marginally significant (p = .07) and the associated random effect of IH on internalizing symptoms became non-significant, but all other effects remained the same (see Table 7).

Table 6.

Minority stressors and coping strategies predicting internalizing symptoms in separate models.

Predictor	Estimate	95% CI	SE	p
IH predicting Internalizing Symptoms
Between-person variability in IH	.10	−.04, .24	.07	.148
Within-person variability in IH (fixed effect)	.16	.02, .31	.07	.026
Within-person variability in IH (random effect)	.11	.03, .18	.04	.008
RS predicting Internalizing Symptoms
Between-person variability in RS	.09	.04, .13	.02	< .001
Within-person variability in RS (fixed effect)	.03	.007, .05	.01	.010
Within-person variability in RS (random effect)	.003	.001, .005	.001	.001
Active Coping predicting Internalizing Symptoms
Between-person variability in Active Coping	.10	−.07, .26	.08	.244
Within-person variability in Active Coping (fixed effect)	< .001	−.11, .11	.06	.997
Within-person variability in Active Coping (random effect)	.10	.06, .13	.02	< .001
Disengaged Coping predicting Internalizing Symptoms
Between-person variability in Disengaged Coping	.22	.06, .38	.08	.008
Within-person variability in Disengaged Coping (fixed effect)	.44	.36, .51	.03	< .001
Within-person variability in Disengaged Coping (random effect)	.10	.07, .12	.01	< .001

Notes. IH = internalized homonegativity; RS = rejection sensitivity; for all models, the random-effects models were preferred and, as such, both the fixed- and random-effect are reported.

Table 7.

Associations between minority stressors and internalizing symptoms (in the same model) and between coping strategies and internalizing symptoms (in the same model).

Predictor	Estimate	95% CI	SE	p
IH and RS predicting Internalizing Symptoms
Between-person variability in IH	−.03	−.15, .09	.06	.641
Within-person variability in IH	.12	−.01, .25	.07	.071
Within-person variability in IH (random effect)	.07	−.04, .11	.06	.192
Between-person variability in RS	.09	.04, .14	.02	< .001
Within-person variability in RS (fixed effect)	.02	.004, .05	.01	.021
Within-person variability in RS (random effect)	.003	.001, .005	.001	.003
Active and Disengaged Coping predicting Internalizing Symptoms
Between-person variability in Active Coping	−.001	−.21, .21	.11	.992
Within-person variability in Active Coping (fixed effect)	.03	.01, .05	.01	.003
Within-person variability in Active Coping (random effect)	.003	.002, .004	.001	< .001
Between-person variability in Disengaged Coping	.21	.02, .41	.10	.029
Within-person variability in Disengaged Coping (fixed effect)	.04	−.14, .22	.09	.669
Within-person variability in Disengaged Coping (random effect)	.37	.29, .45	.04	< .001

Note. IH = internalized homonegativity; RS = rejection sensitivity; CI = confidence interval; SE = standard error; N (participants) = 76; N (events) = 139–140; for all models, the random-effects models were preferred and, as such, both the fixed- and random-effect are reported.

Third, we tested the associations between coping strategies and internalizing symptoms in four models (see Table 8). For these analyses, random-effects models were superior and, as such, results are reported from random-effects models. The between-person and within-person fixed effects of active coping on internalizing symptoms were not significant, but the associated random effect of active coping on internalizing symptoms was significant, indicating that the within-person association between active coping and internalizing symptoms varied across participants. There was a significant between-person effect of disengaged coping on internalizing symptoms, indicating that gay men who reported higher tendencies to utilize both active and disengaged coping also reported higher internalizing symptoms. The within-person fixed effect of disengaged coping on internalizing symptoms and the associated random effect were also significant, indicating that higher internalizing symptoms were reported on weeks when gay men reported utilizing disengaged coping more than their average levels, but this association varied across participants. When active and disengaged coping were included in the same model, the within-person fixed effect of disengaged coping on internalizing symptoms became non-significant and the within-person fixed effect of active coping on internalizing symptoms became significant (see Table 7), which may be due to the high correlation between active and disengaged coping (i.e., multicollinearity).

Table 8.

Minority stressors and coping strategies predicting internalizing symptoms in the same model.

Predictor	Estimate	95% CI	SE	p
Between-person variability in IH	−.04	−.15, .08	.06	.503
Within-person variability in IH	.11	−.04, .26	.08	.166
Between-person variability in RS	.08	.03, .13	.03	.001
Within-person variability in RS	.03	.01, .06	.01	.014
Between-person variability in Active Coping	.01	−.17, .20	.09	.900
Within-person variability in Active Coping	−.41	−1.67, .85	.64	.523
Between-person variability in Disengaged Coping	.12	−.09, .32	.10	.261
Within-person variability in Disengaged Coping	.51	−.72, 1.74	.63	.416

Note. IH = internalized homonegativity; RS = rejection sensitivity; CI = confidence interval; SE = standard error; N (participants) = 76; N (events) = 139; for this model, the fixed-effect model was preferred.

Finally, we included minority stressors and coping strategies in the same fixed-effects model to test their relative associations with internalizing symptoms (see Table 8). The only significant associations were the between- and within-person effects of RS on internalizing symptoms. The between-person effect indicates that gay men who reported higher tendencies to anxiously expect rejection also report higher internalizing symptoms. The within-person effect indicates that higher internalizing symptoms were reported on weeks when gay men reported higher RS than their average levels. Together, findings indicate that RS is a stronger predictor of internalizing symptoms than IH and coping strategies.

Mediation at the between-person level

Fixed effects are reported for all mediation analyses, because the fixed-effects models were superior. Active coping was not tested as a mediator, because the between-person effects of IH and RS on active coping were not significant. We examined disengaged coping as a mediator of the associations between each minority stressor and internalizing symptoms at the between-person level in separate models (see Table 9). There was a significant between-person effect of IH on disengaged coping and a significant between-person effect of disengaged coping on internalizing symptoms (controlling for IH). The indirect effect was significant, indicating that disengaged coping significantly mediated the association between IH and internalizing symptoms at the between-person level. Additionally, there was a significant between-person effect of RS on disengaged coping and a significant between-person effect of disengaged coping on internalizing symptoms (controlling for RS). However, the indirect effect was not significant, indicating that disengaged coping did not significantly mediate the association between RS and internalizing symptoms at the between-person level.

Table 9.

Disengaged coping as a mediator of the associations between minority stressors and internalizing symptoms.

Predictor	Estimate	95% CI	SE	p
Disengaged coping as a mediator of the association between IH and internalizing symptoms at the between-person level
IH predicting Disengaged Coping	.36	.07, .64	.15	.014
Disengaged Coping predicting Internalizing Symptoms (controlling for IH)	.19	.03, .34	.08	.018
Indirect effect of IH on Internalizing Symptoms via Disengaged Coping	.07	.005, .16
Disengaged coping as a mediator of the association between RS and internalizing symptoms at the between-person level
RS predicting Disengaged Coping	.11	.03, .19	.04	.009
Disengaged Coping predicting Internalizing Symptoms (controlling for RS)	.13	−.04, .29	.09	.139
Indirect effect of RS on Internalizing Symptoms via Disengaged Coping	.01	−.004, .04
Disengaged coping as a mediator of the association between IH and internalizing symptoms at the within-person level
IH predicting Disengaged Coping	.66	.25, 1.08	.21	.002
Disengaged Coping predicting Internalizing Symptoms (controlling for IH)	.14	.04, .24	.05	.004
Indirect effect of IH on Internalizing Symptoms via Disengaged Coping	.09	.02, .21
Disengaged coping as a mediator of the association between RS and internalizing symptoms at the within-person level
RS predicting Disengaged Coping	.14	.08, .21	.03	< .001
Disengaged Coping predicting Internalizing Symptoms (controlling for RS)	.11	.04, .19	.04	.
Indirect effect of RS on Internalizing Symptoms via Disengaged Coping	.02	.004, .03

Notes. IH = internalized homonegativity; RS = rejection sensitivity; for all models, the fixed-effect models were preferred.

Mediation at the within-person level

We did not test active coping as a mediator, because the within-person effect of IH on active coping was not significant and the within-person fixed effect of active coping on internalizing symptoms was not significant. We examined disengaged coping as a mediator of the associations between each minority stressor and internalizing symptoms at the within-person level in separate models (Table 9). There was a significant within-person effect of IH on disengaged coping and a significant within-person effect of disengaged coping on internalizing symptoms (controlling for IH). The indirect effect was significant, indicating that disengaged coping significantly mediated the association between IH and internalizing symptoms at the within-person level. Similarly, there was a significant within-person effect of RS on disengaged coping and a significant within-person effect of disengaged coping on internalizing symptoms (controlling for RS). The indirect effect was significant, indicating that disengaged coping significantly mediated the association between RS and internalizing symptoms at the within-person level.

Discussion

The current study advanced the literature on gay men’s mental health by examining coping strategies in response to discrimination as mediators of the associations between minority stressors and mental health using a weekly diary. Half of participants endorsed discrimination during at least one week over the course of seven weeks. Among those who did, most only endorsed it during one week. Although it is encouraging that a sizable proportion of the sample did not endorse any discrimination, the fact that a near equal proportion did indicates that gay men continue to be discriminated against despite increasing societal acceptance. Further, given that the current study only spanned seven weeks, it is likely that more gay men experience discrimination over longer periods of time. This supports the ongoing need for longitudinal research on how gay men cope with discrimination.

Findings demonstrated that there was weekly fluctuation in most of the constructs. For RS and internalizing symptoms, 47–49% of the variability was accounted for by an individual’s weekly fluctuation, highlighting the importance of examining these constructs in real-time (e.g., daily, weekly). These findings are consistent with a recent daily diary study of sexual minority individuals, which found that 38% of the variability in expectations of rejection was due to within-person daily fluctuation (Mohr & Sarno, 2016). While RS has previously been conceptualized as a trait (e.g., Feinstein et al., 2012; Pachankis et al., 2008), these findings indicate that it may be better conceptualized as a state that varies in response to the environment.

Although a recent daily diary study found that 41% of the variability in internalized stigma was accounted for by daily fluctuation (Mohr & Sarno, 2016), we found that only 19% of the variability in IH was accounted for by weekly fluctuation. It is possible that daily fluctuations in IH are masked when examined at the weekly level. Alternatively, given that our sample consisted of gay men and their sample consisted of lesbian, gay, and bisexual individuals, it is possible that the stability of IH varies as a function of gender and/or sexual identity. We also found that 22–32% of the variability in gay men’s use of active and disengaged coping strategies varied from week to week. Although this indicates that they do not respond to all instances of discrimination in the same ways, there does tend to be stability in their responses. It will be important for future research to examine situational influences on anxious expectations of rejection and use of specific coping strategies. These findings help to refine conceptualizations of minority stressors, suggesting that negative thoughts and feelings about one’s sexual orientation may be more stable than anxious expectations of rejection. While more stable, weekly fluctuation in IH still influenced coping and internalizing symptoms, highlighting its importance.

Predictors of Coping Strategies

Consistent with previous research (Nicholson & Long, 1990; Szymanski & Carr, 2008; Szymanski & Owens, 2008), higher IH and RS were associated with higher disengaged coping at the between- and within-person levels. The between-person effect of IH on disengaged coping became non-significant when RS was controlled for, suggesting that RS is a unique predictor of disengaged coping. The within-person effects extend previous research by demonstrating that disengaged coping was utilized more on weeks when individuals reported higher IH and RS than their average levels. Thus, even if an individual reports low average levels of IH and RS, their levels on a particular week can still influence their use of disengaged coping in response to discrimination.

The only significant predictor of active coping was the within-person effect of RS and it was opposite the predicted direction. Active coping was utilized more on weeks when individuals reported higher RS than their average levels. Given that active and disengaged coping were positively associated, this may reflect a tendency to utilize multiple coping strategies in response to anxious expectations of rejection. This is consistent with qualitative findings that gay and bisexual men use multiple strategies to cope with heterosexism in a single situation (McDavitt et al., 2008). The non-significant association between IH and active coping is consistent with Szymanski and Owens (2008), but inconsistent with Nicholson and Long (1990). Nicholson and Long used a measure of active coping that emphasized social support, so it is possible that IH only influences some types of active coping.

Predictors of Internalizing Symptoms

Results partially supported the hypothesis that IH would be associated with internalizing symptoms. Although the between-person effect of IH on internalizing symptoms was not significant, the within-person effect was significant, indicating that higher internalizing symptoms were reported on weeks when individuals reported higher IH than their average levels. This is consistent with a recent daily diary study, which found that sexual minorities reported higher negative affect on days when they reported higher internalized stigma than usual (Mohr & Sarno, 2016). Still, in our study, this effect became marginally significant when RS was controlled for, suggesting that RS has a stronger influence on internalizing symptoms.

The random effect of IH on internalizing symptoms was also significant, indicating that the association varied across participants. However, this effect became non-significant when RS was controlled for, so this finding should be interpreted with caution. Still, we offer a possible explanation. Although the average association between IH and internalizing symptoms was positive, it ranged from negative to positive for individual participants. This suggests that IH may lead to increased internalizing symptoms for some people, but decreased internalizing symptoms for others. It is counterintuitive to consider the possibility that IH may be associated with decreased internalizing symptoms for some people. However, some people who harbor negative thoughts and feelings about their sexual orientation may push themselves to excel in other areas of their lives, contributing to better mental health. Pachankis and Hatzenbuehler (2013) found that sexual minority men derived their self-worth from achievement-oriented domains (e.g., academics, competition) more than heterosexual men, especially those who concealed their sexual orientation for longer. Therefore, if IH has positive consequences for some people, it may be a result of the ways in which they choose to derive self-esteem.

Consistent with minority stress theory (Meyer, 2003) and previous research (Feinstein et al., 2012), results supported the hypothesis that RS would be associated with internalizing symptoms at the between- and within-person levels. The within-person fixed effect extends previous cross-sectional research by demonstrating that, regardless of how much an individual tends to anxiously expect rejection, they experience more distress on weeks when they are more anxious about and expectant of rejection. This is also consistent with a recent daily diary study, which found that sexual minorities reported more negative affect on days when they reported higher expectations of rejection than usual (Mohr & Sarno, 2016). The random effect of RS on internalizing symptoms was also significant, indicating that the within-person association varied across participants (the average was positive, but it ranged from negative to positive for individual participants). This suggests that RS may lead to increased internalizing symptoms for some people, but decreased internalizing symptoms for others. It is possible that some people who anxiously expect rejection seek supportive people and environments, consequently reducing risk for symptoms.

Consistent with studies that have demonstrated that higher avoidant coping was associated with worse mental health among sexual minorities (Lehavot, 2012; Nicholson & Long, 1990; Szymanski & Henrichs-Beck, 2014), results also supported the association between disengaged coping and internalizing symptoms. Thus, regardless of an individual’s general tendency to use disengaged coping strategies, they experience more distress on weeks when they respond to discrimination in this way. In contrast, active coping was not significantly associated with internalizing symptoms. Although previous studies have found that active coping was associated with better mental health among sexual minorities (DeMarco, Ostrow, & DiFranceisco, 1999; Lehavot, 2012; Szymanski & Henrichs-Beck, 2014; Zea, Reisen, & Poppen, 1999), Lehavot (2012) found that maladaptive coping had a stronger influence on depression than adaptive coping. In our study, when active and disengaged coping were included in the same model predicting internalizing symptoms, the within-person fixed effect of disengaged coping became non-significant and the within-person fixed effect of active coping became significant. It is likely that this is due to the high correlation between active and disengaged coping (i.e., multicollinearity), so findings from the model with active and disengaged coping should be interpreted with caution.

When all of the minority stress and coping variables were included in the same model predicting internalizing symptoms, the only significant predictor was RS (at the between- and within-person levels). This suggests that RS is a stronger predictor of internalizing symptoms than IH and coping strategies. While we found some evidence for IH and disengaged coping having influences on internalizing symptoms as well, it is possible that their influences are driven by shared variance with RS.

Mediation Analyses

We tested disengaged coping as a mediator of the between- and within-person associations between minority stressors and internalizing symptoms. Three of the four mediation analyses were significant, supporting the mediating role of disengaged coping. This suggests that IH and RS influence internalizing symptoms, in part, by contributing to disengaged coping in response to discrimination. This is consistent with Hatzenbuehler (2009), who proposed that coping may mediate associations between discrimination and mental health problems. These findings are also consistent with previous research that has demonstrated that maladaptive coping mediated the associations between discrimination and distress (Hatzenbuehler, Nolen-Hoeksema, & Dovidio, 2009), implicit anti-gay attitudes and distress (Hatzenbuehler, Dovidio, Nolen-Hoeksema, & Phills, 2009), and internalized stigma and distress (Kaysen et al., 2014; Szymanski & Henrichs-Beck, 2014; Szymanski & Owens, 2008) among sexual minorities. These findings suggest that when gay men feel badly about their sexual orientation or anxiously expect to be rejected, they utilized disengaged coping strategies in response to experiences of discrimination. In turn, using these coping strategies led to greater psychological distress.

Clinical Implications

The current findings underscore the roles of IH and RS in how gay men cope with discrimination as well as their mental health. Whether or not someone experiences discrimination is largely out of one’s control, but negative thoughts and feelings about one’s sexual orientation and anxious expectations of rejection can be reduced through clinical interventions, such as cognitive-behavioral therapy. Cognitive techniques can be used to help clients develop and utilize more realistic ways of thinking about the self and others, whereas behavioral techniques (e.g., exposure) can be used to help clients tolerate their anxiety in situations that trigger expectations of rejection. Pachankis, Hatzenbuehler, Rendina, Safren, and Parsons (2015) recently developed a transdiagnostic treatment designed to reduce depression, anxiety, risky sexual behavior, and substance use among gay and bisexual men by improving their ability to cope with minority stress. A randomized controlled trial demonstrated a significant reduction in depressive symptoms and a trend toward a reduction in anxiety symptoms in the treatment group relative to a waitlist (Pachankis, et al., 2015). Our findings point to potential mechanisms of change, such that reductions in minority stressors and disengaged coping have the potential to reduce mental health problems. While IH and RS are generally associated with negative outcomes, our findings also suggest that they may be associated with decreased internalizing symptoms for some gay men. It is important that clinicians assess the extent to which negative thoughts and feelings about one’s sexual orientation vary and their influence on mental health for each individual client.

Findings also emphasize the negative consequences of utilizing disengaged coping in response to discrimination. Clinicians can work with gay men to increase their awareness that disengaged coping may increase their distress and to develop alternative ways of coping with discrimination. Additionally, it is noteworthy that RS was the only significant predictor of internalizing symptom when all of the minority stress and coping variables were included in the same model. Anxious expectations of rejection may be a particularly important intervention target for gay men who are experiencing internalizing problems. As such, teaching them how to identify these expectations and their consequences, as well as helping them cope with them in more effective ways, may lower risk for depression and anxiety.

Limitations

Findings should be interpreted in light of several limitations. First, the mean levels of minority stress and internalizing symptoms were relatively low given the possible ranges, suggesting that the sample was well adjusted. Still, given that most hypotheses were supported, it is likely that findings would emerge in a more symptomatic sample. Second, the analytic sample was reduced from 147 to 76 for analyses that included coping. In these analyses, tests of within-person effects were based on the 29 participants who endorsed discrimination (and, thus, provided coping data) during two or more weeks. Despite the reduced sample size, we were still able to detect significant within-person effects of coping, highlighting the robustness of these associations. However, because a limited number of participants endorsed discrimination during two or more weeks, we did not test lagged associations. It will be important for future research to use larger samples to increase statistical power for lagged analyses. Future studies could also focus on micro-aggressions, which may lead to higher rates of endorsement. Third, given that several measures were adapted or developed for the current study, replication will be important. The measure of coping also aggregated multiple strategies into broad categories based on factor analysis and we found slightly different factor structures at the within- and between-person levels. Further, we made several post-hoc modifications to the factors based on the factor loadings and modification indices. Although our modifications were within logical reason, this limits the extent to which the same factors would be found in different samples. It will also be important for future research to examine specific coping strategies rather than aggregating multiple strategies into broad categories. Doing so has the potential to refine our understanding of the specific coping strategies that contribute to better versus worse mental health subsequent to discrimination.

Fourth, given that stress and symptoms were assessed at the same time, it is unclear if stress preceded symptoms. To address this limitation, participants were instructed to report their symptoms since the discrimination, but it is still possible that symptom reports were influenced by symptoms before the discrimination. Fifth, participants ranged in age from 18 to 65 years. Given substantial changes in societal acceptance of gay men over time, there may be age and cohort differences in gay men’s exposure to and management of minority stress. We tested this possibility, but age was not significantly associated with any variables of interest and it did not significantly moderate any associations (results not presented). Studies using different designs (e.g., longitudinal cohort studies, qualitative studies) are needed to better understand potential age and cohort differences. Finally, all participants identified as gay men, two-thirds identified as White, and three-quarters had completed a four-year college degree or higher. It will be important for future research to see if findings extend to other sexual minority groups (e.g., sexual minority women, bisexual individuals) and to represent diverse racial/ethnic and socioeconomic backgrounds.

Despite limitations, the use of a weekly diary design allowed for novel tests of within-person effects of minority stressors on coping strategies used in response to discrimination and internalizing symptoms. Given that gay men continue to experience discrimination despite a social climate that is more accepting than in the past, it is as important as ever to understand influences on mental health in this population. The current findings underscore the need for research designs that capture the fluctuating nature of minority stressors over time in order to understand the decisions that people make in the face of discrimination. Doing so has the potential to advance our understanding of why gay men are at increased risk for mental health problems and the processes that contribute to such risk.

Public Significance Statement:

This study demonstrates that gay men respond to discrimination using more disengaged coping strategies on weeks when they experience more negative thoughts/feelings about their sexual orientation and more anxious expectations of rejection. In turn, their use of disengaged coping strategies contributes to depression/anxiety. Teaching gay men how to identify these thoughts/feelings and their consequences, and helping them cope in more effective ways, has the potential to lower risk for depression/anxiety.