Monosexual and Nonmonosexual Women in Same-Sex Couples’ Relationship Quality During the First Five Years of Parenthood

Abstract

Research on relationship quality in same-sex couples has rarely focused on (a) couples who are parents, who likely experience additional stressors, or (b) couples in which partners differ in sexual identity. Insomuch as nonmonosexual women (i.e., women with non-exclusive sexual orientations) experience unique challenges due to monosexism, relationship quality may be influenced by whether partners share a monosexual or nonmonosexual identity. The current study is a longitudinal, dyadic analysis of 118 female parents within 63 same-sex couples whose relationship quality (relationship maintenance, conflict, love, ambivalence) was assessed at five time points across the first 5 years of adoptive parenthood. Monosexual women were those who identified as exclusively lesbian/gay (n = 68); nonmonosexual women were those who identified as mostly lesbian/gay, bisexual, queer, pansexual, or mostly heterosexual (n = 50). Analyses revealed both actor and partner effects on maintenance and conflict, such that nonmonosexual women reported more maintenance and conflict than monosexual women, and women with nonmonosexual partners reported more maintenance and conflict than women with monosexual partners. Depression was related to greater conflict and ambivalence and less love; internalized sexual stigma was related to greater conflict and ambivalence. In terms of change over time, maintenance and love declined whereas ambivalence increased during early parenthood.

Introduction

A body of research exists on same-sex couples’ relationship quality (Fingerhut & Peplau, 2013), but this work is limited by the fact that many studies use only one partner’s report of relationship quality (Frost & Meyer, 2009; Gaines et al., 2005), most studies are cross-sectional (Frost & Meyer, 2009; Gaines et al., 2005; Mohr & Fassinger, 2006) and, when longitudinal designs are used, they typically follow couples across only two time points (Goldberg & Sayer, 2006; Mohr & Daly, 2008). Furthermore, few studies have explored the relationship quality of same-sex couples who are parents (Bos, Knox, van Rijn-van Gelderen, & Gartrell, 2016; Goldberg & Sayer, 2006; Goldberg, Smith, & Kashy, 2010), which is significant in that becoming a parent introduces unique forms of stress (Canario & Figueiredo, 2016) and same-sex couples are increasingly becoming parents (Gates, 2013).

Also of note is that little work examines relationship quality in same-sex couples where partners do not share the same sexual identity. Differences in sexual orientation, attraction, or relationship history are not typically addressed in studies of same-sex couples; both partners are usually treated as “lesbian” or “gay,” even if they do not identify as such (Ross & Dobinson, 2013). Bisexual individuals and other nonmonosexual people in monogamous relationships are typically defined based on who they choose as a partner, rendering their personal sexual identities invisible (Hartman-Linck, 2014). Speaking to the lack of attention to this issue, research on “mixed orientation” relationships only explores relationships where one partner identifies as heterosexual and the other as lesbian, gay, bisexual, queer (LGBQ), or some other label within the sexual minority spectrum (Buxton, 2001; Kays, Yarhouse, & Ripley, 2014; Schwartz, 2012; see Hernandez, Schwenke, & Wilson, 2011 for a review). In turn, this work tends to focus on marriages where one partner “comes out” to their heterosexual partner (as opposed to marriages where both partners are aware that one partner identifies as non-heterosexual), and tends to address topics such as rates of dissolution after disclosure (Yarhouse, Gow, & Davis, 2009) and sexual intimacy (Kays et al., 2014). The notion that within same-sex couples, partners could differ in sexual identity, and be aware of this difference, is simply not acknowledged. As Vencill and Wiljamaa (2016) note, it is necessary to broaden the definition of mixed orientation relationships to include mixed orientation same-sex relationships specifically; currently, no research on this topic exists.

This study explores relationship quality (relationship maintenance, conflict, love, ambivalence) across five time points among same-sex female couples who are parents. We assess whether relationship quality outcomes differ by monosexual/nonmonosexual status, and whether they vary depending upon the combination of sexual identifications (both monosexual; both nonmonosexual; one monosexual, one nonmonosexual). Monosexual sexual identities are defined as exclusive heterosexual or gay/lesbian identities, which emphasize “polar” romantic or sexual attractions to one gender or sex; nonmonosexual sexual identities, such as bisexual, queer, mostly lesbian/gay, and mostly heterosexual, reflect non-exclusive attractions, or romantic or sexual attractions to more than one gender or sex (Flanders, Robinson, Legge, & Tarasoff, 2016). In our study, monosexual women (n = 68) are those who identify as exclusively lesbian/gay, and nonmonosexual women (n = 50) are those women who identify as anything else: namely, mostly lesbian/gay, mostly heterosexual, or bisexual, queer, or pansexual.

Several factors influenced our decisions to group participants as monosexual versus nonmonosexual. First, these terms are among the mostly widely utilized by scholars and practitioners (e.g., see Dyar, Feinstein, Schick, & Davila, 2017; Flanders, Tarasoff, Legge, Robinson, & Gos, 2017; Persson, Pfaus, & Ryder, 2015). Second, prior research has documented important distinctions in the experiences and outcomes of monosexual and nonmonosexual women (e.g., Persson et al., 2015), likely in part because of the unique stigmas that individuals with non-exclusive sexualities encounter in society at large and in their interpersonal relationships specifically (Flanders et al., 2016). Third, we organized participants in this way for methodological reasons: some reduction of sexual identity categories was necessary in order to accommodate a sophisticated and adequately powered analysis.

We acknowledge that recent scholarship has noted problematic aspects of the terms monosexual and nonmonosexual, whereby monosexuality is centered and nonmonosexuality is defined in opposition to/against monosexual (Flanders, 2017; Galupo, Mitchell, & Davis, 2015) and we note the existence of useful alternative terms to nonmonosexual, such as plurisexual (Galupo et al., 2015; Mitchell, Davis, & Galupo, 2015). We also acknowledge that nonmonosexual women themselves did not use this term to describe their sexual identities; however, within the larger category of nonmonosexual, specific sexual identities (e.g., bisexual, queer) were not endorsed by all women. Using these terms would privilege one identity over others and erase other identities, and, thus, we opted for a more general “umbrella” term.

Next we discuss relevant research on (a) relationship quality and well-being outcomes according to monosexual/nonmonosexual status, (b) predictors of relationship quality in same-sex couples, and (c) relationship quality over time among new parents.

Monosexual/nonmonosexual identity status and personal/relational well-being

As a group, sexual minority women have been found to report higher levels of mental health problems as compared to heterosexual women, which in large part is explained by sources of minority stress (Plöderl & Tremblay, 2015; Strutz, Herring, & Halpern, 2015), including victimization and internalized sexual stigma (e.g., homonogativity, binegativity; Feinstein, Goldfried, & Davila, 2012; Hatzenbuehler, McLaughlin, Keyes, & Hasin, 2010). When looking within sexual minority women, nonmonosexual women tend to report more mental health symptoms than monosexual women (Colledge, Hickson, Reid, & Weatherburn, 2015; Persson, Pfaus, & Ryder, 2015). These higher levels may reflect, in part, the impact of monosexism, where those who are nonmonosexual are punished for not conforming to dominant assumptions and norms surrounding sexuality (i.e., the notion that attraction to only one gender is possible), and, in turn, persecution within both gay and heterosexual communities (Hayfield, Clarke, & Halliwell, 2014). Nonmonosexual people who disclose their sexual identities may encounter stereotypes, held by LG and heterosexual persons, that characterize them as hypersexual, confused, or “going through a phase” (Flanders et al., 2016; Ross, Dobinson, & Eady, 2010). Nonmonosexual people also experience erasure in that their identities are not “visible” to others (i.e., they are assumed to be heterosexual or LG depending on their partner’s gender); and, if they are believed to be LG, they are additionally vulnerable to homophobic prejudice (Ross et al., 2010).

Qualitative research with bisexual and other nonmonosexual individuals has sometimes explored perceived challenges in romantic relationships, such as partners believing stereotypes of bisexual people as promiscuous or indecisive (Gustavson, 2009; Lahti, 2015; Ross et al., 2010). Nonmonosexuality may be experienced as stressful for monosexual partners, who are sometimes described as “critical” and “have[ing] problems” with their partners’ bisexual or queer identities (Gustavson, 2009, p. 422), perhaps reflecting broader tensions within the gay community (Hayfield et al., 2014), whereby “from [the] specific lesbian standpoint, the lesbian community is essential to identity and. . .bisexual women insert some uncertainties in a lesbian collective” (Gustavson, 2009, pp. 422–23). In her interviews with five bisexual women in long-term relationships, two of whom were partnered with women, Lahti (2015) observed that “it was possible for bisexuality to be talked about in the interviews with the female couples. . .as an identity or sexual orientation, but not as a desire, at least not toward a gender other than their partner’s. . .bisexual women drew on the discourse of romantic love and stressed that their sexual orientation was toward their partner” (p. 443). Some bisexual women may remain silent about their nonmonosexual identities to avoid tension, positioning themselves in solidarity with lesbian communities (Ault, 1996), whereas others may engage in bisexual activism (i.e., be vocal about their nonmonosexuality; Gustavson, 2009; Hartman-Linck, 2014), which could create conflict.

Such findings, taken together, suggest that couples in which one partner is nonmonosexual and one partner is monosexual might experience unique relationship challenges. For example, they may engage in higher levels of relationship maintenance (e.g., talking about and processing their relationship). They may also experience greater ambivalence. Supporting this possibility, Feinstein and colleagues (2014) found that monosexual (heterosexual and LG) survey respondents indicated less willingness than bisexual respondents to engage in romantic or sexual activities with bisexual partners; and, heterosexual and LG respondents were generally less willing to be in a relationship with a bisexual partner than they were to have sex with or to date one.

Thus, same-sex couples in which one or both partners are nonmonosexual may show differences in various relationship domains (i.e., maintenance, conflict, love, ambivalence). Yet little empirical evidence on the topic exists, and thus, our analysis of sexual identity status in relation to these outcomes is exploratory.

Predictors of relationship quality in same-sex couples

Same-sex couples’ relationship quality may be affected by aspects of minority stress (Meyer, 1995; Meyer & Dean, 1998). Especially harmful to same-sex couples is internalized sexual orientation stigma (Frost & Meyer, 2009; Mohr & Daly, 2008). LGB people who are ambivalent about their LGB identity, or who hold negative views about LGB people (e.g., they are not capable of intimacy or maintaining long-lasting relationships) may find it harder to bond with, commit to, or be satisfied with a relationship partner (Frost & Meyer, 2009; Mohr & Daly, 2008). Internalized sexual stigma has been linked to poor relationship quality in non-parent (Mohr & Fassinger, 2006) and parent (Tornello, Johnson, & O’Connor, 2013) same-sex couples. Longitudinal work has found that internalized sexual stigma is related to decreases in relationship satisfaction over time, an association that may be due to the adverse effects of internalized sexual stigma on emotion/affect and communication (Mohr & Daly, 2008). Thus we examine internalized sexual stigma as a substantive predictor of relationship quality, given its significance in the literature on same-sex couples’ relationship quality.

Prior work suggests that relationship variables (e.g., relationship duration) and individual demographic variables (e.g., education, income, age, race) may also predict relationship quality. Mohr and Fassinger (2006) found that relationship duration was positively related to relationship quality in same-sex couples. A study of gay male couples found a significant correlation between higher income and better relationship quality (Elizur & Mintzer, 2003). A study of same-sex couples who were parents found that younger parent age was associated with greater relationship quality (Tornello et al., 2013), and a study of non-parent same-sex couples found that lower education levels were related to higher reported intimacy (Kurdek, 1998). Regarding race, some work has documented unique forms of stress that interracial same-sex couples may encounter in their relationships (Rostosky et al., 2008), yet other research (Jeong & Horne, 2009) has not found differences in relationship quality based on race or racial match (interracial versus same-race). Also of note is that research on same-sex couples who were also adoptive parents found that adopting a non-infant (older) child was positively associated with risk of relationship dissolution (Goldberg & Garcia, 2015), suggesting the significance of child factors in relationship quality. Given these findings, we control for various demographic factors (i.e., relationship duration, income, age, education, race, and child age) in this study.

Relationship quality over time

As stated, little longitudinal work has examined same-sex couples’ relationship quality, particularly among parents. However, research on both non-parent (Kurdek, 1998, 2008) and parent (Goldberg & Sayer, 2006; Goldberg et al., 2010) same-sex couples has documented declines in relationship quality over time (e.g., increases in conflict; declines in love), including across the transition to parenthood (Goldberg & Sayer, 2006). Turning to the more robust literature on heterosexual parents, this work is relatively consistent in showing that relationship quality declines, on average, during the first few years of parenthood (Kurdek, 1993; Doss, Rhoades, Stanley, & Markman, 2009). Specifically, love declines, whereas conflict and ambivalence (i.e., relationship uncertainty) increase, on average (Doss et al., 2009; Lawrence, Cobb, Rothman, Rothman, & Bradbury, 2008). Relationship maintenance behaviors (e.g., communicating about one’s needs; discussing the quality of one’s relationship; Braiker & Kelley, 1979) also tend to decline across the transition (Dainton, 2007). Notably, though, some work has found that stability in relationship quality, not decline, is a more typical trajectory for new parents (Foran, Hahlweg, Kliem, & O’Leary, 2013).

Thus, in addition to examining levels of relationship quality, we also examine trajectories of relationship quality over time, to determine whether sexual identity status is related to these outcomes during early parenthood. And, given that among same-sex couples who are parents, mental health symptoms have emerged as predictors of declines in relationship quality (Goldberg & Sayer, 2006; Goldberg et al., 2010), and given the strong relationship between mental health and relationship quality in the parenting literature more broadly (Proulx, Helms, & Buehler, 2007), we examine depression as a predictor of relationship quality.

The current study

The current study is a longitudinal, dyadic analysis of 118 female parents within 63 same-sex couples (in eight couples, only one partner had data) whose relationship quality was assessed at five time points (3 months after adopting their first child, 1 year after, 2 years after, 3 years after, and 5 years after). Monosexual women were those who identified as exclusively lesbian or gay (n = 68); nonmonosexual women were those who identified as mostly lesbian/gay, bisexual, queer, or mostly heterosexual (n = 50). Couples varied in composition, with 30 women in nonmonosexual-nonmonosexual relationships (43.2%), 37 women in nonmonosexual-monosexual relationships (31.4%), and 51 women in monosexual-monosexual relationships (25.4%). In addition to examining the role of sexual identity status, we examined internalized sexual stigma and depression (a time varying covariate) as predictors.

Of note is that women who dissolved their relationships over the course of the study (i.e., the first five years of parenthood) could not be included in this sample (i.e., because they did not have measure of relationship quality at each time point). Predictors of relationship dissolution among the sample, and differences between couples who stayed together and split up, are discussed in Goldberg and Garcia (2015).

Method

Participant recruitment

To be included in the study, which was approved by the internal review board (IRB) at Clark University, same-sex couples had to be adopting their first child and both partners had to be first-time parents (see Goldberg & Garcia, 2015 for a full description of recruitment methods). Adoption agencies in the United States were asked to provide study information to clients who had not yet adopted. Census data were used to identify states with a high percentage of same-sex couples and effort was made to contact agencies in those states. Over 30 agencies provided information to clients, often in the form of a brochure that invited them to participate in a study of the transition to adoptive parenthood. Clients contacted the researcher for details.

Procedure

Members of each couple were interviewed separately over the telephone three months after they were placed with a child; they were also asked to (separately) complete questionnaires (T1). Members of each couple were also sent questionnaires to complete 1 year post-placement (T2). They were interviewed and completed questionnaires again 2 years post-placement (T3). Finally, they completed questionnaires 3 years post-placement (T4) and 5 years post-placement (T5). Participants were compensated for their participation at each time point.

Description of the sample

Sixty-eight (57.6%) of the women in the sample were monosexual (i.e., exclusively lesbian/gay), and 50 (42.4%) were nonmonosexual. Within the nonmonosexual group, 35 women (70.0% of the group) identified as “mostly lesbian/gay,” 10 women (20.0%) identified as bisexual, three (6.0%) identified as queer, one (2.0%) identified as pansexual, and one (2.0%) identified as “mostly heterosexual.” Within the monosexual group, all 68 identified as completely lesbian/gay. Couple types were as follows: both partners monosexual (41.3%, n = 26 couples), both partners nonmonosexual (25.4%, n = 21 couples) and one partner monosexual, one partner nonmonosexual (33.3%, n = 16 couples).

Regarding individual level variables (i.e., variables that varied for partners within the couple; Table 1), women’s average annual personal income was $58,106 (SD =$47,729, range: $0-$250,000) and their average education level was 4.45 (where 4 = bachelor’s degree; SD = 1.01, range: 2 = high school diploma to 6 = doctoral degree). Women’s average age at the time of the adoption was 39.42 (SD = 6.00, range: 27.85–56.53). The sample was 90.7% white. Multilevel modeling was used to examine potential differences in individual-level demographic variables by sexual identity status; no significant differences emerged (age and income were significantly correlated within couples, ICC = 0.33 and 0.29, p’s < .05, respectively). Fisher’s exact test was used to test for differences in race by sexual identity status due to the low expected cell count for nonmonosexual women of color; no significant differences were found.

Table 1.

Individual-Level Demographics by Sexual Identity Status

	Nonmonosexual (N = 50)	Monosexual (N = 68)	Total (N = 118)
	M (SD)	M (SD)	M (SD)
Age	38.57 (5.09)	40.03 (6.56)	39.42 (6.00)
Personal Income	$53,728.26 ($54,942.93)	$61,449.64 ($41,613.86)	$58,106.36 ($47,729.72)
Education Levela	4.44 (1.01)	4.46 (1.01)	4.45 (1.01)
Race (% white)	92.0%	89.7%	90.7%

Note.

^a1 = less than high school education, 2 = high school graduate, 3 = associate’s degree/some college, 4 = bachelor’s degree, 5 = master’s degree, 6 = Ph.D./J.D./M.D. None of these variables significantly differed by sexual identity status.

Regarding dyadic/family level variables (Table 2), the average family (combined) income for female same-sex couples was $112,525 (SD = $74,895, range: $0-$320,000). Average relationship duration was 7.60 years (SD = 3.82). A total of 50.8% of couples used private domestic adoption, 34.9% of couples used public domestic adoption, and 14.3% used international adoption. The average age of children at the time of adoption was 21.29 months (SD = 43.96 months; range: newborn-16 years). A total of 46.0% of participants adopted boys, 44.4% adopted girls, and 9.5% adopted mixed gender (boy/girl) siblings.¹ Of the participants’ children, 71.4% were of color and 27.0% were white; one child was missing race information.

Table 2.

Demographics by Couple Type

	Both Monosexual	Both Nonmonosexual	Mixed orientation	Total
	M (SD)	M (SD)	M (SD)	M (SD)
Family Income	$126,457.22 ($65,963.05)	$88,964.29 ($75,204.46)	$116,688.79 ($82,117.23)	$112,525.82 ($74,895.46)
Relationship Duration	7.32 (4.23)	9.06 (3.85)	6.82 (3.05)	7.60 (3.82)
Child Age at Adoption (in Months)	23.43 (52.77)	11.4 (24.35)	25.71 (43.56)	21.29 (43.96)
Child Age at Adoption (% newborn)	34.6%	53.3%	19.0%	33.19%
Child Race (% white)	30.8%	25.0%	23.8%	27.4%
Child Gender
Boy	53.8%	50.0%	33.3%	46.0%
Girl	38.5%	43.8%	52.4%	44.4%
Boy and girl	7.7%	6.3%	14.3%	9.5%
Adoption Type
Public Domestic	34.6%	31.3%	38.1%	34.9%
Private Domestic	53.8%	62.5%	38.1%	50.8%
International	11.5%	6.3%	23.8%	14.3%

Note: Using one-way ANOVA for continuous variables and chi-square tests for discrete variables, no significant differences by group emerged in any of the demographic variables.

Descriptive statistics for the four relationship outcomes and two substantive predictors (depression, internalized sexual stigma), by monosexual versus nonmonosexual status, appear in Table 3. Correlations among the relationship outcomes at each time point appear in Table 4. Of the 118 participants who had any relationship quality data—and thus were included in analyses—relationship quality was 6.7% missing at T2, 9.3% missing at T3, 16.10% missing at T4, and 30.5% missing at T5. Participants who did not provide relationship quality responses by T5 were not significantly different from those who did in age, p = .621, income, p = .517, or education level, p = .870. Nor were they different in race, p = .456 (Fisher’s exact), or sexual identity status. p = .612.

Table 3.

Descriptive Statistics for Outcome Variables by Sexual Identity Status and Time

		Nonmonosexual (N = 50)	Monosexual (N = 68)
		M (SD)	M (SD)
Love	T1	7.83 (0.76)	7.76 (0.85)
	T2	7.48 (0.89)	7.62 (0.81)
	T3	7.46 (0.88)	7.53 (1.05)
	T4	7.32 (0.95)	7.33 (1.13)
	T5	6.87 (1.58)	7.33 (1.28)
Conflict	T1*	4.21 (1.19)	3.51 (1.28)
	T2*	4.41 (1.30)	3.66 (1.36)
	T3	4.41 (1.44)	3.90 (1.19)
	T4*	4.34 (1.41)	3.83 (1.30)
	T5*	4.14 (1.23)	3.53 (1.17)
Ambivalence	T1	1.92 (0.97)	1.86 (0.90)
	T2	2.22 (1.21)	2.01 (1.19)
	T3	2.33 (1.46)	2.12 (1.16)
	T4	2.54 (1.69)	2.18 (1.37)
	T5+	2.64 (1.68)	2.04 (1.35)
Maintenance	T1+	6.18 (0.97)	5.60 (1.32)
	T2	5.99 (1.28)	5.52 (1.22)
	T3	6.06 (1.22)	5.56 (1.34)
	T4	5.67 (1.34)	5.29 (1.25)
	T5	5.40 (1.28)	5.34 (1.39)
Depression	T1	0.61 (0.51)	0.48 (0.37)
	T2**	0.73 (0.54)	0.48 (0.39)
	T3	0.60 (0.47)	0.56 (0.42)
	T4**	0.63 (0.51)	0.44 (0.35)
	T5	0.46 (0.36)	0.40 (0.34)
Internalized Stigma	T1	0.52 (0.72)	0.34 (0.40)

Note. Multilevel models at each time point were conducted to test for differences between women who were nonmonosexual and women who were monosexual and the stars next to the time label reflect the significance level for each test.

⁺p < .10,

^*p < .05,

^**p < .01. Standard deviations appear in the parentheses next to the mean.

Table 4.

Correlations among Outcome Variables at Each Time Point

Time	Variable	1.	2.	3.	4.
T1	1. Love	1
	2. Conflict	−0.42***	1
	3. Ambivalence	−0.59***	0.38***	1
	4. Maintenance	0.43***	0.04 .	−0.13	1
T2	1. Love	1
	2. Conflict	−0.61***	1
	3. Ambivalence	−0.69***	0.58***	1
	4. Maintenance	0.13 .	0.21* .	−0.02	1
T3	1. Love	1
	2. Conflict	−0.52***	1
	3. Ambivalence	−0.76***	0.57***	1
	4. Maintenance	0.20* .	0.16+ .	−0.001	1
T4	1. Love	1
	2. Conflict	−0.63***	1
	3. Ambivalence	−0.80***	0.72***	1
	4. Maintenance	0.36***	−0.03 .	−0.15	1
T5	1. Love	1
	2. Conflict	−0.43***	1
	3. Ambivalence	−0.72***	0.63***	1
	4. Maintenance	0.45***	−0.03 .	−0.30**	1

Note.

⁺p < .10,

^*p < .05,

^**p < .01,

^***p < .001

Measures

Outcomes

Relationship quality.

Relationship quality was assessed using the Relationship Questionnaire (Braiker & Kelley, 1979), which contains 4 subscales: relationship maintenance (5 items), conflict (5 items), love (10 items), and ambivalence (5 items). All four domains were treated as outcomes. Items are answered on a 9-point scale (1= not at all to 9 = very much). Sample items are: “How much do you tell your partner what you want or need from the relationship?” (maintenance), “How often do you and your partner argue?” (conflict), “To what extent do you have a sense of ‘belonging with your partner?’” (love), and “How ambivalent are you about continuing in the relationship with your partner?” (ambivalence). Cronbach’s alphas for maintenance ranged from 0.52 (at T2) to 0.70 across all 5 time points. Alphas for conflict ranged from 0.71 to 0.79. Alphas for love ranged from 0.83 to 0.93. Alphas for ambivalence ranged from 0.62 (at T1) to 0.86.

Predictors

Monosexual/nonmonosexual status.

Sexual identity was effects coded (monosexual = 1, nonmonosexual = −1).

Internalized sexual stigma.

Internalized sexual stigma was assessed at T1 with a 9-item measure developed by Martin and Dean (1988). Items such as “If someone offered me the chance to be completely heterosexual, I would accept the chance” were administered with a 5-point response scale, ranging from 1 = disagree strongly to 5 = agree strongly. This measure has good convergent validity and good internal consistency (Herek & Glunt, 1995). Higher mean scores indicate higher internalized sexual stigma. The alpha for the scale was .90.

Depressive symptoms.

Depression was measured at each time point using the Center for Epidemiologic Studies Depression Scale (Radloff, 1977), a Likert-type scale that assesses depressive symptoms (20 items). Items such as “I felt sad” were responded to on a 4-point scale, from 0 = rarely or none of the time to 3 = most or all of the time. The CES-D has established validity and good internal consistency. Alphas ranged from 0.88 to 0.91 across time points.

Controls

Relationship duration.

At T1, participants were asked, “How long have you been in a committed relationship with your partner?” Relationship duration was the length of time, in years, that each individual indicated they had been in their current relationship.

Age. Participants’ age, in years, at T1.

Race. Race was recoded as an indicator for white where white = 1 and participants of color (POC) = 0.²

Education.

Education was measured on a scale of 1–6 where 1 = less than high school education, 2 = high school diploma, 3 = associate’s degree/some college, 4 = bachelor’s degree, 5 = master’s degree, and 6 = PhD/MD/JD.

Income.

Participants’ T1 personal annual income, in dollars.

Child age.

Age was effects coded: newborn (1) versus older (0). This particular age grouping was chosen in light of prior working showing that, among female couples with adopted children, women who ultimately dissolved their relationships were more likely to have adopted a non-newborn child (Goldberg & Garcia, 2015; Goldberg, Moyer, Black, & Henry, 2015).

Analytic strategy

We used multilevel modeling to account for the nonindependence due to women nested within couples crossed with time (Kenny, Kashy, & Cook, 2006). The dyads in our sample are indistinguishable, that is, the two members of the dyad cannot be meaningfully differentiated on some variable (e.g., gender, sexual identity status). Although some women are monosexual and some are nonmonosexual, it is not the case that in every dyad one member is monosexual and the other member is nonmonosexual. Thus, to examine change over time in relationship quality, we use the dyadic growth curve model for indistinguishable dyads as described in Kashy, Donnellan, Burt, and McGue (2008). In addition, we used an Actor-Partner Interdependence Model (APIM; Kenny et al., 2006) approach in that we estimated the effect of one’s own monosexual status (actor effect), the effect of one’s partner’s monosexual status (partner effect), as well the effect of being the same sexual identity status or not (the actor-partner interaction) on relationship quality. We refer to this last effect as dyad type in the remainder of the paper.

Dyadic growth curve models are mixed linear models with estimates of both fixed and random effects. For the fixed effects, an intercept and a slope is estimated for each member within a dyad. For distinguishable dyads (e.g., a sample of mixed-gender couples; a sample of patients and their caregivers), it is possible to get separate estimates of intercepts and slopes for each type of member (e.g., men’s intercept, women’s intercept, men’s slope, women’s slope). For indistinguishable dyads, the two estimates for the intercept (i.e., person 1’s intercept and person 2’s intercept) and the two estimates for the slope (i.e., person 1’s slope and person 2’s slope) are pooled due to the arbitrary selection of who is person 1 and who is person 2.

In addition to these two fixed effects, the pooled intercept and the pooled slope, we estimated the fixed effects of actor’s sexual identity status, partner’s sexual identity status, dyad type, internalized sexual stigma (measured at T1 and grand-mean centered), depression (time-varying: measured at all 5 time points), the two-way interaction of internalized sexual stigma with actor’s sexual identity status, the two-way interaction of sexual stigma with partner’s sexual identity, the three-way interaction of sexual stigma and actor’s and partner’s sexual identity, three additional interactions to explore the differences in the slope by sexual identity status, and the control variables described above, for a total of 17 fixed effect estimates.³ Due to the small sample of women, 118 women measured over 5 times point (n = 590 measurements), and the relatively large number of parameters needed to estimate dyadic growth curve models, we adopted the following analysis strategy. All control variables were included in initial analyses (Full Models) and then trimmed if their coefficients were found to not significantly differ from zero for all four relationship quality outcomes, resulting in the final Models that are reported in Tables 5–8.

Table 5.

Final Model Fixed Effects Estimates for Relationship Maintenance

Variable	b	df	t	p
Intercept	5.93	78.50	36.86	<.001
Time (in years)	−0.10	41.70	−3.54	0.001
Yearly Depression	0.002	386.00	0.01	0.990
Internalized Sexual Stigma	−0.09	98.10	−0.38	0.706
Actor’s Sexual Identity Status	−0.22	107.00	−1.90	0.060
Partner’s Sexual Identity Status	−0.25	105.00	−2.20	0.030
Dyad Type	0.05	52.60	0.38	0.702
Intern Sex Stigma × Actor’s Sexual Identity Status	−0.10	90.00	−0.44	0.663
Intern Sex Stigma ×Partner’s Sex Identity Status	0.03	90.10	0.12	0.903
Intern Sex Stigma × Dyad Type	−0.01	98.50	−0.05	0.959
Time × Actor’s Sexual Identity Status	0.03	66.80	1.16	0.252
Time × Partner’s Sexual Identity Status	0.04	66.20	1.31	0.196
Time × Dyad Type	−0.01	42.40	−0.21	0.834

Note. Dyad type is the interaction of actor’s sexual identity status (monosexual or nonmonosexual) and partner’s sexual identity status. Sexual identity status was effects coded: monosexual = 1 and nonmonosexual = −1.

Table 8.

Final Model Fixed Effects Estimates for Relationship Ambivalence

Variable	b	df	t	p
Intercept	1.60	61.80	12.00	<.001
Time (in years)	0.15	31.30	3.73	<.001
Yearly Depression	0.67	299.00	5.76	<.001
Internalized Sexual Stigma	0.56	97.70	2.37	0.020
Actor’s Sexual Identity Status	0.02	86.30	0.21	0.832
Partner’s Sexual Identity Status	−0.08	84.20	−0.75	0.458
Dyad Type	0.09	40.70	0.81	0.425
Intern Sex Stigma × Actor’s Sexual Identity Status	0.36	88.70	1.54	0.127
Intern Sex Stigma × Partner’s Sexual Identity Status	−0.19	88.70	−0.82	0.413
Intern Sex Stigma × Dyad Type	−0.28	98.00	−1.17	0.245
Time × Actor’s Sexual Identity Status	−0.03	74.00	−0.82	0.416
Time × Partner’s Sexual Identity Status	−0.03	73.70	−0.83	0.411
Time × Dyad Type	−0.02	31.40	−0.57	0.572

Note. Dyad type is the interaction of actor’s monosexual status and partner’s monosexual status. Sexual identity status was effects coded: monosexual = 1 and nonmonosexual = −1.

In addition to these fixed effects, dyadic growth curve models with indistinguishable dyads also have seven random effects and an error variance. These seven random effects include: 1) the pooled variance of the intercepts, 2) the pooled variance of the slopes, 3) the covariance of the two intercepts, 4) the covariance of the two slopes, 5) the within-person covariance of the intercept and slope, 6) the between-person covariance of the intercept and slope, and 7) the covariance in dyad members’ residuals for each time point. SAS 9.4 with restricted maximum likelihood estimation and the Satterthwaite correction to degrees of freedom was used to fit all models.

Results

We first estimated four models that included the fixed effects of all predictor variables (described above) and control variables, as well as all seven random effects on each of the four relationship quality variables (i.e., maintenance, conflict, love, ambivalence). In these Full Models, none of the coefficients for any of the control variables reached statistical significance for any of the four outcome variables. Thus, relationship duration, age, race, education, personal income, and child age were all trimmed from the model. The Final Models for each of the four outcomes are now reported in turn. A complete report of the fixed effects estimates is contained in Tables 5–8 and Figure 1 depicts the trajectories over time for the four outcomes.

Figure 1.

The reported relationship maintenance over time by actor and partner monosexual status.

Relationship maintenance

Fixed effects.

There was a statistically significant effect of time on reports of relationship maintenance, b = −0.10, SE = 0.03, p = .001, indicating an overall negative slope, i.e., a decrease in women’s relationship maintenance reports by 0.11 points per year. Depressive symptoms that year were not significantly related to reports of relationship maintenance, b = 0.002, SE = 0.14, p = .990, nor was internalized sexual stigma (reported at T1), b = −0.09, SE = 0.24, p = .706.

There was a marginally significant effect of actor sexual identity status on relationship maintenance such that nonmonosexual women reported more maintenance behaviors than monosexual women, b = −0.22, SE = 0.12, p = .060. The partner effect of sexual identity status on maintenance was statistically significant, such that women whose partners were nonmonosexual also reported more maintenance behaviors, b = −0.25, SE = 0.11, p = .030. There was no significant effect of dyad type on maintenance, b = 0.05, SE = 0.14, p = .702—that is, it was not the case that the most maintenance was reported in monosexual-nonmonosexual dyads. The overall negative slope, i.e., the decrease in maintenance, did not differ across actor sexual identity status, partner sexual identity status, nor dyad type (p ranged from .196 to .834). There were also no significant interactions of sexual identity status and internalized sexual stigma (p ranged from .663 to .959).

Random effects.

There was significant variance in maintenance reports at T1 (intercept), ${\sigma }_{int}^2$ = 0.80, SE = 0.20, p < .001, as well as a marginally significant positive covariance between dyad members’ intercepts, ${\sigma }_{int1,int2}$ = 0.37, SE = 0.21, p = .075—i.e., women had varying levels of reported relationship maintenance at T1, and if a woman reported a higher level of maintenance then her partner also reported a higher level of maintenance. The covariance in yearly errors between dyad members was statistically significant, ${\sigma }_{e1,e2}$ = 0.17, SE = 0.07, p = .020, indicating that if a woman reported a particularly high level of maintenance in a specific year, her partner also reported a particularly high level of maintenance that year. No other random effects were statistically significant.

Conflict

Fixed effects.

In contrast to the findings for maintenance, there was no statistically significant change over time in conflict, b = −0.003, SE = 0.03, p = .929. There was a statistically significant positive association between depression reported in a given year and conflict reported in that same year, such that women who were more depressed reported higher levels of relational conflict, b = 0.90, SE = 0.13, p < .001. In addition, there was a statistically significant main effect of internalized sexual stigma on reports of conflict, b = 0.45, SE = 0.22, p = .045, such that higher internalized sexual stigma at T1 was related to higher levels of conflict overall.

There was a statistically significant effect of actor sexual identity status on conflict, such that nonmonosexual women reported more conflict in their relationships than monosexual women, b = −0.24, SE = 0.12, p = .045. The partner effect of sexual identity status on conflict was also statistically significant, such that women whose partners were nonmonosexual also reported more conflict, b = -0.26, SE = 0.12, p = .030. As with maintenance, there was no significant effect of dyad type on conflict, b = -0.13, SE = 0.14, p = .348. There were no significant interactions of sexual identity status and internalized sexual stigma (p ranged from .113 to .949).

Random effects.

There was significant variance in conflict reports at T1 (intercept), ${\sigma }_{int}^2$ = 0.82, SE = 0.21, p < .001, and the covariance in yearly errors between dyad members was statistically significant, ${\sigma }_{e1,e2}$ = 0.12, SE = 0.06, p = .048. No other random effects were statistically significant.

Love

Fixed effects.

A statistically significant effect of time on reports of love indicate that love decreased over the five time points by 0.16 points per year, b = −0.16, SE = 0.03, p < .001. Similar to the findings for conflict, there was also a significant effect of depression on love such that higher levels of depression were related to lower levels of love, b = −0.29, SE = 0.09, p < .001.

In contrast to the effects for maintenance and conflict, there was no effect of internalized sexual stigma on love, b = −0.003, SE = 0.14, p = .984, nor any statistically significant effects of sexual identity status: For the actor effect, b = 0.05 (ns), for the partner effect, b = −0.04 (ns), and for dyad type, b = 0.04 (ns). All interaction effects were not statistically different from zero.

Random effects.

Many of the random effects of love were statistically significant. As with maintenance and conflict, there was significant variance in reports of love at T1 (intercept), ${\sigma }_{int}^2$ = 0.20, SE = 0.08, p = .017. There was also significant variance in the slopes, the change in love, across women, ${\sigma }_{slope}^2$ = 0.02, SE = 0.01, p = .036, and a statistically significant covariance of the intercept and slope within-person, ${\sigma }_{int1,slope1}$ = 0.07, SE = 0.02, p < .001—i.e., if a woman was more in love with her partner at T1, her slope (change in love) also tended to be larger (less negative). There was also a positive covariance of the intercept and slope between-person ${\sigma }_{int1,slope2}$ = 0.05, SE = 0.02, p = .006—if one woman was more in love at the intercept, this was related to her partner having a more positive (less negative) slope (i.e., less decline in love). There was no significant covariance between dyad members’ slopes. The covariance between women in a couple’s intercepts for love was statistically significant and positive, ${\sigma }_{int1, int2}$ = 0.18, SE = 0.02, p = .041, and the covariance in yearly errors between dyad members was statistically significant, ${\sigma }_{e1,e2}$ = 0.17, SE = 0.04, p < .001. Random effects for love were converted into correlations and standard deviations which appear in Table 9.

Table 9.

Random Effect Estimates from the Indistinguishable Dyads Dyadic Growth Model for Love

	1.	2.	3.	4.
1. Person’s 1 Intercept	0.45*
2. Person’s 1 Slope	1***a	0.13*
3. Person’s 2 Intercept	0.87*	0.87**	0.45*
4. Person’s 2 Slope	0.87**	−0.21	1***a	0.13*

Note.

^aCorrelation has been fixed 1. The standard deviations for each random effect appear on the diagonal.

^*p < .05,

^**p < .01,

^***p < .001.

Ambivalence

Fixed effects.

There was a statistically significant positive slope for relationship ambivalence, b = 0.15, SE = 0.04, p < .001, such that ambivalence about the relationship increased over time by 0.15 points each year. As with conflict and love, depression was significantly associated with relationship ambivalence, b = 0.67, SE = 0.12, p < .001, such that more depression was related to more ambivalence.

Consistent with the findings for conflict, there was a statistically significant main effect of internalized sexual stigma, b = 0.56, SE = 0.24, p = .020, such that women who reported higher sexual stigma at T1 reported more relationship ambivalence. There were no significant effects of sexual identity status, p from .425 to .832, nor interactions of sexual stigma and sexual identity status, p from .127 to .413.

Random effects.

There was significant variance in women’s reports of ambivalence at T1 (intercept), ${\sigma }_{int}^2$ = 0.53, SE = 0.16, p < .001, and significant variance in the change over time in ambivalence reports, ${\sigma }_{slope}^2$ = 0.04, SE = 0.02, p = .045. The covariance in yearly errors between dyad members was also statistically significant, ${\sigma }_{e1,e2}$ = 0.31, SE = 0.06, p < .001. No other random effects were statistically significant.

Discussion

This study makes a number of empirical contributions. First, it expands our understanding of “mixed orientation” relationships, which have historically focused largely on heterosexual women married to gay/bisexual men (Hernandez et al., 2011). It thereby heeds the call of Vencill and Wiljamaa (2016) to expand our conceptualization of mixed orientation relationships to consider the reality of diverse sexual identities in the context of same-sex relationships. Second, it examines several dimensions of relationship quality over time, among female same-sex adoptive couples, thereby contributing to the literatures on same-sex couples’ relationship quality (e.g., Kurdek, 1998) and same-sex parents’ relationship quality (e.g., Goldberg & Sayer, 2006). Third, it demonstrates the application of a sophisticated statistical modeling technique, dyadic growth curve modeling with indistinguishable dyads (Kashy et al., 2008), to the study of relationship quality in same-sex couples.

To summarize the findings for change in relationship quality, maintenance and love decreased across the first several years of parenthood, and ambivalence increased. Conflict did not change significantly over time. That there were changes in most but not all relationship domains is consistent with prior research on relationship quality among same-sex couples (Goldberg & Sayer, 2006) and parents with young children (Doss et al., 2009; Kurdek, 1993). Yet the lack of change in conflict, in contrast to other domains, is important, as it points to the importance of assessing multiple domains of relationship quality.

Regarding our findings related to sexual identity status, we found that concordance versus discordance in status did not impact any aspect of relationship functioning. Thus, monosexual-nonmonosexual couples did not experience better or worse outcomes than nonmonosexual-nonmonosexual and monosexual-monosexual couples. However, sexual identity did play a role in relationship quality in that nonmonosexual women, and partners of nonmonosexual women, reported higher levels of maintenance and conflict. With no effect of dyad type, interestingly, there was no evidence that one’s partner’s nonmonosexual status affects a monosexual women’s relationship quality any differently than it affects a fellow nonmonosexual women’s relationship quality. Thus, it seems that for both women who identify as not completely lesbian/gay, and their partners, the reality of their non-exclusive attractions may ultimately impact certain relationship dynamics. Perhaps nonmonosexual women and their partners are aware of stereotypes related to nonmonosexuality (e.g., the notion that bisexual people are confused and/or less committed to relationships; Ross et al., 2010) and such awareness motivates behavioral efforts to maintain the relationship, as well as prompting more arguments (e.g., due to worries about how one’s own or one’s partner’s nonmonosexual status might threaten the relationship). Yet it is also possible that one partner’s nonmonosexuality functions to open up conversations about sexuality – which may lead to greater communication about one’s needs in the relationship, and the relationship in general, as well as to more conflicts and tensions (e.g., perhaps as a result of this greater openness and the potentially provocative discussions such openness promotes). Therapists should ideally assess both partners’ sexual self-identifications, and address (perhaps first individually, and then with both partners) whether, how, and how much the couple discusses their mutual sexual identities, and the nature and sequelae of such discussions.

Of note here are qualitative findings that some bisexual women partnered with women feel pressure to remain silent about their nonmonosexual identities (e.g., to avoid tension; Ault, 1996; Lahti, 2015). Perhaps, then, the high levels of both relationship maintenance and conflict that we observed among nonmonosexual women and partners of nonmonosexual women reflect the consequences of discussing these identities. Of course, they could also reflect women’s awareness of but lack of communication about these identities. Notably, though, the lack of association between sexual identity and ambivalence or love suggests that whatever is driving participants’ increased relational maintenance and conflict (i.e., awareness of stereotypes about bisexuality and/or non-exclusive attractions; discussions about sexuality) does not translate into relationship uncertainty or lack of affection, for nonmonosexual women or their partners. Thus, couples can maintain highly committed, loving relationships regardless of sexual identity status (Hernandez et al., 2011). Such findings have implications for clinicians, who should be sensitive to the possibility that both nonmonosexual clients and their partners may come to therapy holding beliefs that reflect monosexism (e.g., they may worry about their own or their partner’s ability to maintain long-lasting, fulfilling relationships; Ross et al., 2010; Wilde, 2014). In turn, therapists should not only address these beliefs (e.g., their origins and dynamics) but should be prepared to offer psychoeducation about the lack of evidence to support them.

In addition to the change in relationship quality over time and the increased maintenance and conflict reported by nonmonosexual women and their partners, we also found many interesting random effects, many of which assess interdependence in relationship quality within couples. Consistently we found interdependence between partners’ reports in each year—if one woman experienced high relationship quality, then her partner was also experiencing high relationship quality. The most interesting random effects were found for love, which represents a more affective component of relationship quality (whereas maintenance and conflict are more behavioral). There was a positive association between partners’ reports of love at the first assessment—but most interestingly, the more a woman reported loving her partner at the first assessment, the less her partner declined in love over time. There was also a similar within-person association—if a woman started the study reporting more love, she declined less over time (or, if she was lower on love to start with, then love declined faster).

Turning to the findings for our substantive predictors, internalized sexual stigma was related to higher ambivalence and conflict, i.e., the more negative dimensions of relationship quality. This set of findings extends a small literature documenting the deleterious effects of internalized sexual stigma on relationship outcomes (e.g., Frost & Meyer, 2009; Tornello et al., 2013). Regarding ambivalence, it appears that internalization of sexual stigma may prompt more feelings of doubt or uncertainty about the relationship, even in the context of shared parenting. Likewise, with regard to conflict, it may be that unresolved feelings about one’s sexuality prompts negative ways of dealing with challenges in the relationship—consistent with prior research showing a link between internalized sexual stigma and poor communication (Gaines et al., 2005). These findings have implications for therapists who work with same-sex couples, particularly couples at risk for relationship dissolution (i.e., due to high levels of ambivalence and conflict). Sensitively assessing and addressing one or both partners’ internalized sexual stigma could have beneficial effects on both individual and relational well-being. By decreasing internalized sexual stigma, commitment and healthier conflict negotiation may be enhanced (Zheng & Zheng, 2016).

We also found that higher levels of depression were related to greater conflict and ambivalence, and less love. This set of findings echoes prior work documenting associations between mental health and relationship quality (Proulx et al., 2007) and among same-sex parents specifically (Goldberg et al., 2010) and has important clinical implications. First, it suggests that improving mental health, at any point in the trajectory of a relationship, could improve relationship quality. Second, it suggests that although depression may negatively affect relationship quality, the behaviors needed to maintain relationships (i.e., maintenance behaviors) are not necessarily affected. This is promising news for interventionists who work with distressed couples.

Limitations and conclusions

There are a number of limitations of this study. First, we did not assess salience of sexual identity (i.e., the salience of women’s monosexual/nonmonosexual identity), which could be important to relationship quality, above and beyond sexual identity itself (King & Smith, 2004). Second, we only examined female couples, and so we cannot say anything about how our findings might generalize to male same-sex couples. Third, we did not examine women’s sexual identity labels at each time point; it is possible that some women changed (e.g., from monosexual to nonmonosexual, or vice versa) over the course of the first five years of parenthood, and this could be important to their experiences of relationship quality. Fourth, we did not include couples who split up in our sample; thus, our study only examines relationship quality changes among couples who ultimately stayed together—and thus the conclusion that can be drawn from our findings are limited. Fifth, participants who stayed in the study may have differed from those who dropped out; thus, our findings are further limited to the couples that chose to maintain participation.

Additionally, our sample was relatively affluent (with a combined income of over $100,000) and well-educated; thus, our findings may not be generalizable to female couples in other income and educational brackets. Also, the alphas for maintenance were relatively low, particularly at the second measurement point, and findings for this domain should be viewed with caution. In addition, our sample size restricted us from examining the array of specific identities within the larger category of “nonmonosexual” – and whether these identities had implications for relationship quality. Furthermore, we do not have qualitative data to help explain the role sexual identity played in these relationships (i.e., accounts of how sexual identity is or is not discussed among the couples, and how they perceive sexual identity as affecting relationship maintenance or conflict). Finally, we wish to acknowledge that our grouping variable to describe women’s sexual identities—i.e., monosexual versus nonmonosexual—is necessarily imperfect. We considered a variety of umbrella terms to capture women’s varying sexual identities and determined that this particular set of terms was the least problematic in terms of balancing practical utility, scientific integrity, and respect for the multiple identities reported by participants.

Despite these limitations, our study makes a contribution to several literatures, and holds implications for sex and relationship therapists who work with members of the LGBTQ community. Our findings suggest that sexual identity status may have implications for certain dimensions of relationships among female same-sex couples—namely, relationship maintenance and conflict—but are unrelated to commitment to the relationship and positive feelings toward one’s partner. Therapists who work with women who identify as nonmonosexual, and their partners, should be aware of these findings in order to gently and effectively interrogate monosexist beliefs and assumptions on the part of clients, as well as to help clients understand, discuss, and manage the implications of differing sexual identity statuses.

Table 6.

Final Model Fixed Effects Estimates for Relationship Conflict

Variable	b	df	t	p
Intercept	3.67	73.20	23.47	<.001
Time (in years)	−0.003	40.80	−0.09	0.929
Yearly Depression	0.90	401.00	6.70	<.001
Internalized Sexual Stigma	0.45	84.20	2.03	0.045
Actor’s Sexual Identity Status	−0.24	92.30	−2.04	0.045
Partner’s Sexual Identity Status	−0.26	90.40	−2.21	0.030
Dyad Type	−0.13	48.60	−0.95	0.348
Intern Sex Stigma × Actor’s Sexual Identity Status	0.22	88.30	1.01	0.314
Intern Sex Stigma × Partner’s Sexual Identity Status	−0.35	88.20	−1.60	0.113
Intern Sex Stigma × Dyad Type	0.01	84.50	0.06	0.949
Time × Actor’s Sexual Identity Status	0.01	84.40	0.39	0.694
Time × Partner’s Sexual Identity Status	0.05	83.80	1.65	0.102
Time × Dyad Type	0.002	41.20	0.06	0.956

Note. Dyad type is the interaction of actor’s sexual identity status and partner’s sexual identity status. Sexual identity status was effects coded: monosexual = 1 and nonmonosexual = −1.

Table 7.

Final Model Fixed Effects Estimates for Love

Variable	b	df	t	p
Intercept	7.94	66.90	77.01	<.001
Time (in years)	−0.16	42.10	−5.61	<.001
Yearly Depression	−0.29	284.00	−3.39	<.001
Internalized Sexual Stigma	0.00	80.60	−0.02	0.984
Actor’s Sexual Identity Status	0.05	80.30	0.84	0.401
Partner’s Sexual Identity Status	−0.04	77.70	−0.70	0.484
Dyad Type	−0.08	45.10	−0.93	0.360
Intern Sex Stigma × Actor’s Sexual Identity Status	0.04	66.20	0.29	0.775
Intern Sex Stigma × Partner’s Sex Identity Status	−0.19	66.70	−1.45	0.152
Intern Sex Stigma × Dyad Type	−0.11	82.80	−0.81	0.422
Time × Actor’s Sexual Identity Status	0.003	77.90	0.14	0.892
Time × Partner’s Sexual Identity Status	0.02	77.50	0.81	0.423
Time × Dyad Type	0.01	42.40	0.20	0.839

Note. Dyad type is the interaction of actor’s sexual identity status and partner’s sexual identity status. Sexual identity status was effects coded: monosexual = 1 and nonmonosexual = −1.