Daily Autonomy Support and Sexual Identity Disclosure Predicts Daily Mental and Physical Health Outcomes

Decisions to Come Out

Unlike people with a visible stigma (e.g., obesity, physical disability), individuals with a concealable stigma (e.g., a nonheterosexual identity; mental illness) regularly face decisions about whether, how much, and to whom they disclose (Pachankis, 2007). Coming out as LGB is a dynamic, ongoing process as opposed to one discrete event (e.g., Bohan, 1996; Mohr & Fassinger, 2000). For example, a diary study by Beals, Peplau, and Gable (2009) found that, on average, lesbian and gay participants explicitly considered disclosing their sexual orientation to others 3 times over a 2-week period.

Moreover, LGB individuals are selective in how much and to whom they disclose their sexual identity. For example, 51% of LGB individuals are not out to most people at work (Human Rights Campaign, 2010). Similarly, D’Augelli (2006) found that only 23% of youth were out to everyone in their life. As these findings suggest, coming out or being out is not all-or-nothing—outness, or one’s degree of openness regarding sexual orientation, often varies with different people in one’s life (Mohr & Fassinger, 2000).

Decisions about disclosing a concealable stigma can be stressful and affect well-being. Frable, Platt, and Hoey (1998) found that individuals with concealable stigmas reported lower day-to-day social confidence and self-esteem, and greater anxiety and depression than those with a visible stigma or no stigma. This may be because deciding to disclose a stigmatized identity can mean choosing between two undesirable outcomes: risking rejection, or concealing an important part of oneself.

Risks of Coming Out

Negative attitudes or intolerance toward LGB individuals can make disclosure a risky task, keeping many people “in the closet.” For example, sexual prejudice can exist in workplaces, motivating LGB individuals to keep their sexual identity concealed. Indeed, 58% of LGB individuals reported hearing derogatory comments from coworkers (Human Rights Campaign, 2010). D’Augelli (2006) found 38% of LGB youth feared being verbally attacked, and 28% feared physical assault at school. In addition, 81% reported experiencing verbal harassment, 38% reported physical threats, and 15% reported physical assaults.

Even with close others, the threat of rejection can keep individuals from disclosing their LGB identity. D’Augelli (2002) found that approximately one quarter of mothers and one third of fathers reacted to their child’s disclosure with intolerance or nonacceptance. In their sample, 39% reported losing a friend because of their sexual orientation, which also emerged as a risk factor for suicide attempts and worse mental health. D’Augelli, Hershberger, and Pilkington (1998) found that LGB youth who had come out to family experienced more verbal and physical abuse, and were more suicidal than those youth who concealed their sexual orientation. Such consequences make it clear why some choose to conceal their LGB identity.

Reasons to Come Out

Given these risks, why would LGB individuals choose to come out at all? The consensus across literatures is that although concealment may be effective at helping avoid the risks of disclosure, it can be costly to mental and physical health (see Pennebaker & Chung, 2011). Concealment of sexual orientation has been found to predict greater depression and anxiety (e.g., Schrimshaw, Siegel, Downing, & Parsons, 2013; Ullrich, Lutgendorf, & Stapleton, 2003), more smoking (Pachankis, Westmaas, & Dougherty, 2011), and more rapid HIV disease progression (Cole, Kemeny, Taylor, & Visscher, 1996; Ullrich et al., 2003). Experimental studies (Critcher & Ferguson, 2014) found that those who had to conceal their sexual orientation showed performance decrements on cognitive and physical tasks compared with those who did not conceal, illustrating that concealment taxes cognitive and self-regulatory resources (Baumeister, Vohs, & Tice, 2007; Quinn, 2006). In addition, concealment affects relationship quality and social support (Uysal, Lin, Knee, & Bush, 2012), particularly with the LGB community, which is an important source of support for LGB individuals that can help buffer psychological distress (Meyer, 2003; Pachankis, 2007). In short, concealing prevents people from reaping the intrapersonal and interpersonal benefits of freely being oneself.

These pitfalls of concealment suggest, conversely, many of the benefits of coming out—improving mental and physical well-being, freeing up cognitive and emotional resources, and bringing more closeness and connectedness into relationships. More than just avoiding the costs of concealment, coming out can facilitate self-acceptance, and ameliorate some of the distress caused by stigma (e.g., Cain, 1991; Ragins, 2004). Indeed, at the daily level, LGB individuals experienced greater well-being on days they disclosed their sexual orientation (Beals et al., 2009). Coming out enables people to more fully be themselves with others, as well as to incorporate their sexual identity with other aspects of themselves (Meyer, 2003). This is important insofar as integration of one’s experiences and identities has long been seen as important for health and wellness (e.g., Rogers, 1961; Weinstein, Deci, & Ryan, 2011).

Outcomes of Disclosure Depend on Reactions from Others

Increasingly, research has shown that a confidant’s reaction largely determines people’s experiences following a sensitive disclosure (Chaudoir & Fisher, 2010; Rodriguez & Kelly, 2006). For example, women who disclosed that they had an abortion to close but nonsupportive others (e.g., family member, friend) felt more depressed and coped with the abortion in less effective ways compared with those who told supportive others or those who did not disclose at all (Major & Gramzow, 1999). Adolescents disclosing an LGB identity to parents and caregivers who reacted with acceptance reported lower depression, fewer suicide attempts, less drug use, and less sexual risk–taking behaviors (C. Ryan, Huebner, Diaz, & Sanchez, 2009; C. Ryan, Russell, Huebner, Diaz, & Sanchez, 2010) as compared with adolescents who had parents and caregivers with rejecting responses to disclosure. Reactions from important people may have more of an impact on wellness than the act of disclosing itself. For example, D’Augelli (2002) found that parents’ reactions to their children’s LGB identity disclosure were predictive of youth’s mental health symptoms, whereas the disclosure itself had less impact, even after controlling for quality of the parent–child relationship.

Although studies have looked at antecedents of disclosure, as well as outcomes of disclosure, little empirical research has looked at these two processes together (Chaudoir & Fisher, 2010). This work aims to fill that gap. Given that the consequences of disclosure for an LGB individual largely depend on the reactions of the confidants, it seems plausible that they look for cues of threat or support to predict when disclosure will result in benefits versus harm. To understand how relationships can make people feel more or less safe to disclose, we focus on the SDT concept of autonomy support, or acceptance and support for being oneself (Lynch, La Guardia, & Ryan, 2009).

Autonomy Support and Being Oneself

According to SDT, people have basic psychological needs for autonomy, competence, and relatedness. Satisfaction of these needs promotes people’s natural propensity toward psychological growth and wellness, whereas having these needs thwarted contributes to ill-being and psychopathology (R. M. Ryan, Deci, Grolnick, & La Guardia, 2006). The need for autonomy refers to the extent that people are behaving in accord with their beliefs and feelings, and being themselves. The opposite of autonomy is feeling controlled, or a sense of pressure to act in certain ways. Acting autonomously is associated with a number of positive outcomes such as better mental health (e.g., R. M. Ryan et al., 2006), physical health (e.g., Williams, Grow, Freedman, Ryan, & Deci, 1996), and better relationships (Patrick, Knee, Canevello, & Lonsbary, 2007).

When people feel support for autonomy, they feel accepted for who they are. Autonomy support helps people to behave in ways that are consistent with their beliefs and interests and express themselves authentically (La Guardia & Ryan, 2007; Lynch et al., 2009). Indeed, autonomy support aids integration and congruence within individuals (R. M. Ryan, 1995; R. M. Ryan & Deci, 2000). For example, Weinstein and colleagues (2011) found that contextual support for autonomy helped people to integrate positive and negative memories or events into their self-concept, even the ones they judged as shameful, which had benefits to well-being.

Perceiving someone as controlling might lead an individual to conceal a stigmatized identity rather than risk rejection or negative judgments. One common form of control is conditional regard, or conveying that one is only loveable under certain conditions (Roth, Assor, Niemiec, Ryan, & Deci, 2009). LGB individuals have an identity that may be regarded as unacceptable or unlovable, potentially making it feel less safe to be oneself with conditionally regarding others.

Autonomy and Disclosure

Autonomy has been theoretically and empirically linked to disclosure as both antecedent and outcome. People are more likely to present different aspects of their personality with autonomy-supportive others (Lynch et al., 2009). Directly related to the present work, Legate, Ryan, and Weinstein (2012) found that autonomy support was a robust predictor of coming out as LGB across varied relationship contexts. Importantly, the benefits of coming out were limited to autonomy-supportive contexts, whereas benefits to wellness from disclosing were absent in controlling contexts.

Uysal and colleagues (Uysal, Lin, and Knee, 2010; Uysal et al., 2012) found that self-disclosure enhanced personal and relationship well-being because it satisfied basic psychological needs for autonomy, relatedness, and competence, whereas concealment led to worse relationship well-being because it thwarted these psychological needs. Closely related to this research, Beals and colleagues’ (2009) experience-sampling study found that gay and lesbian individuals experienced greater well-being on days when they disclosed versus concealed their sexual identity, in part because they felt supported and understood—what SDT would conceptualize as satisfying psychological needs for autonomy and relatedness. This research builds off of work by Beals and colleagues (2009) and Legate and colleagues (2012) by examining day-to-day fluctuations in disclosure and psychological well-being and for the first time explores day-to-day fluctuations in physical symptoms based on environmental support and decisions about disclosure.

The Current Research

Our aim in the present research was to understand the social conditions surrounding disclosure experiences within daily interactions, testing the link between disclosure and physical health in addition to psychological well-being, and addressing the question of why disclosing relates to wellness. First, we set out to replicate a cross-sectional finding from Legate et al. (2012) using an experience-sampling methodology—that it is not harmful to people’s well-being to vary in how out they are across their relationships. This null hypothesis was a basic premise of the research: People will naturally vary in their disclosure according to the support provided by the social environment. Building on that premise, we further tested the hypothesized path model depicted in Figure 1. Specifically, we posited a multilevel mediation model in which people will vary in how out they are across their daily interactions based on the autonomy support they perceive (Path a1), with higher levels of autonomy support in a conversation predicting higher levels of disclosure in that conversation. In turn, we predicted that higher disclosure would predict greater need satisfaction (Path b), which in turn would predict greater well-being (Path c1) and fewer physical symptoms (Path c2). That is, we expected that autonomy support would have an indirect effect on well-being and physical symptoms through disclosure (first mediator) and need satisfaction (second mediator). In addition, given the strong links between autonomy support and need satisfaction in previous studies (e.g., Adie, Duda, & Ntoumanis, 2011; R. M. Ryan & Deci, 2000), we hypothesized that interacting with autonomy-supportive people would also directly predict greater need satisfaction (Path a2), and in turn greater psychological well-being (Path c1) and fewer physical symptoms (Path c2). Although we intended to test these indirect effects at both the within- and between-person levels using multilevel modeling, we expected these indirect effects to operate primarily at the level of interactions—the within-person level.

OPEN IN VIEWER

Figure 1.

Hypothesized multilevel path model tested at both within- and between-person levels.

Method

Results

Preliminary Analyses

Descriptive statistics and intercorrelations for primary study variables are presented in Table 1. ANOVAs examined for sexual orientation differences across study variables. Results showed that across the diary period, bisexuals perceived less autonomy support, disclosed their sexual orientation less, felt less need satisfaction, and reported more physical symptoms, F(2, 68) = 3.62-7.76, ps = .001-.03, as compared with lesbians and gay men (who did not differ). Across the 2-week period, perceived autonomy support, levels of disclosure, and need satisfaction differed across relationship type, F(11, 1995) = 32.42-35.50, ps < .001. Close friends provided the most autonomy support of any targets (ps < .001), and participants were most out (ps < .001), and felt the most autonomy and relatedness in interactions with them (ps < .001). Related, people overwhelmingly interacted with close friends more than any of the other 11 targets (see Figure 2). Because of the many differences across sexual orientation and relationship types, we included these variables as controls in multilevel analyses.

OPEN IN VIEWER

Table 1.

Means, Standard Deviations, and Correlations of Study Variables at the Between-Person Level.

	M	SD	1	2	3	4	5	6	7	8	9
1. BL well-being	3.60	0.90	—
2. BL physical symptoms	1.76	2.74	−.14	—
3. Mdiary autonomy support	3.46	0.34	.23 a	.04	—
4. Mdiary disclosure	3.29	0.50	.28*	−.09	.63***	—
5. SDdiary disclosure	0.63	0.28	−.16	.13	−.38**	−.70***	—
6. Mdiary need satisfaction	3.37	0.38	.27*	−.01	.83***	.73***	−.47***	—
7. Mdiary well-being	3.57	0.37	.62***	−.11	.31*	.38**	−.09	.43***	—
8. Mdiary physical symptoms	0.91	0.89	−.47***	.21 a	−.28*	−.37**	.10	−.34**	−.66***	—
9. Total diaries completed	29.84	10.4	.17	.09	.39**	.29*	−.16	.42***	.14	−.13	—
10. Age	23.52	6.91	.11	−.09	−.05	.16	−.28*	.14	.12	.09	.10

Note. Correlations based on N = 69. BL refers to variable assessed at baseline; Mdiary refers to person-level mean aggregations across the diary period; SDdiary refers to the standard deviation across the diary period.

a

Marginally significant.

*

p < .05. **p < .01. ***p < .001.

OPEN IN VIEWER

Figure 2.

Number of interactions with each target across 2 weeks.

To examine our first hypothesis—that variability in disclosure across interactions would be unrelated to mental and physical wellness—means and SDs of disclosure were computed for each person across the 14-day period. Ordinary least squares regressions were used to regress psychological well-being and physical symptoms (aggregated over the 14-day period) onto SD scores for disclosure, controlling for mean levels of disclosure and sexual orientation (dummy coded). Surprisingly, we found that more variability in disclosure across the diary period (M = .63, SD = .28) was related to greater well-being, β = .36, t(66) = 2.32, p = .02, and fewer physical symptoms, β = −.35, t(66) = −2.25, p = .03, after controlling for average levels of disclosure (well-being: β = .61, t[66] = 3.79, p < .001; physical symptoms: β = −.53, t[66] = −3.35, p = .001). Therefore, more variability in disclosure after controlling for average levels of disclosure was related to greater wellness outcomes.

Multilevel Modeling

Mplus software (Version 7.4; Muthén & Muthén, 1998-2014) was used to test the hypothesized multilevel path model (Figure 1). In multilevel structural equation modeling (SEM), variance is divided into within-person and between-person components, which allows us to see how much variance is explained at the conversation level (Level 1) as well as at the mean level across the diary period (Level 2). This modeling approach allowed us to simultaneously test the proposed multistep mediation hypotheses at both Levels 1 and 2 (see Preacher, Zyphur, & Zhang, 2010). Multilevel models not only accommodate the nested structure of the data, but they are also better suited than ordinary least squares regression to handle missing data (Bolger & Shrout, 2007; Little & Rubin, 1987). Expectably, a source of missing data was the number of diary surveys provided—63 participants (93%) reported on one or more personal interactions each day, 43 participants (63%) reported on two or more personal interactions each day, and 11 participants (16%) reported the maximum of three personal interactions each day. As suggested by Preacher and colleagues (2010), we examined intraclass correlation coefficients to determine the variance explained at the between-person level (.25-.42), verifying that there was enough variance for models at both between- and within-person levels.

Level 1 effects

Within specific diary assessments following social interactions, we built a model directly corresponding to our conceptual model (see Figure 1). Thus, autonomy support was allowed to predict both disclosure and need satisfaction, disclosure predicted need satisfaction, and need satisfaction predicted both well-being and physical symptoms. To control for the stability of the two outcome variables (well-being and physical symptoms) across conversations, we included levels of those variables from the previous assessment as predictors at Level 1 (modeling autoregressive paths for the outcomes). As the time intervals between assessments varied, we also included a variable coding time (in hours) between assessments as a predictor and allowed that variable to moderate the autoregressive paths (expecting those paths to be weaker for longer intervals between assessments; for example, Reis, Sheldon, Gable, Roscoe, & Ryan, 2000). In other words, an individual’s well-being score from their second diary occurrence was predicted by their well-being score on their first diary occurrence, the time between the first and second assessments, and the interaction between the first well-being assessment and time.

To control for potential differences across relationship types, we included dummy variables at Level 1 representing three relationship types—close friends, family (that combined mom, dad, sibling, and other relative), and stranger (that combined strangers and acquaintances) as these three dummy-coded relationships account for 55% of all conversations. We specifically selected family and close friends to see the effects of close relationships, and to examine how the model held across even the most distant relationship type, we selected strangers and acquaintances. In each of these variables, the interactions with the targeted relationship type were coded 1, and all others were coded as 0. Thus, by including those three dummy variables at Level 1, highly close and highly distant relationships were compared with all other relationship types (which can be considered to have intermediate levels of closeness, for example, coworkers, casual friends, roommates). We modeled these as main effects, and also tested for potential moderations with autonomy support when predicting disclosure and need satisfaction. This examined whether the effect of autonomy support on disclosure and need satisfaction was stronger for certain relationships—directly testing that moderation through the included interaction terms.

Level 2 effects

Following recommendations for testing multilevel mediation (Preacher et al., 2010) in addition to testing our conceptual meditational model at Level 1, we also tested the same meditational associations at Level 2 to examine whether average levels of each of the constructs in the model (Figure 1) would show a comparable pattern of mediation. As past research shows bisexuals are at higher risk of psychological distress (e.g., Semlyen, King, Varney, Hagger, & Johnson, 2016) and our own data revealed that bisexuals differed from lesbians and gay men on many variables, we included dichotomous variables at Level 2 coding sexual orientation using bisexuals as the reference group (i.e., separately coding gay and lesbian respondents) allowing those variables to predict endogenous variables. As the number of diaries completed varied widely across participants, we also included the total number of diaries the participant completed (grand-mean centered) as a Level 2 predictor of endogenous variables. In addition, baseline levels of well-being predicted average well-being, and baseline physical symptoms predicted average physical symptoms across the diary period (both baseline variables were grand-mean centered). Dichotomous variables coding sexual orientation (at Level 2) and relationship type (at Level 1) were entered uncentered. As the main variables of interest were modeled at both the within- and between-person levels, they were entered uncentered per the recommendations of Preacher and colleagues (2010). To focus the tests of our mediation hypotheses on patterns of results across all participants, the conversation-level effects (i.e., Level 1) were set as fixed. For all multilevel results, unstandardized regression coefficients and their 95% confidence intervals (CIs) are presented in the text and in Table 2.

OPEN IN VIEWER

Table 2.

Multilevel Mediation Model Coefficients.

Level 1 model: Examining associations within conversations			Level 2 model: Examining associations between persons
Predictor	B (SE)	95% CI	Predictor	B (SE)	95% CI
First mediator: Disclosure
Aut Sup	.21 *** (.05)	[0.11, 0.31]	Aut Sup	.61 *** (.17)	[0.27, 0.95]
Family	−.27* (.11)	[−0.48, −0.06]	Total diaries	.002 (.005)	[−0.01, 0.01]
Friend	.41*** (.06)	[0.29, 0.52]	Gay	.25* (.12)	[0.01, 0.49]
Stranger	−.22** (.08)	[−0.38, −0.05]	Lesbian	.35* (.15)	[0.06, 0.64]
Aut Sup × Family	.03 (.08)	[−0.13, 0.20]
Aut Sup × Friend	.19* (.09)	[0.02, 0.37]
Aut Sup × Stranger	.23 a (.13)	[−0.02, 0.48]
Second mediator: Need satisfaction
Aut Sup	.40 *** (.04)	[0.33, 0.48]	Aut Sup	.59 *** (.12)	[0.36, 0.81]
Disclosure	.36 *** (.03)	[0.30, 0.41]	Disclosure	.30 *** (.07)	[0.16, 0.45]
Family	.09 a (.05)	[−0.01, 0.20]	Total diaries	.003 (.003)	[−0.003, 0.01]
Friend	.06 a (.03)	[−0.003, 0.12]	Gay	−.01 (.05)	[−0.11, 0.10]
Stranger	.02 (.03)	[−0.05, 0.08]	Lesbian	−.05 (.06)	[−0.18, 0.07]
Aut Sup × Family	.17** (.05)	[0.06, 0.27]
Aut Sup × Friend	.11 a (.06)	[−0.01, 0.22]
Aut Sup × Stranger	.03 (.06)	[−0.08, 0.15]
First outcome: Well-being
Need satisfaction	.28 *** (.03)	[0.22, 0.34]	Need satisfaction	.30 * (.14)	[0.03, 0.58]
Prior well-being	.13*** (.03)	[0.07, 0.18]	Baseline well-being	.13*** (.04)	[0.06, 0.20]
Time lag	.01 (.01)	[−0.01, 0.03]	Total diaries	−.001 (.004)	[−0.01, 0.01]
Time × Prior Well-Being	−.07*** (.02)	[−0.10, −0.03]	Gay	.02 (.08)	[−0.14, 0.18]
Family	−.06 a (.03)	[−0.12, 0.002]	Lesbian	−.02 (.08)	[−0.19, 0.14]
Friend	−.07* (.03)	[−0.12, −0.01]
Stranger	.03 (.03)	[−0.04, 0.10]
Second outcome: Physical symptoms
Need satisfaction	−.13 * (.05)	[−0.24, −0.02]	Need satisfaction	−.62 a (.36)	[−1.33, 0.09]
Prior symptoms	.19***(.05)	[0.10, 0.29]	Baseline symptoms	.03* (.01)	[0.01, 0.05]
Time lag	.02 (.02)	[−0.02, 0.06]	Total diaries	−.003 (.01)	[−0.02, 0.01]
Time × Prior Symptoms	−.03** (.01)	[−0.06, −0.01]	Gay	−.33 (.22)	[−0.76, 0.10]
Family	.21* (.09)	[.04, 0.38]	Lesbian	−.24 (.25)	[−0.73, 0.25]
Friend	.01 (.05)	[−0.08, 0.10]
Stranger	−.03 (.08)	[−0.18, 0.12]

Note. Bs represent the unstandardized regression coefficients and SEs are their standard errors; Gay and Lesbian (both coded 1) are sexual orientation dummy codes (with bisexuals as the reference group); Friend, Family, and Stranger are dummy codes for close and distant relationship types (coded 1, with all other groups as the reference); Aut Sup × Family, Friend, and Stranger represent interaction terms for these relationship-type dummy codes and autonomy support. Primary predictors are in bold-faced text and modeled at both Levels 1 and 2, covariates are not. CI = confidence interval; Aut Sup = autonomy support.

a

Marginally significant.

*

p < .05. **p < .01. ***p < .001.

Model fit

To evaluate model fit, we followed recommendations by Kline (2005) and Marsh, Hau, and Wen (2004): a comparative fit index (CFI) and Tucker–Lewis index (TLI) above .95, and the root mean square error of approximation (RMSEA) and standardized root mean square residual (SRMR) below .08. Using these benchmarks, our model showed good fit to the data, χ²(34, N = 69) = 50.63, p = .03, CFI = .99, TLI = .98, RMSEA = .02, SRMR_within = .01 SRMR_between = .07. The model also accounted for reasonably large amounts of variance in the predicted (i.e., endogenous) variables: 66% of within-person and 88% of between-person variance for disclosure, 87% within and 97% between for need satisfaction, 20% within and 61% between for physical symptoms, and 89% within and 94% between for well-being.

Predicting disclosure

The first path (a1) in our hypothesized model (Figure 1) was significant at both levels. Thus, at the individual conversation level (Level 1), perceiving autonomy support during conversations predicted more disclosure in those conversations (B = .21, SE = .05, p < .001, CI = [0.11, 0.31]). Similarly, across all conversations (Level 2), perceiving more autonomy support across the entire diary period predicted higher average levels of disclosure across all conversations (B = .61, SE = .17, p < .001, CI = [0.27, 0.95]). The main effects of relationship type suggested that people disclosed sexual orientation significantly more in conversations with close friends (B = .41, SE = .06, p < .001, CI = [0.29, 0.52]) and significantly less with family (B = −.27, SE = .11, p = .01, CI = [−0.48, −0.06]) and strangers (B = −.22, SE = .08, p = .01, CI = [−0.38, −0.05]) as compared with relationships of intermediate levels of closeness (e.g., coworkers, roommates). The interaction terms testing differences in links between autonomy support and disclosure across specific types of relationships suggested that perceiving autonomy support in conversations with close friends—and marginally, from strangers—was more strongly linked to disclosure (friends: B = .19, SE = .08, p = .03, CI = [0.02, 0.37]; strangers: B = .23, SE = .13, p = .07, CI = [−0.02, 0.48]) as both effects suggested a stronger positive association between autonomy support and disclosure for those two relationship types when compared with relationships of intermediate closeness. Thus, relationship type significantly moderated that predictive link. The number of diaries completed did not significantly predict average disclosure across conversations (p > .50) but sexual orientation did—gay men and lesbians disclosed more than bisexuals across the diary period, Bs = .25 and .35, respectively, ps < .05. ²

Predicting need satisfaction

As hypothesized, disclosing was associated with more need satisfaction in the conversation (Path b). This was true for disclosure predicting need satisfaction at the level of conversations (B = .36, SE = .03, p < .001, CI = [0.30, 0.41]) and for average levels of disclosure predicting need satisfaction across the diary period (B = .30, SE = .15, p < .001, CI = [0.03, 0.58]). People also experienced more need satisfaction with autonomy-supportive others (Path a2) at the level of individual conversations (B = .40, SE = .04, p < .001, CI = [0.33, 0.48]) and averaged across the diary period (B = .59, SE = .12, p < .001, CI = [0.36, 0.81]). Examining differences across relationship types, people marginally felt more need satisfaction in conversations with close friends (B = .06, SE = .03, p = .06, CI = [−0.003, 0.12]) and family (B = .09, SE = .05, p = .09, CI = [−0.01, 0.20]) compared with relationships of intermediate closeness. Links between autonomy support and need satisfaction varied meaningfully across relationship types: Perceiving autonomy support in conversations with family and close friends was especially linked to need satisfaction following conversations with those people (family: B = .17, SE = .05, p = .002, CI = [0.06, 0.27]; friends: B = .11, SE = .06, p = .07, CI = [−0.01, 0.22]) when compared with relationships of intermediate closeness. However, the number of diaries completed and sexual orientation did not significantly predict average need satisfaction across conversations (ps > .30).

Predicting well-being

Consistent with hypothesis, experiencing need satisfaction predicted higher well-being (Path c1). This was true for predicting well-being following each conversation (B = .28, SE = .03, p < .001, CI = [0.22, 0.34]) and for predicting average levels of well-being across the diary (B = .30, SE = .14, p = .03, CI = [0.03, 0.58]). This was true even after controlling for levels of well-being following the previous diary assessment (B = .13, SE = .03, p < .001, CI = [0.07, 0.18]), and more recent reports showed a stronger predictive link between previous and current well-being (B = −.07, SE = .02, p < .001, CI = [−0.10, −0.03]). Surprisingly, having conversations with friends and, marginally family, predicted lower well-being (friends: B = −.07, SE = .03, p = .02, CI = [−0.12, −0.01]; family: B = −.06, SE = .03, p = .06, CI = [−0.12, 0.002]) compared with relationships of intermediate closeness. At Level 2, baseline well-being predicted higher average well-being across the diary period (B = .13, SE = .04, p < .001, CI = [0.06, 0.20]). In contrast, the number of diaries completed and sexual orientation did not significantly predict average well-being across conversations (ps > .50).

Predicting physical symptoms

Experiencing need satisfaction in a conversation predicted fewer physical symptoms following that conversation (Path c2: B = −.13, SE = .05, p = .02, CI = [−0.24, −0.02]) but average levels of need satisfaction only marginally predicted fewer average symptoms across the diary period (B = −.62, SE = .36, p = .09, CI = [−1.33, 0.09]). At Level 1, symptoms reported during the prior interaction predicted more current symptoms (B = .19, SE = .05, p < .001, CI = [0.10, 0.29]), and that predictive link was especially strong when prior symptoms were reported more recently (B = −.03, SE = .01, p = .002, CI = [−0.06, −0.01]). Turning to differences across relationship types, conversations with family predicted more physical symptoms (B = .21, SE = .09, p = .02, CI = [0.04, 0.38]) compared with relationships of intermediate closeness. At Level 2, baseline physical symptoms predicted greater average symptoms across the diary period (B = .03, SE = .01, p = .01, CI = [0.01, 0.05]). In contrast, sexual orientation and total number of diaries failed to emerge as significant predictors (ps > .10, CIs pass through 0).

Indirect effects

In this same path model, we estimated all possible indirect effects at the within-person (i.e., at the level of individual conversations) and between-person (i.e., at the level of averages across all conversations) levels, which reduces Type 1 error and a potential confounding of the mediation effect (Preacher et al., 2010). We calculated 95% Monte Carlo CIs for all two- and three-path indirect effects using Tofighi and MacKinnon’s (2016) web-based utility, as this method is recommended in smaller samples and does not assume that indirect effects are normally distributed (Preacher et al., 2010). Although all within-person and some between-person indirect effects were significant (Table 3), we were most interested in the mediation paths of autonomy support predicting well-being and physical symptoms through disclosure and need satisfaction. These indirect effects were significant at the within-person level for both well-being (B = .02, SE = .006, CI = [0.01, 0.03]) and physical symptoms (B = −.01, SE = .005, CI = [−0.02, −0.002]), supporting our model at the level of individual conversations. In addition, one of the corresponding between-person indirect effects emerged as significant (well-being: B = .06, SE = .03, CI = [0.004, 0.14]) whereas the other effect was not significant (physical symptoms: B = −.11, SE = .08, CI = [−0.30, 0.02]). In sum, the reason that everyday experiences of receiving autonomy support predicted greater well-being and fewer physical symptoms was because they helped people be more out about their sexual orientation, which in turn satisfied needs for autonomy and relatedness. Average levels of disclosure and need satisfaction also helped explain the link between average levels of autonomy support and well-being across the diary period. Importantly, this mediation model seemed to operate more robustly at the level of daily interactions (as opposed to people’s mean levels of autonomy support and disclosure), suggesting that everyday opportunities to experience autonomy support and disclose have potent links to everyday mental and physical health.

OPEN IN VIEWER

Table 3.

Indirect Effects at the Within- and Between-Person Levels.

Indirect path		Within-person (Level 1)		Between-person (Level 2)
		B (SE)	95% CI	B (SE)	95% CI
a1×b	autonomy support → disclosure → need satisfaction	.07* (.02)	[0.04, 0.12]	.18* (.07)	[0.07, 0.33]
a2×c1	autonomy support → need satisfaction → well-being	.11* (.02)	[0.08, 0.15]	.18* (.09)	[0.02, 0.37]
a2×c2	autonomy support → need satisfaction → physical symptoms	−.05* (.02)	[−0.10, −0.01]	−.37 (.22)	[−0.83, 0.05]
b×c1	disclosure → need satisfaction → well-being	.10* (.01)	[0.07, 0.13]	.09* (.05)	[0.01, 0.20]
b×c2	disclosure → need satisfaction → physical symptoms	−.05* (.02)	[−0.09, −0.01]	−.19 (.12)	[−0.44, 0.03]
a1×b×c1	autonomy support → disclosure → need satisfaction → well-being	.02* (.006)	[0.01, 0.03]	.06* (.03)	[0.004, 0.14]
a1×b×c2	autonomy support → disclosure → need satisfaction → physical symptoms	−.01* (.005)	[−0.02, −0.002]	−.11 (.08)	[−0.30, 0.02]

Note. Bs represent the unstandardized regression coefficients of the indirect effects, SEs are their standard errors, and 95% CIs are the 95% confidence intervals calculated using the Monte Carlo method.

*

Indicate significance at p < .05 based on the Monte Carlo method.

Discussion

In this study, we examined antecedents and consequences of everyday experiences of being out. Although past studies have highlighted the relations between disclosure and mental and physical health, novel to this study, we investigated factors affecting moment-to-moment decisions to disclose and their effects on mental and physical health using hypotheses derived from SDT. An experience-sampling design was used to capture experiences of coming out or being out as they occurred, limiting retrospective bias and clarifying day-to-day patterns in decisions about disclosure and their links to wellness.

Although the sample was generally out about their sexual identity, people did vary in how out they were to various targets across the 2 weeks. As expected, this variability was not harmful to wellness. Unexpected, however, was that more variability predicted greater well-being and fewer physical symptoms. Taken together, these results suggest that selectively revealing one’s sexual orientation to others does not have clear costs, and may at times yield adaptive benefits. This finding supports a basic premise of this research that even though coming out is usually a good thing for health and wellness, it is adaptive to do so selectively.

A major goal of this research was to examine how receiving autonomy support is associated with decisions about sexual identity disclosure, and how these decisions in turn are associated with daily mental and physical health. Perceptions of autonomy support during interactions indeed predicted being more out and feeling more need satisfaction in a conversation, which in turn predicted greater well-being and fewer physical symptoms. In other words, when people perceived that someone accepted them for who they are, they were more out about their sexual orientation with that person, they felt they could be themselves (autonomy) and felt closer (relatedness) to that person, and they felt better—mentally and physically—following those conversations. Importantly, these indirect paths predicting well-being and physical symptoms were consistently operating at the level of conversations, whereas these indirect paths were inconsistent at the level of averages across the diary period. This is in line with recent findings looking at first coming out experiences—when confidantes reacted positively to people’s first sexual identity disclosure, it predicted higher current levels of well-being because people experienced more autonomy need satisfaction in that relationship (W. S. Ryan, Legate, & Weinstein, 2015). Taken together, these findings suggest that each disclosure opportunity comes with the possibility of boosting need satisfaction, well-being, and buffering against physical health symptoms, and that having accepting, autonomy-supportive others in one’s life reduces barriers to disclosing.

Novel to this research is our focus on everyday physical health. Our data are consistent with research revealing the deleterious costs of concealing sexual orientation on physical health (e.g., Cole et al., 1996; Juster, Smith, Ouellet, Sindi, & Lupien, 2013), but this is the first study examining day-to-day fluctuations in physical health that correspond to day-to-day disclosure decisions and experiences of environmental support. This suggests that it is not just overall concealment or a lack of support that produces negative health outcomes, but that momentary decisions to conceal from others and feeling a lack of support correspond to expressing more physical symptoms that same day.