Internalized Homophobia and Relationship Quality among Lesbians, Gay Men, and Bisexuals

Abstract

We examined the associations between internalized homophobia, outness, community connectedness, depressive symptoms, and relationship quality among a diverse community sample of 396 lesbian, gay, and bisexual (LGB) individuals. Structural equation models showed that internalized homophobia was associated with greater relationship problems both generally and among coupled participants independent of outness and community connectedness. Depressive symptoms mediated the association between internalized homophobia and relationship problems. This study improves current understandings of the association between internalized homophobia and relationship quality by distinguishing between the effects of the core construct of internalized homophobia and its correlates and outcomes. The findings are useful for counselors interested in interventions and treatment approaches to help LGB individuals cope with internalized homophobia and relationship problems.

Internalized homophobia represents “the gay person’s direction of negative social attitudes toward the self” (Meyer & Dean, 1998, p. 161) and in its extreme forms, it can lead to the rejection of one’s sexual orientation. Internalized homophobia is further characterized by an intrapsychic conflict between experiences of same-sex affection or desire and feeling a need to be heterosexual (Herek, 2004). Theories of identity development among lesbians, gay men, and bisexuals (LGB) suggest that internalized homophobia is commonly experienced in the process of LGB identity development and overcoming internalized homophobia is essential to the development of a healthy self-concept (Cass, 1979; Fingerhut, Peplau, & Hgavami, 2005; Mayfield, 2001; Rowen & Malcolm, 2002; Troiden, 1979; 1989). Furthermore, internalized homophobia may never be completely overcome, thus it could affect LGB individuals long after coming out (Gonsiorek, 1988). Research has shown that internalized homophobia has a negative impact on LGBs’ global self-concept including mental health and well being (Allen & Oleson, 1999; Herek, Cogan, Gillis, & Glunt, 1998; Meyer & Dean, 1998; Rowen & Malcolm, 2002).

Recent research on internalized homophobia and mental health has adopted a minority stress perspective (DiPlacido, 1998; Meyer 1995; 2003a). Stress theory posits that stressors are any factors or conditions that lead to change and require adaptation by individuals (Dohrenwend, 1998; Lazarus & Folkman, 1984; Pearlin, 1999). Meyer (2003a, b) has extended this to discuss minority stressors, which strain individuals who are in a disadvantaged social position because they require adaptation to an inhospitable social environment, such as the LGB person’s heterosexist social environment (Meyer, Schwartz, & Frost, 2008). In a meta-analytic review of the epidemiology of mental health disorders among heterosexual and LGB individuals Meyer (2003a) demonstrated differences between heterosexual and LGB individuals and attributed these differences to minority stress processes.

Meyer (2003a) has defined minority stress processes along a continuum of proximity to the self. Stressors most distal to the self are objective stressors—events and conditions that happen regardless of the individual’s characteristics or actions. For the LGB person these stressors are based in the heterosexist environment, such as prevailing anti-gay stereotypes, prejudice, and discrimination. These lead to more proximal stressors that involve, to various degrees, the person’s appraisal of the environment as threatening, such as expectations of rejection and concealment of one’s sexual orientation in an effort to cope with stigma. Most proximal to the self is internalized homophobia: the internalizations of heterosexist social attitudes and their application to one’s self. Coping efforts are a central part of the stress model and Meyer has noted that, as it applies to minority stress, individuals turn to other members and aspects of their minority communities in order to cope with minority stress. For example, a strong sense of connectedness to one’s minority community can buffer the ill effects of minority stress.

Meyer and Dean (1998) have referred to internalized homophobia as the most insidious of the minority stress processes in that, although it stems from heterosexist social attitudes, it can become self-generating and persist even when individuals are not experiencing direct external devaluation. It is important to note that despite being internalized and insidious, the minority stress framework locates internalized homophobia in its social origin, stemming from prevailing heterosexism and sexual prejudice, not from internal pathology or a personality trait (Russell & Bohan, 2006).

Internalized Homophobia and Relationship Quality

As a minority stressor, internalized homophobia has also been linked to several negative outcomes in romantic relationships and non-romantic intimate relationships of LGB individuals. At the core of the prevailing stigma surrounding being LGB are unsubstantiated notions that LGB people are not capable of intimacy and maintaining lasting and healthy relationships (Meyer & Dean, 1998). The anxiety, shame, and devaluation of LGB people and one’s self are inherent to internalized homophobia and are likely to be most overtly manifested in interpersonal relationships with other LGB individuals (Coleman, Rosser, & Strapko, 1992). To the extent that LGB people internalize these notions, they could manifest in intimacy-related problems in many forms.

Experiencing these negative feelings in the context of sexual and other intimate interactions is likely to decrease the quality of and satisfaction with one’s relationships. To alleviate these feelings, individuals may avoid lasting and deep relationships with other LGB people and/or seek avenues for sexual expression devoid of intimacy and interpersonal closeness. Within coupled romantic relationships, one’s partner and shared experiences serve as constant reminders of one’s own sexual orientation. Internalized homophobia can thus lead to problems related to ambivalence, relational conflict, misunderstandings, and discrepant goals (Mohr & Fassinger, 2006). Also, individuals who view themselves negatively because they are LGB, are likely to be perceived as less attractive relationship partners than individuals who have more positive views of themselves.

Empirical evidence supports these theoretical claims. With regard to romantic relationships, Meyer and Dean (1998) demonstrated that gay men with higher levels of internalized homophobia were less likely to be in intimate relationships, and when they were in relationships, they were more likely to report problems with their partners than gay men with lower levels of internalized homophobia. Similarly, Ross and Rosser (1996) demonstrated that among gay and bisexual men internalized homophobia was negatively associated with relationship quality and the length of individuals’ longest relationships. Other researchers have shown that internalized homophobia negatively affects relationship functioning by reducing individuals’ efforts to maintain relationships in the face of partner conflict (Gains, Henderson, Kim, Gilstrap, Yi, Rusbut, et al., 2005). Internalized homophobia has been linked to poor relationship quality within both male and female same-sex relationships (Balsam & Szymanski, 2005; Otis, Rostosky, Riggle, & Hamrin, 2006).

With regard to non-romantic relationships, internalized homophobia can affect the quality of LGB individuals’ friendships, familial relationships, and other social relationships. For example, a higher level of internalized homophobia has been linked to loneliness (Szymanski & Chung, 2001), less social support in general, and less support specifically from other LGBs (as a proportion of all support received; Shidlo, 1994).

Research suggests that internalized homophobia also affects gay and bisexual men’s experience of sexual intimacy. Higher levels of internalized homophobia are associated with greater sexual depression, sexual anxiety, sexual image concern, and fear of sexuality as well as lower levels of sexual esteem and sexual satisfaction and are predictive of sexual problems among gay and bisexual men (Dupras, 1994; Meyer, 1995). Although there is less research about sexual intimacy among women, internalized homophobia has also been implicated in sexual problems among lesbians and bisexual women (Nichols, 2004).

Distinguishing Internalized Homophobia from Its Outcomes and Correlates

Researchers have disagreed about what constitutes internalized homophobia and how it is distinct from associated constructs (Currie, Cunningham, & Findlay, 2004; Meyer & Dean, 1998; Nungesser, 1983; Ross & Rosser, 1996; Shildo, 1994: Szymanski & Chung, 2001). Most significantly, some have included in the definition of internalized homophobia the degree to which the person is out about his/her sexual orientation (we refer to this as “outness” here) and connected to the LGB community (Mayfield, 2001; Shildo, 1994; Williamson, 2000). Also, some have considered depression and suicidal thoughts (Nungesser, 1983; Shildo, 1994) as well as hopelessness about one’s future (Szymanski & Chung, 2001) as part of internalized homophobia because, as we showed above, these are often associated with internalized homophobia.

The minority stress model differs from these perspectives in that it conceptualizes internalized homophobia and outness as two separate minority stressors and community connectedness as a mechanism for coping with minority stress. Depression is conceptualized as a potential outcome of internalized homophobia (Meyer, 2003a). Applying the minority stress model to understand how internalized homophobia is distinctly related to relationship quality is important given the lack of consistency in the field regarding associations between outness, community connectedness, depression, and relationship quality. For example, outness has been shown to be indicative of better relationship quality by some researchers (Caron & Ulin, 1997; Lasala, 2000), while others have found that outness was not related to relationship quality (Balsam & Szymanski, 2005; Beals & Peplau, 2001). Although community connectedness has been an important aspect of internalized homophobia in some models, we were aware of no studies that explicitly examine its association with relationship quality independently of other aspects of internalized homophobia. Further, researchers have yet to examine the unique ways in which internalized homophobia is related to relationship problems in LGB lives, independent of depressive symptoms.

The treatment of outness as an aspect of internalized homophobia stems from psychologists’ view that coming out is a positive developmental stage in LGB identity development (Cass, 1979). Coming out to important people in one’s life may indicate that one has overcome personal shame and self-devaluation associated with being LGB. But, we contend, lack of outness should not be taken to indicate the opposite and therefore should not be conceptualized as a part of internalized homophobia (Eliason & Schope, 2007).

Being out regarding one’s sexual orientation follows self-acceptance, but even after completely accepting one’s self as lesbian, gay, or bisexual, an LGB person may decide not to be out in certain situations. Outness is often solely a function of situational and environmental circumstances that are unrelated to internal conflict. Disclosing an LGB orientation is affected by opportunities for and expected risks and benefits from the disclosure. For example, others’ knowledge of one’s sexual orientation was shown to be related to external pressures such as having experienced discrimination and physical and verbal abuse (Frost & Bastone, 2007; Schope, 2004), suggesting that choosing not to disclose can be self-protective. A good example of this are men and women in the U.S. military who are barred from coming out by law and risk dismissal if they come out (Herek & Belkin, 2005). Another example pertains to LGB individuals in the work place. Rostosky and Riggle (2002) demonstrate that coming out at work is a function not only of individuals’ levels of internalized homophobia, but also their perceiving a safe and nondiscriminatory work environment. Clearly, concealing sexual orientation in an unsafe environment is a sign of healthy adjustment to environmental constraints and should not be considered indicative of internalized homophobia. As Fassinger and Miller (1996) note, “disclosure is so profoundly influenced by contextual oppression that to use it as an index of identity development directly forces the victim to take responsibility for his or her own victimization” (p. 56, in Eliason & Schope, 2007).

Similar issues arise in conceptualizing internalized homophobia when considering its relationship to affiliation with the lesbian, gay, and bisexual community. A sense of connectedness with similar others may serve to remind LGB people that they are not alone, provide social support for dealing with stress, and allow them to make more favorable social comparisons (Crocker & Major, 1989; Lewis, Derlega, Clarke, & Kuang, 2006; Smith & Ingram, 2004). Individuals with a higher level of internalized homophobia may be less likely to feel connected with the gay community, but this is not always the case. Although few studies examine this relationship, it is plausible that, similar to outness, participation in the gay community is related to opportunities for and risk in doing so. For example, individuals in areas lacking a strong numeric representation of LGB individuals may not have a high level of connectedness to the gay community simply because there is little or no presence of similar others. Also, it is plausible that connection to the LGB community may have a different level of importance for single and coupled LGB individuals. Single LGBs may rely on community to serve social support functions, however coupled individuals may not rely on the community as much in this regard. Thus, lack of connection with the community is not necessarily a reflection of internalized homophobia and should be considered as a separate construct so that researchers can tease apart these constructs in understanding their associations with relationship quality.

The associations between internalized homophobia, depressive symptoms, and relationship quality are obscured by conceptualizations of internalized homophobia that involve a considerable amount of overlap with depressive symptoms. Studies have consistently demonstrated a direct relationship between internalized homophobia and depressive symptoms (e.g., Igartua, Gill, & Montoro, 2003; Meyer, 1995; Shildo, 1994; Szymanski, Chung, & Balsam, 2001). These findings are in accordance with the minority stress model, which conceptualizes internalized homophobia as a minority stressor which causes mental health problems including depressive symptoms (Meyer, 2003a).

Few, however, have empirically studied whether or not internalized homophobia and depressive symptoms are independently related to relationship quality (Biss & Horne, 2005). Studies have linked increased depressive symptoms with problems in intimate relationships (Burns, Sayer, & Moras, 1994; Davila, Karney, Hall, & Bradbury, 2003; Golan, Friedman, & Miller, 2002). Additional research on the interpersonal aspects of depression has demonstrated that individuals who are depressed bring about negative affect, anxiety, and tension within their relationship partners, which in turn, causes relationship problems in the form of misunderstandings and rejection (Coyne, Kahn, & Gotlib, 1987; Coyne, Kessler, Tal, Turnbull, Wortman, Greden, 1987). These findings suggest that internalized homophobia may lead to increased depressive symptoms that, in turn, reduce relationship quality.

The Current Study

We examined the association between internalized homophobia and the quality and closeness of individuals’ interpersonal relationships with friends and family and within romantic relationships. Specifically, we investigated internalized homophobia’s association with sexual problems, loneliness, and the quality of individual’s interpersonal relationships and, among coupled individuals, relationship strains (e.g., relational conflict, misunderstandings). We assessed internalized homophobia, outness, community connectedness, and depressive symptoms as separate, independent constructs in the minority stress experience. We then examined the extent to which depressive symptoms mediated the relationship between internalized homophobia and relationship quality.

Our hypothesized model is outlined in Figure 1. Specifically, we hypothesized that internalized homophobia would positively affect relationship problems independent of outness, community connectedness, and depressive symptoms (path a). We hypothesized that depressive symptoms would partially mediate the effect of internalized homophobia on relationship problems (paths b and c). Consistent with previous theory and research, we expected that a higher level of internalized homophobia would be associated with less outness and less affiliation with the LGB community. We did not have specific hypotheses regarding the effects of outness and community connectedness¹ on relationship problems (paths d and e), but we isolated the effects of these factors so that we could examine the independent effect of internalized homophobia on relationship problems.

Figure 1.

Theoretical model explaining relationship problems and relationship strain among LGB individuals.

Method

The data analyzed in the current study were obtained as part of Project Stride, a large epidemiological study that investigated the relationships between stress, identity, and mental health among diverse LGB and heterosexual populations in New York City. Participants in Project Stride were 396 LGB and 128 heterosexual individuals. The current study only includes data from the LGB participants. (Detailed information about Project Stride is available online at http://www.columbia.edu/~im15/).

Participants and Procedure

Participants (N = 396) were sampled between February 2004 and January 2005 from venues in New York City chosen to represent a wide diversity of cultural, political, ethnic, and sexual communities. Sampling venues included business establishments (e.g., bookstores, cafes), social groups, and outdoor areas (e.g., parks), as well as snowball referrals. Recruitment of participants occurred in 2 phases. In the first phase, 25 outreach workers visited a total of 274 venues in 32 different New York City zip codes. For each potential participant, recruiters completed a brief screening form that would determine eligibility for participation in the study. In the second phase, eligible participants were contacted by research interviewers and invited to participate in a face-to-face interview. Participants were eligible if they were 18–59 years-old, New York City residents for two years or more who could communicate in English and self-identified as: a) lesbian, gay, or bisexual; b) male or female; and c) white, black or Latino (participants may have used other identity terms in referring to these social groups). We used quota sampling to ensure approximately equivalent numbers of participants across gender, race/ethnicity, and age group (18–30 and 31–59). The response rate was 60%, defined according to the American Association for Public Opinion Research (AAPOR, 2005; formula RR2) as the number of complete and partial interviews divided by the number of complete and partial interviews, refusals, and eligible non-contacts (individuals who screened eligible in phase 1 whom we could not contact for an interview). The cooperation rate was 79%, calculated in the same way as the response rate, but excluding non-contacts (AAPOR, 2005, formula COOP2). Response and cooperation rates did not vary greatly by sexual orientation, race/ethnicity, or gender (χ² s ≤ 0.78, ps ≥ .38).

Recruitment efforts were successful at reaching individuals who resided in diverse New York City neighborhoods and avoiding concentration in particular “gay neighborhoods” that is often characteristic of sampling of LGB populations. Participants resided in 128 different New York City zip codes; no more than 4% of the sample resided in any one zip code area. Participants’ mean age was 32.43 years (SD = 9.24). By design, the sample included about equal numbers of men and women (n = 198), and White (n = 134, 34%), Black (n = 131, 33%), and Latino (n = 131, 33%) participants. The median per capita income was $27,500, 16% (n = 68) were unemployed, and 22% (n = 86) had a high school education or less. A total of 71 (18%) identified as bisexual and the rest as gay or lesbian (including similar terms, such as queer or homosexual). Approximately half (n = 184, 47%) of the participants were in a relationship (73 men and 111 women). The mean length of their relationships was 3.21 years (SD = 3.50, Median = 2). A total of 26 men and 50 women reported living with their partners; five men and 21 women were married or registered as domestic partners.

Participants completed in-person interviews lasting a mean of 3.82 hours (SD = 55.00 minutes). Interview were conducted by interviewers trained to be sensitive to the concerns of the LGB community aided by the use of a Computer-Assisted Personal Interview (CAPI). The research protocol was reviewed and approved by the Western Institutional Review Board. Participants signed a written informed consent form after the study procedure had been explained to them and were paid $80 upon completing the interview.

Measures

Internalized homophobia (IHP)

The IHP scale was originally developed by Martin and Dean (1992) to assess the extent to which LGB individuals reject their sexual orientation, are uneasy about their same-sex desires, and seek to avoid same-sex attractions and sexual feelings (Herek & Glunt, 1995; Meyer, 1995; Meyer & Dean, 1998). This measure was designed to assess the construct of internalized homophobia as we defined it above in the context of the minority stress model: distinct from mental health outcomes and isolated from concerns with community connectedness and outness. The original scale consisted of 9 items. To avoid confounding between internalized homophobia and community connectedness, of particular interest to our study, we eliminated one item from the original measure that reads “I often feel it best to avoid personal or social involvement with other gay men.” The 8-item scale included, for example, how often participants have “wished you weren’t gay, “felt alienated from yourself because of being gay,” and “felt that being gay is a personal shortcoming.” Participants rated the frequency with which they experienced such thoughts and feelings in the year prior to the interview on a 4-point scale ranging from 1 = “often” to 4 = “never.” Items were worded so that the subject of the question matched the participant’s self-reported sexual identity label so that “gay” in the examples above was replaced with “lesbian,” “bisexual,” or “queer,” as relevant to the participant. Scores were recoded so that higher scores indicated more internalized homophobia. Previous studies have demonstrated that scores on this scale have internal consistency reliability of .79 (Meyer, 1995; Meyer & Dean, 1998) to .83 (Lewis, Derlega, Griffin, & Krowinski, 2003). Internal consistency for scores on internalized homophobia in the current study was .86. In a sample of gay men and lesbians, Herek et al. (1998) demonstrated convergent validity for the scale through significant correlations with individual self-esteem (for gay men), and collective self-esteem (for both gay men and lesbians). In a study of gay fathers, Sbordone (1993; as cited by Shildo, 1994) reported that this measure of internalized homophobia significantly correlates with another widely used measure of internalized homophobia: the Nungesser Homosexuality Attitudes Inventory (Nungesser, 1983).

Depressive symptoms

The Center for Epidemiological Studies depression scale (CES-d; Radloff, 1977) is a 20-item measure of depressive symptoms experienced over a one week period prior to the interview. Items were phrased in such a way that participants were asked how often during the past week they “could not get going,” “felt depressed,” “felt hopeful about the future,” and “felt people dislike you.” Participants responded on a 4-point scale ranging from 1 = “rarely or none of the time (<1 day)” to 4 = “most or all of the time (5–7 days).” Previous studies have demonstrated that scores on this scale have internal consistency reliability of .85 in the general population (Radloff, 1977) and .87 to .92 among LGBs (Frost, Parsons, & Nanin, 2007; Lewis et al., 2003). Numerous studies have demonstrated the convergent validity of the CES-d among both clinical and non-clinical samples in the form of large correlations with clinical reports of depression, DSM depression diagnoses, the Hamilton Rating Scale for Depression, and the Symptom Checklist-90 (for a review of validity evidence see McDowell & Newell, 1996; Roberts & Vernon, 1983). Although the scale has been shown to correlate moderately to highly with other measures of anxiety and psychological distress, it has been successful in identifying depression in several clinical and community samples (McDowell & Newell, 1996) and as a result is one of the most widely used measures of depressive symptoms. Internal consistency for scores on the CES-d in the current study was .92.

Outness

This measure assessed the degree of disclosure of sexual orientation to (a) family, (b) straight friends, (c) LGB friends, and (d) co-workers (Meyer, Rossano, Ellis, & Bradford, 2002). Participants described the extent to which they were “out of the closet” to each of these groups on a scale of 1 = “out to none” to 4 = “out to all.” The measure has good face validity, using simple language and referring to behaviors that are commonly discussed among LGB individuals. Preliminary evidence of validity from the current study is provided by the significant negative correlations between outness and internalized homophobia and community connectedness (see Table 1). Internal consistency for scores on the 4 outness items in the current study was .75.

Table 1.

Means, Standard Deviations, and Correlations among the Overall Scales.

	All (N = 396)		Coupled (n = 184)		Pearson Correlations
Variable	M	SD	M	SD	1	2	3	4	5	6	7
1. IHP	1.40	0.57	1.40	0.55	--	−.43**	−.30**	.20**	.20**	.14**	.23**
2. Outness	3.31	0.71	3.29	0.67	−.41**	--	.25**	−.06	−.11	−.11	−.15**
3. Connectedness	3.28	0.53	3.31	0.52	−.38**	.29**	--	−.05	−.05	−.07	−.18**
4. Depressive Symptoms	1.71	0.56	1.71	0.52	.23**	−.07	.00	--	.37**	.31**	.39**
5. Sex Problems	2.10	0.84	--	--	--	--	--	--	--	.15**	.26**
6. Loneliness Strain	1.47	0.58	--	--	--	--	--	--	--	--	.36**
7. Positive Relations with Others	2.85	1.32	--	--	--	--	--	--	--	--	--
8. Relationship Strain	--	--	1.38	0.40	.20**	−.18*	.05	.24**	--	--	--

Note: Correlations for the entire sample are presented above the principal diagonal of the correlation matrix. Correlations for coupled participants only are presented below the principal diagonal of the correlation matrix. Sex Problems, Loneliness Strain, and Positive Relations with Others were assessed for all participants in the study, but not analyzed separately for coupled participants only. Relationship strain was assessed for coupled participants only.

^*p < .05,

^**p < .01.

Connectedness to the LGB community

Community connectedness was assessed with an 8-item scale, adapted from a 7-item community cohesion scale used in the Urban Men’s Health Study (UMHS), a multi-city study of gay men’s psychological and physical health (Mills et al., 2001). We added one item—“You feel a bond with other [men who are gay or bisexual]” taken from Herek and Glunt’s (1995) community consciousness scale—to the UMHS scale to capture symbolic affiliation that did not denote activity. The scale was further modified to specify participation in New York City’s LGB community. To aid participants in answering these questions, they were read by the interviewer a definition of community as used in the scale, which stated, “I don’t mean any particular neighborhood or social group, but in general, groups of gay men, bisexual men and women, and lesbians.” Participants rated on a scale of 1 = “agree strongly” to 4 = “disagree strongly” how much they agreed with items such as “Participating in NYC’s LGBT community is a positive thing for me” and “I really feel that any problems faced by NYC’s LGBT community are also my problems.” Scores were recoded so that higher scores indicated more connectedness. Scores on the measure demonstrated internal consistency of .78 in the UMHS (Barrett & Pollack, 2005) and with the addition of the new item scores on this measure in the current study had a Cronbach’s alpha of .80. Although there is no published data explicitly addressing the validity of this measure, in the current study connectedness to the LGB community was significantly negatively correlated with internalized homophobia and outness (see Table 1) and significantly positively correlated with the number of LGB-related community or recreational groups participants were members of or active in (measured by a 9-item checklist developed by Mills et al., 2001 for the UMHS; r = .31, p < .001).

Relationship and loneliness strain

Participants’ experiences of chronic strains across a multitude of dimensions were assessed based on Wheaton’s (1999) conceptualization of chronic strain. In compiling items for an inventory of chronic strain we followed the procedure used by Turner and others (e.g., Turner & Avison, 2003; Turner, Wheaton, & Lloyd, 1995) aiming to make the items culturally relevant to our participants and theoretically relevant to the overall study (Meyer et al., 2008). In Project Stride, our inventory of chronic strains contained questions designed to assess strain across three following life domains: Job/work; unemployment; finances; education; parenting; residence; relationships; loneliness; significant other’s illness/health; caretaking responsibilities; relationship with parents; wanting children; and general strain (for a more detailed description of the full inventory see Meyer et al., 2008). We used items from two of these life domains as indicators of relationship quality in the current study: loneliness and relationship strain. Relationship strain was assessed among coupled participants only using seven items, such as “You have a lot of conflict with your partner/boyfriend/girlfriend” and “Your partner/boyfriend/girlfriend expects too much out of you.” These items were taken directly from the relationships section of an inventory of chronic strains previous developed by Turner and colleagues for use in the Physical Challenges and Health Study (the full inventory can be assessed online at http://www.fiu.edu/~lchrc/challenge.htm). We did not include one item from the relationships section of this inventory (i.e., “Your sexual needs are not fulfilled by this relationship”) because it overlapped with the measure of sex problems. Loneliness strain was assessed for all individuals in the study using two items. One item was taken from the social life and recreation section of Turner’s inventory (i.e., “You don’t have enough friends”) and we added an item which read “You are alone too much.” Participants were asked to indicate how true each statement was for them on a scale of 1 = “very true” to 3 = “not true.” Scores were recoded so that higher scores indicated more of each construct. Internal consistency for relationship and loneliness strain in the current study were .86 and .62, respectively. Because inventories of chronic strains vary from study to study and are most often analyzed in aggregate, no evidence for the validity of these two sets of items as distinct measures is available.

Positive relations with others

We used the 3-item positive relations with others scale, which is one of Ryff’s (Ryff & Keyes, 1995) six psychological well-being scales. The six psychological well-being scales were developed to integrate theories of life course development and positive mental health conceptions of psychological well-being using a construct-oriented approach to personality assessment. The positive relations with others subscale assessed the degree of warmth and trust in individuals’ interpersonal relationships, broadly conceived. Participants were asked whether they agreed or disagreed with each of three items, such as “I have not experienced many warm and trusting relationships with others” on a scale of 1 = “disagree completely” to 7 = “agree completely.” Scores for positive relations with others were reverse coded, so higher scores indicated more problems with having positive relations with others. Internal consistency for scores on this measure in the current study was .54. This is consistent with what others have found for this subscale and stems from the authors’ desire to create a brief measure and maintain the multidimensionality of the overall psychological well-being construct at the expense of internal consistency (Chrouser Ahrens, & Ryff, 2006; Ryff & Keyes, 1995). The items for the positive relations with others scale have been consistently demonstrated as representing a distinct single factor in Ryff’s 6-factor model of psychological well-being (Ryff & Keyes, 1995; Ryff & Singer, 2006). The positive relations with others scale has also been shown to differentiate between the other five psychological well-being scales as well as indicators of happiness, life satisfaction, and depression (Ryff & Keyes, 1995). The subscale correlates highly with personality trait agreeableness (Schmutte & Ryff, 1997). Furthermore, providing evidence of validity based on expected demographic differences, Marks and Lambert (1998) showed that individuals who are separated or divorced are significantly lower on positive relations with others than married individuals.

Sex problems

We used the sex problems subscale of the Psychiatric Epidemiology Research Interview (PERI) (Dohrenwend, Shrout, Egri, & Mendelsohn, 1980). The PERI contains 25 subscales, 8 of which measure general components of psychological distress (e.g., anxiety, sadness, hopelessness). The remaining 17 subscales, including sex problems, measured other symptoms (e.g., guilt, rule breaking, enervation). The sex problems subscale indicated the frequency of inhibited sexual desire, excitement, or orgasm over the 12 months prior to the interview (4 items for women and 5 items for men). Participants were given response choices ranging from 1 = “never” through 5 = “very often” to questions such as “How often have you had no interest in sex?” Scores on this measure in a previous study of minority stress among gay and bisexual men demonstrated internal consistency at the level of .72 (Meyer, 1995). Furthermore, internal consistency for scores on the measure across seven waves of data collected between 1985 and 1991 among gay men in New York City ranged from .69 to .73 (Martin & Dean, 1992). Internal consistency for scores on sex problems in the current study was .71 and .74 for men and women, respectively. The subscale was designed to measure a construct independent of the other subscales within the PERI measure. Specifically, in a sample of NYC heads of households, although sex problems correlated moderately with the anxiety subscale (r = .33), sex problems correlated with the other 23 subscales at a level of r = .28 or lower, demonstrating discriminant validity (Dohrenwend et al., 1980).

Results

Data Preparation and Preliminary Analyses

Scale scores were distributed approximately normally, with the exception of the item measuring outness to LGB friends (skewness = −3.70, kurtosis = 14.43). To ensure univariate normality among the variables in the study, we computed a new “out to friends” score, which combined the average of the out to LGB friends and out to straight friends items. This new variable was distributed approximately normally (skewness = −1.57, kurtosis = 1.93). Descriptive statistics and Pearson correlations among the overall scale scores are presented in Table 1, for all participants and for coupled participants only. Missing data were minimal (i.e., = 2% on all measures), were determined to be random, and missing values were replaced with mean substitution using the mean from each participant’s corresponding demographic group based on age, gender, race/ethnicity.

We conducted structural equation modeling (SEM) analyses in AMOS^® using latent variables to test the hypotheses outlined above (Frazier, Tix, & Barron, 2004). Because we did not have multiple scale indicators of several of the predictor variables in the study, we created observed indicators by computing item parcels, or average scores on a subset of the scale items, using the item-to-construct technique for creating item parcels (Little, Cunningham, Shahar, & Widaman, 2002). Three parcels were created for each of the following scales: Internalized Homohpobia (two, 3-item parcels, and one, 2-item parcel); Center for Epidemiological Studies –Depression scale (two, 7-item parcels, and one, 6-item parcel); community connectedness (two, 3-item parcels, and one, 2-item parcel); and relationship strain for coupled participants only (two, 2-item parcels and one, 3-item parcel). This involved conducting an exploratory factor analysis individually with each scale forcing a single-factor solution. Based on the factor loadings, we assigned the three highest loading items to be the basis for each of the three parcels. Next, we assigned the remaining three highest loading items to the parcels in reverse order. We repeated this process until all items were assigned to parcels. Parceling in the current study was deemed appropriate given the constructs used in the study were unidimensional, and our focus was on the relationships between latent constructs, not the measurement and interrelatedness of each individual item (Little et al., 2002).

We used AMOS^® to test for multivariate normality among the observed variables to be used in testing the hypothesized SEM models. The data did not demonstrate multivariate normality. We observed a multivariate kurtosis value of 47.35 with a critical ratio of 20.86, which exceeded the cutoff point of 1.96 that tests multivariate normality. When conducting SEM analyses with data that are not multivariate normal, the chi-square indicator of model fit is overestimated and the standard estimates used to test the significance of parameter estimates are underestimated. To correct for this problem, we used a bootstrapping procedure (Bollen & Stein, 1992; 1993) in AMOS^® to calculate (a) the average standardized path coefficients, their standard errors, and associated probability values based on estimates from 10,000 samples dawn randomly from the 396 participants in the study, and (b) the Bollen-Stine adjusted probability values for the chi-square tests of model fit.

SEM Tests of the Association between Internalized Homophobia and Relationship Problems

At each step in the SEM analyses testing the relationship between internalized homophobia and relationship problems, we fit models separately² for (a) all participants in the study (N = 396) with a latent outcome of relationship problems, and (b) coupled participants only (n = 184) with a latent outcome of relationship strain. We followed the two-step process recommended by Anderson and Gerbing (1988) which requires first demonstrating adequate fit of the measurement models using confirmatory factor analysis followed by testing the fit of the proposed structural models. In addition to the model chi-square, we used four additional indicators of good model fit: The relative chi-square, which is computed by dividing the model chi-square by the degrees of freedom for the model (values less than three; Carmines & McIver, 1981); the root-mean-square error of approximation (RMSEA; values below .06; Hu & Bentler, 1999); standardized root-mean-square residual (SRMR; values below .08; Hu & Bentler, 1999); and the comparative fit index (CFI; values above .95; Hu & Bentler, 1999).

Measurement models

Table 2 presents the factor loadings for the observed indicators for the measurement models as well as the correlations between the latent variables in the models. The measurement model including relationship problems for all participants in the study fit the data well, χ² (80, n = 396) = 133.50, p < .05 (Relative χ² = 1.67; CFI = .979; RMSEA = .041, 90% CI = .028, .053; SRMR = .035). In this model, the factor loadings for all observed indicators were statistically significant. Furthermore, internalized homophobia, depression, and relationship problems were significantly correlated with one another. Outness and community connectedness were also significantly correlated with internalized homophobia as well as relationship problems. The measurement model including relationship strain for coupled participants only, also fit the data well, χ² (80, n = 184) = 98.24, p = .44 (Relative χ² = 1.23; CFI = .985; RMSEA = .035, 90% CI = .000, .057; SRMR = .044). Factor loadings for all observed indicators were statistically significant in this model as well. Among coupled participants only, internalized homophobia, depression, and relationship problems were significantly correlated with one another. Outness and community connectedness were also significantly correlated with internalized homophobia, but were not correlated with relationship strain. The results of both models indicated that all observed variables were adequate indicators of their corresponding latent constructs. Thus, we proceeded to fit the structural models.

Table 2.

Factor Loadings and Correlations among Latent Variables Obtained from the Measurement Models

			Latent Factor Correlations
Latent Factors and Observed Variables	Standardized Factor Loadings		1	2	3	4	5
	All (N = 396)	Coupled (n = 184)
1.Internalized Homophobia			--	.20**	−.35**	−.53**	.37**
IHPP1	.87**	.84**
IHPP2	.86**	.86**
IHPP3	.75**	.67**
2.Depressive Symptoms			.25**	--	−.07	−.08	.70**
CESDP1	.91**	.90**
CESDP2	.90**	.88**
CESDP3	.86**	.85**
3.Connectedness			−.43**	.01	--	.29**	−.24**
CONNECTP1	.75**	.81**
CONNECTP2	.82**	.81**
CONNECTP3	.71**	.65**
4.Outness			−.56**	−.10	.33**	--	−.26**
OUTFAM	.64**	.54**
OUTCOW	.70**	.74**
OUTFRND	.85**	.80**
5.Relationship Problems			--	--	--	--	--
Sex Problems	.46**	--
Loneliness Strain	.47**	--
Positive Relations w/Others	.62**	--
6.Relationship Strain			.27**	.34**	.06	−.20	--
CHRRELP1	--	.73**
CHRRELP2	--	.78**
CHRRELP3	--	.80**

Note: All values were calculated using bias-corrected bootstrapping with 10,000 samples. Correlations for the entire sample are presented above the principal diagonal of the correlation matrix. Correlations for coupled participants only are presented below the principal diagonal in the correlation matrix. Relationship Problems were assessed for all participants in the study. Relationship strain was assessed for coupled participants only. IHPP1-IHPP3 = Internalized Homophobia item parcels; CESDP1-CESDP3 = CES-d item parcels; CONNECTP1 - CONNECTP3 = Community connectedness parcels; OUTFAM = Degree of being out to family; OUTCOW = Degree of being out to co-workers; OUTFRND = Degree of being out to friends; CHRRELP1 - CHRRELP3 = Relationship chronic strain item parcels.

^*p < .05,

^**p < .01.

Structural models

To test our hypotheses outlined in Figure 1, we followed the steps provided by Holmbeck (1997; see also Frazier et al., 2004) using the bootstrapping techniques described above. Model 1 tested the direct and uncontrolled relationship between internalized homophobia and relationship problems (Figure 1, path a only). Model 2 tested the direct effect of internalized homophobia on relationship problems controlling for community connectedness and outness (Figure 1, paths a, d, and e). Model 3 examined the extent to which depressive symptoms completely mediated the relationship between internalized homophobia and relationship problems (Figure 1, paths b, c, d, and e). These models were compared to Model 4 –the full hypothesized model – which modeled the relationship between internalized homophobia and relationship problems as partially mediated by depressive symptoms (Figure 1, all paths). Each model was tested separately for all participants in the study and separately for coupled participants only. The results are reported in Table 4. In our description of these models below, we use the suffix a to refer to models tested among all participants in the study and b to refer to models tested among coupled participants only.

For all participants in the study, we observed a significant uncontrolled direct effect of internalized homophobia on relationship problems, β = .33, SE = .08, p < .01 (model 1a). This relationship remained substantial and statistically significant when community connectedness and outness were controlled for, β = .24, SE = .09, p < .05 (model 2a). Both models fit the data well (see Table 4). The complete mediation model (model 3a) also fit the data well; however, this model fit the data slightly less well than the hypothesized partial mediation model (model 4a). The addition of the direct effect of internalized homophobia on relationship problems did not significantly improve the fit of the model (Δχ² (1) = 2.8, p = .09). The hypothesized partial mediation model predicting relationship problems among all participants in the study is presented in Figure 2. As hypothesized, internalized homophobia was associated with increased depressive symptoms and depressive symptoms were associated with increased relationship problems. Internalized homophobia explained approximately 4% of the variance in depressive symptoms in this model. The direct effect of internalized homophobia on relationship problems was not statistically significant. These relationships were controlled for the effects of community connectedness and outness, which were both not significantly associated with relationship problems. This model accounted for 57% of the variance in relationship problems among all participants in the study.

Figure 2.

Structural model explaining relationship problems among all participants in the study (N = 396).

Note: Numbers represent standardized path coefficients obtained from bootstrapping using 10,000 samples. ** p < .01.

For coupled participants only, we also observed a significant direct effect of internalized homophobia on relationship strain, in both the uncontrolled model, β = .27, SE = .10, p < .01 (model 1b), and controlled model, β = .31, SE = .16, p < .05 (model 2b). In fact, the effect size of the direct effect of internalized homophobia on relationship strain slightly increased when community connectedness and outness were controlled for. Both models fit the data well (see Table 4). The complete mediation model (model 3b) also fit the data well, but again, slightly worse than the hypothesized partial mediation model (model 4b). The addition of the direct effect of internalized homophobia on relationship strain did not significantly improve the model (Δχ² (1) = 3.37, p = .07). The hypothesized partial mediation model predicting relationship strain among coupled participants only is presented in Figure 3. Internalized homophobia was associated with increased depressive symptoms and increased depressive symptoms were associated with increased relationship problems. Internalized homophobia explained approximately 6% of the variance in depressive symptoms among coupled participants only. The direct effect of internalized homophobia on relationship strain was not statistically significant. Unlike the model fit to the total sample, in the model fit for coupled individuals, a higher degree of community connectedness was associated with increased relationship strain. As in the total sample, outness was not significantly related to decreased relationship problems. This model accounted for 18% of the variance in relationship strain among coupled participants.

Figure 3.

Structural model explaining relationship strain among coupled participants (N = 184).

Note: Numbers represent standardized path coefficients obtained from bootstrapping using 10,000 samples. * p < .05, ** p < .01.

Significance of indirect effects

The results of the above models suggest that the effects of internalized homophobia on relationship problems and relationship strain are mediated by depressive symptoms. We therefore followed the procedure for testing the significance of the mediated effects of internalized homophobia outlined by Shrout and Bolger (2002) and Mallinckrodt, Abraham, Wei, and Russell (2006). To calculate the estimates of indirect effects, we multiplied the standardized path coefficient linking internalized homophobia to the mediator (i.e., depressive symptoms) by the standardized path coefficient linking the mediator to the outcome (i.e., relationship problems or relationship strain) obtained from the bootstrapping procedure using 10,000 samples. We also obtained the standard errors and 95% bias corrected confidence intervals around these estimates. According to Shrout and Bolger (2002), if the bias-corrected 95% confidence intervals around the estimates for the indirect effects do not contain 0, it is possible to conclude that the indirect effects are statistically significant at the level of p < .05. For all participants in the study, the standardized indirect effect of internalized homophobia on relationship problems was significant, (.20) X (.66) = .13, SE = .04, CI = .06, .22. Among coupled participants only, the standardized indirect effect of internalized homophobia on relationship strain was also significant, (.24) X (.28) = .07, SE = .04, CI = .01, .12.

Discussion

We aimed to assess the association between internalized homophobia and relationship problems using the minority stress model as our theoretical reference. Using this theoretical perspective, we conceptualized internalized homophobia as a minority stressor, separating the core construct of internalized homophobia from what the minority stress model describes as its outcomes and correlates. We suggested that three factors that some researchers have seen as overlapping with internalized homophobia—outness, community connectedness, and depressive symptoms—should be seen as distinct constructs. Our results demonstrate the utility of the minority stress model in delineating the relationships among these constructs.

As hypothesized, internalized homophobia was associated with greater relationship problems among all participants and among coupled individuals, specifically. These findings are consistent with previous research that has shown a negative relationship between internalized homophobia and relationship quality (e.g., Meyer, 1995; Meyer & Dean, 1998; Otis et al., 2006; Shildo, 1994; Szymanski et al, 2001). However, we also showed that outness, community connectedness, and depressive symptoms are important to consider as factors independent of internalized homophobia.

As hypothesized, the direct effects of internalized homophobia significantly attenuated when we accounted for the mediating role of depression, suggesting that internalized homophobia leads to relationship problems primarily by increasing depressive symptoms. This is important to consider in interpreting the results from previous studies that demonstrated no effect of internalized homophobia on indicators of relationship quality controlling for psychological well-being (Biss & Horne, 2004). By theorizing and analyzing internalized homophobia and its mental health outcomes at the same level, researchers misrepresent the nature of the relationship between the two constructs, obscuring the meditational role of mental health outcomes in the association between internalized homophobia and relationship quality.

Outness had a strong negative relationship with internalized homophobia but it was not significantly associated with indicators of relationship quality among all participants or among the subgroup of coupled participants. This indicates that although internalized homophobia and outness are related constructs they are not synonymous with one another. It is internalized homophobia, not outness, that has an impact on relationship quality among LGB individuals. Although we had no specific hypothesis about the relationship between outness and relationship problems, this finding is not surprising given recent calls by psychologists to avoid seeing outness as an indication of internalized homophobia (Eliason & Schope, 2007) and research that showed that internalized homophobia was negatively related to relationship quality, but outness was not (Balsam & Szymanski, 2005).

Similarly, although internalized homophobia was significantly correlated with community connectedness, it impeded relationship quality independent of connectedness. Unlike outness, community connectedness was associated with relationship quality for participants in the study. In predicting relationship strain among coupled participants, greater affiliation was associated with increased relationship strain. We had no specific hypotheses about the relationship between community connectedness and relationship strain, but we find this result intriguing and deserving of further research. One possible explanation for this finding is that individuals who are highly connected with the LGB community may be less invested in their relationships with their partners, which may make their partners feel less valued or neglected. It is also possible that individuals who are experiencing problems in their relationships turn to the LGB community for support and sanctuary. In sum, our results show that depressive symptoms, outness, and community connectedness are separate, though related, constructs that have unique roles in the experiences of LGB individuals.

Although our results demonstrate that depressive symptoms seem to completely mediate the relationship between internalized homophobia and relationship quality, especially among coupled participants, it is important not to overlook the remaining direct association between internalized homophobia and relationship quality. Although it was not statistically significant, the addition of this association to the models demonstrated improvement that approached statistical significance. Kline (2004) has noted that significance testing should not be the only consideration in reviewing results so that not only Type I but also Type II errors (not recognizing a relationship because of lack of power) would be minimized. Thus the direct effect of internalized homophobia on relationship problems may well have clinical significance and deserves further study. The careful clinician working with LGB individuals should be cognizant that even after effectively treating depressive symptoms, underlying internalized homophobia needs to be addressed. As Gonsiorek (1988) noted, internalized homophobia can often persist and continue to affect LGB lives after individuals have successfully come out and have found positive connections with other LGB individuals. Working with an LGB client experiencing problems in interpersonal relationships, clinicians should pay careful attention to internalized homophobia, even if the individual has come out to important others and demonstrates positive participation in the LGB community.

Limitations

Several study limitations must be kept in mind when interpreting our findings. As with any cross-sectional study, causal claims cannot be drawn from the correlational data we present despite our presentation of the data in structural equation models that suggest causality. Furthermore, there are plausible alternative models that may fit the data as well or better than the models we tested. That being said, we believe that a causal role for internalized homophobia as specified by our model is the most parsimonious based on existing clinical and theoretical writings and empirical findings.

Some measures used in the study demonstrated less than desirable reliability and have limited information regarding validity among LGB populations (i.e., loneliness, relationship strain, and positive relations with others). In some cases, low reliability may have lead to an underestimation of the relationships between constructs assessed in our study. Future studies should work to develop and validate measures of relationship quality among LGB individuals and incorporate additional, previously validated, measures of related constructs in order to address this limitation.

Although our study suggests that IHP is a significant source of relationship problems among LGB individuals, it was not meant to assess the full spectrum of factors that may affect relationship quality or how such factors may interact with one another. To understand relationship quality, it is important to consider many other factors such as differing commitment levels, disapproval from family and friends, and other stressors.

Also, our data was limited to the individual level. We therefore could not study factors that can only be observed in the dyadic level (Mohr & Fassinger, 2006, Otis et al., 2006). Perhaps more important is that the measures used as indicators of relationship quality do not provide an exhaustive representation of the construct (e.g., we did not investigate relationship satisfaction among coupled participants). Future studies should include more comprehensive measures to test relationship quality and tap more domains of the general intimacy construct, especially outside of the context of romantic relationships, as problems in relationships generally defined (e.g., friendships and familial relationships) have rarely been the focus of such studies.

An important strength of the current study is that we examined the experiences of diverse populations of LGB individuals. This is an advantage over studies that typically investigate whites only, thereby improving the external validity of our findings. However, sampling in the community limits how much can be generalized to clinical populations. Although we believe that the results are important for understanding treatment strategies, counselors should assess these findings critically, recognizing that the men and women in the sample were contacted using community sampling strategies and they may be significantly different from a clinical population. For example, they may be more out, higher in community connectedness, lower on internalized homophobia, and have fewer depressive symptoms than a clinical sample.

Conclusions

It is important to recognize that our conceptualization of internalized homophobia locates it as a social stressor that is related to negative social stigma surrounding LGB lives prevalent in society today (Meyer, 2003a). Like others, we wish to caution against conceptualizing internalized homophobia as a trait that is internal to the person. We view the term internalized homophobia as relating to the process whereby prevailing heterosexism becomes applied to the self (Russell & Bohan, 2006). Further, the use of the term “internalized homophobia” has been debated. Some writers note that it incorrectly suggests a phobic reaction, obscuring the focus on the external stressor. For this and other reasons writers have suggested alternative terms, such as “internalized heterosexism,” “internalized sexual prejudice,” and “ internalized sexual stigma.” However, attempts at alternative terminology have yet to prove superior to internalized homophobia (Herek, 2004). Until a consensus emerges we prefer the term used since the early development of the concept (Malyon, 1982).

Our investigation has special relevance given the media, political, and governmental attention that is being paid to same-sex relationships at present in the form of anti-same-sex marriage campaigns. Although LGB people are gaining rights they did not have previously and the social stigma surrounding LGB lives has declined (Savin-Williams, 2005), negative representations of LGB intimate relationships remain. The persistence of social stigma surrounding LGB people and intimacy remains a significant challenge to public and mental health practitioners and researchers working with LGB populations. Riggle, Thomas, and Rostosky (2005) noted that the current debate on marriage rights creates a “majority tyranny” that is on its own psychologically harmful to LGB individuals. The discourse of opponents of marriage rights devalues relationships of LGB people and reaffirms what Meyer and Dean (1998) called “heterosexist opportunity structures” that privilege heterosexual relationships and discourage same-sex relationships. Notions that LGB individuals, gay men in particular, are incapable of intimacy and long-term relationships and are likely to die alone without family are among the most fundamental stereotypes of LGB people (Meyer & Dean, 1998). Internalization of such societal discourse into one’s self-concept as an LGB individual likely exacerbates the negative effect of internalized homophobia on relationship quality.

Stressors related to heterosexism and its ill effects, including internalized homophobia, must be combated at all levels (Ouellette, 1998). But in addition to efforts being made to combat social stigma at the macrosocial level, attention needs to be paid to helping LGB individuals negotiate this stigma and develop positive self-concepts in the face of it through counseling and preventive services. Good guidelines for effective treatment of LGB individuals have been developed (APA Division 44/Committee on Lesbian, Gay, & Bisexual Concerns Task Force, 2000). Our study suggests that efforts targeted at reducing internalized homophobia and its effects on LGB lives need to be specific, and extend beyond helping LGB people come out of the closet and establish ties with the larger community, as models of identity development may suggest. Counselors working with LGB clients who struggle with internalized homophobia should focus on helping them develop more positive self-regard, combat resultant depressive symptoms, and develop healthy social support networks and intimate relationships (Gonsiorek, 1988).

Table 3.

Fit statistics for Structural Equation Models Predicting Relationship Problems and Relationship Strain

	Fit Statistics
Primary Outcome and Model	χ2	Df	P	χ2/df	RMSEA	90% CI	SRMR	CFI
Relationship Problems for All Participants (N = 396)
Model 1a - Direct Effect (IHP only)	20.79	8	.04	2.60	.064	.031, .098	.041	.982
Model 2a - Direct Effect (IHP & Correlates)	73.59	48	.10	1.53	.037	.018, .053	.035	.983
Model 3a - Complete Mediation	136.67	83	.03	1.65	.040	.028, .052	.038	.979
Model 4a - Hypothesized Model (Partial Mediation)	133.87	82	.03	1.63	.040	.027, .052	.035	.980
Relationship Strain for Coupled Participants (n = 184)
Model 1b - Direct Effect (IHP only)	7.60	8	.60	0.95	.000	.000, .084	.023	.999
Model 2b - Direct Effect (IHP & Correlates)	52.68	48	.56	1.10	.023	.000, .055	.041	.994
Model 3b - Complete Mediation	103.91	83	.39	1.25	.037	.000, .058	.053	.982
Model 4b - Hypothesized Model (Partial Mediation)	100.54	82	.43	1.23	.035	.000, .057	.049	.984

RMSEA = Root mean square error of approximation; 90% CI = Upper and lower bounds for the RMSEA; SRMR = Standardized root mean square residual; CFI = Comparative Fit Index. Probability values are adjusted for multivariate non-normality using Bollen-Stine corrections from bootstrapping with 10,000 samples.