https://orcid.org/0000-0001-7448-1191
Abstract
Cohabitation and marriage are critical milestones during the transition to adulthood; however, there is limited research on the timing of young adults’ first same-sex unions. There is some evidence that same-sex unions may be delayed, particularly for men. Further, formation of both same- and different-sex dating relationships, common among sexual minority young adults, may also extend to cohabitation and marriage. We used the National Longitudinal Study of Adolescent to Adult Health to predict the timing of a first romantic union, defined as a cohabitation or marriage, among sexual minority young adults. We then distinguished between women and men and the timing of a different-sex versus a same-sex union. Compared with heterosexual young adults, lesbian and gay young adults entered a union at later ages (driven by men), whereas bisexual young adults entered a union at younger ages (driven by women). Lesbian and gay young adults who entered a first union with a same-sex partner did so at later ages than those who entered a first union with a different-sex partner. Results suggest that patterns of sexual minority dating relationship formation might extend to unions.
Introduction
Romantic unions, or cohabitation and marriage, are critical for identity and intimacy development during the transition to adulthood (Arnett 2000), and “off-time” unions (those not occurring at the typical stage in life) can lead to worse health outcomes (Beckmeyer 2015) and later relationship difficulties (e.g., Collins et al. 2009). About 60% of young adults cohabit (Manning 2013; Mernitz 2018), and 74% of women and 82% of men enter cohabitation, not marriage, as their first union type (Manning et al. 2014). The age of first cohabitation has remained relatively stable over time: on average, women enter earlier into cohabitation, around age 22, whereas men enter cohabitation around age 24 (Manning et al. 2014). In comparison, the average age at first marriage has steadily increased: in the early 2000s, women got married around age 25 and men married around age 27, and by the 2010s, women married around age 27 and men married around age 30 (U.S. Census Bureau 2023a).
Research on the timing of a romantic union rarely considers sexual minority young adults’ union formation, which may differ from these broader union formation trends. Sexual minority young adults who enter cohabitation or marriage face discrimination and victimization regarding their union (Frost 2011) and may receive less acceptance of, and support for, their union from family and peers (Diamond et al. 1999; Macapagal et al. 2015). In light of this, sexual minority young adults may be discouraged from entering unions in young adulthood, which may delay union formation for this population. The number of sexual minority young adults has grown over recent decades, with about one in five Generation Z young adults identifying as lesbian, gay, bisexual, or transgender (Jones 2022). Examining longitudinal relationship transitions among this population has important implications for understanding broader developmental patterns for a growing demographic that continues to experience marginalization.
Sexual minority young adults may also form different-sex unions to minimize stigma, solidify their sexual identity, or disguise their same-sex attraction from others (Furman 2002; Glover et al. 2009) or, in the case of bisexual young adults, because different-sex union formation is consistent with their sexual orientation. Prior research has documented entrance into different- and same-sex dating relationships among sexual minority youth (Bauermeister et al. 2010; Russell and Consolacion 2003). Yet, cohabitation and marriage require significantly more investment than dating relationships (e.g., Huang et al. 2011), which may discourage sexual minority young adults from entering both same- and different-sex unions. Conversely, recent social changes, such as the growing acceptance of same-sex unions (Doherty et al. 2017) and the increase in the number of contemporary youth, especially young women, who self-identify their sexuality as bisexual or fluid (Galupo et al. 2015), might contribute to a growing prevalence of youth entering both same- and different-sex unions given that youth might have greater access to and desire for both same- and different-sex partners, and fewer barriers to entering same-sex unions.
We employ the National Longitudinal Study of Adolescent to Adult Health (Add Health) to explore the timing of a first romantic union for sexual minority young adults. We predict the timing of a first union for lesbian and gay or bisexual young adults and examine whether the timing of a first union changes depending on whether the partner is a same-sex or different-sex partner. Further, because there are well-established gender differences in the timing of different-sex union formation in general populations (e.g., Manning et al. 2014) and documented gender differences in entrance into same-sex dating relationships (Bauermeister et al. 2010), we also explore the timing of union formation by gender.
Timing of Union Formation for Sexual Minority Young Adults
Measuring the prevalence of U.S. same-sex couples is now possible owing to the addition of questions that allow researchers to identify same-sex partners in nationally representative surveys. Using data from the 2010 U.S. Census, Gates and Cooke (2011) estimated that there were slightly less than 650,000 same-sex couples living together in the United States; 80% of these were cohabiting and 20% were married. Estimates from the 2011 American Community Survey are slightly lower, suggesting that there were approximately 605,000 same-sex couples in the United States (1% of all coupled households) (Vespa et al. 2013). Yet, the nationwide legalization of same-sex marriage in tandem with more favorable societal attitudes toward same-sex unions, particularly among young adults compared with older adults (Doherty et al. 2017), has contributed to increased numbers of same-sex legal unions (Gates and Brown 2015). Contemporary data from the 2021 American Community Survey suggest that there are now more than 1.2 million same-sex couples in the United States, and 59% of these couples are married (U.S. Census Bureau 2023b). Despite the growing numbers of same-sex unions and better data on their prevalence, there is only descriptive evidence to suggest that sexual minority young adults delay union formation (Prince et al. 2020).
From a minority stress theory perspective, union formation may be delayed among sexual minority young adults, particularly for lesbian and gay individuals. According to this theory, sexual minority attraction, behavior, and identity are stigmatized, and sexual minority individuals experience unique stressors in their daily lives not experienced by heterosexual populations, such as discrimination and prejudice (Meyer 2003). These stressors may directly or indirectly contribute to delayed union formation. For instance, same-sex couples in our data were unable to legally marry in many U.S. states during a time when their heterosexual peers were beginning to marry (Obergefell v. Hodges 2015) and may have come of age believing that they would not be able to legally marry. Not only did this unequal access impact their union formation process for marriage, it also might have influenced the timing of a first cohabitation. Because the ability to marry has implications for relationship formation and stability (Carpenter 2020; Chen and van Ours 2020), sexual minority young adults in our sample may view cohabitation differently—more akin to marriage—than their heterosexual peers, contributing to delayed union formation.
Minority stressors may also impact young adults’ partner availability; prior to adulthood, young people may not feel comfortable or safe revealing their sexual minority status and thus have reduced access to same-sex partners and limited means to extend their in-person dating markets to find them (Diamond et al. 1999). These barriers to union formation are likely more pronounced for lesbian and gay young adults because bisexual young adults partner with different-sex partners. Although contemporary sexual minority young adults may have greater partner availability owing to increased acceptance of same-sex relationships (Doherty et al. 2017) and the advent of online dating, a strategy employed by sexual minority people to extend their dating markets (Korchmaros et al. 2015; Rosenfeld 2018), the lower prevalence of sexual minority youth relative to heterosexual youth may make finding romantic partners difficult. Thus, sexual minority union formation may be particularly delayed for lesbian and gay young adults, even in situations or places that are more friendly toward sexual minority people.
Gender Differences
Research on gender differences in union formation finds that women tend to form unions earlier than men (Manning et al. 2014), but it is unclear whether these findings would extend to sexual minority young adults’ union formation during the transition to adulthood. Traditional and evolving gender norms likely alter union formation patterns for both sexual minority women and men. For sexual minority men, engaging in same-sex behavior violates strict masculinity norms that expect men to be sexually interested only in women (Connell 1995; Russell and Consolacion 2003), and navigating these gender norms might result in delayed unions. Indeed, men report low prevalence rates for both same-sex romantic relationships (Bauermeister et al. 2010) and romantic unions (Gates 2014), even when they self-identify as sexual minorities. In these instances, sexual minority men may delay unions compared with sexual minority women because they do not want to face societal repercussions and status loss for going against gender norms.
Even bisexual men might delay union formation compared with bisexual women. In general, sexual fluidity is characterized as inherently feminine (Eisner 2013), and men tend to report more sexually exclusive (i.e., either 100% heterosexual or 100% homosexual) attraction and behavior (Diamond et al. 1999; Vrangalova and Savin-Williams 2012). Further, the proportion of men who report a bisexual identity has remained relatively low and stable over time (England et al. 2016). Even in instances where bisexual men enter unions with different-sex partners, which is consistent with their sexual identity, their union formation timing might mirror that for heterosexual men. Past studies on heterosexual populations document delayed union formation among men relative to women (e.g., Manning et al. 2014), suggesting that bisexual men might enter a union at older ages than bisexual women.
Sexual minority women might enter a union earlier than men owing to gendered socialization. Indeed, gender relational theory suggests that gender informs how men and women are socialized to behave and interact with one another in the context of a dating relationship or romantic union (Ferree 2010). Women are socialized to be relational and desire dating relationships and unions, whereas men are socialized to initiate dating relationships and progress these relationships into unions (Orth and Rosenfeld 2018; Sassler and Miller 2011). Within different-sex relationships, these gendered socialization scripts can create tension: men are given the power to initiate and progress relationships, but have less desire to do so, and women have the desire but are less willing to initiate discussions about forming relationships. Although women are more likely to initiate romantic unions than dating relationships, they often do so indirectly and men ultimately determine whether, and when, a union is formed (Sassler and Miller 2011). Sexual minority women's union formation could differ from this established script, although the socialization process occurs similarly. In these instances, lesbian women who are socialized to be more relational might be more likely to enter a union faster than men because both partners may equally desire, and lack barriers to, union formation.
Unlike sexual minority men, sexual minority women may face less overt stigma regarding entrance into unions because feminine gender roles allow for more flexibility in relationships (Diamond 2002). Indeed, women are more sexually fluid in sexual attraction, behavior, and identity than men (Katz-Wise 2015), and this fluidity is becoming increasingly more common among women (Galupo et al. 2015), suggesting that women may have more options for same-sex partners than men. Contemporary women are more likely to report a bisexual identity and report ever having sex with women (alone, or in addition to having male sexual partners) than women were in the past (England et al. 2016). Women are also more likely than men to enter same-sex dating relationships (Bauermeister et al. 2010) and romantic unions (Gates 2014). Despite these rapid historical changes in sexual attraction and behavior, women with histories of both same- and different-sex behavior or a bisexual identity are more likely to enter unions with a different-sex partner (Herek et al. 2010), perhaps because different-sex partners are more available than same-sex partners. Taken together, these findings suggest that lesbian or bisexual women may be more likely to enter unions earlier than sexual minority men.
Entrance Into Same- Versus Different-Sex Unions
Entrance into both same- and different-sex dating and sexual relationships is uncommon, but not nonexistent, among youth, particularly during the transition to adulthood (Bauermeister et al. 2010; Fish and Pasley 2015; Russell and Consolacion 2003). Growing acceptance toward same-sex unions (Doherty et al. 2017) and greater diversity in young adult sexual identities (Galupo et al. 2015) could create environments where youth increasingly engage in both same- and different-sex unions. Although the motivation for entering both dating relationship types is unknown, scholars have speculated that these motivations can be both beneficial (e.g., for identity development) and maladaptive (e.g., for concealment of same-sex attraction). For instance, entrance into both same- and different-sex dating relationships could reflect an attraction to both men and women (Glover et al. 2009) or aid in the development of a sexual identity (Furman 2002; Morgan 2013). At the same time, different-sex dating relationships could allow youth to disguise their same-sex attraction from others (Furman 2002) or reduce stigma regarding their sexual identity (Glover et al. 2009).
Romantic unions, however, are notably different from dating relationships in that they require significantly more investment. These investments can be tangible, such as sharing the costs of living together (Huang et al. 2011), or intangible, such as investing more psychologically into the romantic union (Finkel et al. 2014). These greater investments may discourage sexual minority young adults from entering a different-sex union if youth identify as lesbian or gay, but not if they identify as bisexual. Additionally, because first romantic unions occur after first dating relationships, sexual minority young adults who have used both different- and same-sex dating relationships to develop or solidify an identity (Collins 2003; Morgan 2013) might not need to enter different-sex unions for these purposes.
Confounding Variables
Among sexual minority youth, the timing of a first union likely differs by demographic, socioeconomic, and contextual factors. Different-sex union formation has been broadly linked to both family structure and race and ethnicity in nationally representative populations (Lichter et al. 2010; Sandberg-Thoma and Kamp Dush 2014), and same-sex couples are more likely to be interracial and more educated (Gates 2014), suggesting that there may be differences in union formation for these groups. Further, same-sex couples are also more likely to live in more tolerant U.S. regions (e.g., those where same-sex marriage was legalized early; Gates 2009), possibly contributing to earlier union formation in these areas because of decreased stigma (Riggle et al. 2017) and greater access to potential same-sex partners (Diamond et al. 1999).
Current Study
We use the National Longitudinal Study of Adolescent to Adult Health to examine timing of a first union among sexual minority young adults who identify as lesbian/gay or bisexual. In addition, we explore whether union formation timing differs for men and women and for those who form a union with a same-sex or a different-sex partner. We propose the following hypotheses:
Hypothesis 1: Lesbian or gay young adults will enter a first union at older ages relative to heterosexual young adults; bisexual young adults will enter a first union at similar ages relative to heterosexual young adults.
Hypothesis 2: Sexual minority women will enter any union, regardless of sexual orientation or partner gender, at younger ages than men.
Hypothesis 3: Sexual minority adults who enter a union will do so at younger ages when partnering with a different-sex partner than a same-sex partner; this association will be more pronounced for bisexual young adults.
Methods
Data
Add Health is a nationally representative panel study of 20,745 U.S. youth designed to examine how individual, familial, and contextual characteristics influence health and well-being (Harris 2013). Add Health recruited adolescents nationwide from 80 high schools and 52 middle schools; adolescents were in grades 7 through 12 during the 1994–1995 school year (Wave I). After the initial Wave I, youth were contacted for participation again in 1996 (Wave II), in 2001–2002 (Wave III; 18–26 years old), in 2008 (Wave IV; 24–32 years old), and in 2016–2018 (Wave V; 33–43 years old). We use data from Waves I, III, IV, and V.
We restrict the initial 20,745 participants at Wave I to those who completed Wave III, IV, or V (n = 18,378) and omit those who report a first union prior to age 16 (n = 18,268). We further restrict the sample to only those with valid cross-sectional weights at Wave IV (n = 14,690); we use Wave IV cross-sectional weights because most romantic union records come from Wave IV (54%). We conducted an attrition analysis predicting Wave IV attrition from Wave III sexual orientation and other characteristics (results not shown). We found no difference in attrition by sexual orientation status; however, women, those with more education at Wave III and an intact family in adolescence (Wave I), and those living outside the Northeast at Wave III were less likely to attrit. Older participants (at Wave III), racial minority participants, and those not employed at Wave III were more likely to attrit.
Missing information on other variables accounted for 6% of the data; we handled this missing data by using multivariate imputation using chained equations in Stata 16.1 (creating 25 imputed datasets). In this process, missing data are imputed through an iterative set of predictive models by treating each variable as a dependent variable and regressing all other variables onto it (Johnson and Young 2011). We tailored all models to the measurement level of each variable (e.g., dichotomous variables are estimated using logistic regression, and continuous variables are estimated using ordinary least-squares regression), which has been shown to produce reliable estimates (Johnson and Young 2011).
Measures
Age at First Union
To measure the beginning date of any cohabitation or marriage reported on or after age 16, we use data from the relationship histories provided at Waves III and IV and information on a current or most recent partnership at Wave V. At Waves III and IV, respondents were asked about their lifetime number of cohabitations and marriages, the month and year these unions began, and demographic information, such as the biological sex, of each partner. At Wave V, respondents were asked about their current or most recent relationships, including cohabitation or marriage, the month and year that union began, and the biological sex of that partner.
We identify the earliest cohabitation or direct marriage not preceded by cohabitation for each respondent and whether that union was with a same-sex or a different-sex partner, at each wave. For marriages following cohabitation, we use respondent's age at first cohabitation. If respondents did not report a first union by Wave III, we supplement their information from Wave IV. If there is no union record at Waves III and IV, we then use information from Wave V; 39% of relationship information comes from Wave III, 54% comes from Wave IV, and 7% comes from Wave V. Changing the sample construction to center on Wave IV does not affect the results.
Sexual Minority Status
Sexual minority status is measured by respondents’ answers to the question, “Please choose the description that best fits how you think about yourself,” which was asked at Waves III, IV, and V. Individuals are categorized as heterosexual if they exclusively identified as “100% heterosexual (straight)” or “mostly heterosexual (straight) but sometimes attracted to people of your own sex” across three waves. Those who reported a bisexual identity at any wave, and did not ever identify as lesbian/gay, are categorized as bisexual. Respondents who reported “100% homosexual (gay)” or “mostly homosexual but sometimes attracted to people of the opposite sex” at any wave are categorized as lesbian/gay. For individuals who reported a mix of bisexual and lesbian/gay identities (n = 83), we prioritize the identity reported at the last available wave and categorize them as lesbian/gay.
Demographic, Family, Socioeconomic, and Contextual Characteristics
Demographic characteristics included dichotomous indicators of respondent's gender and race and ethnicity (at Wave I). Gender was coded as 1 = female. Race and ethnicity were measured as “Black,” “Hispanic,” “Asian,” “other racial category,” or “White” (the reference category).
Family characteristics included growing up in an intact family in adolescence and parental education (both measured at Wave I). Respondents who reported having two biological or adoptive parents were coded as 1 = having an intact family. Parental education was coded as 1 = less than a high school degree, 2 = high school degree or GED, 3 = some college, and 4 = college degree or more.
Socioeconomic characteristics included dichotomous indicators of employment and respondent education. Respondent employment was coded as 1 = not currently employed (at Wave IV). Respondent's highest educational attainment was measured at Wave V as 0 = less than a high school degree, 1 = high school degree or GED, 2 = some college, 3 = college degree, and 4 = more than a college degree. Lastly, the census region where they resided at Wave IV was categorized as Midwest, South, West, or Northeast (the reference category).
Analytic Plan
We used Kaplan–Meier survival estimates and multivariate Cox proportional hazards models to test our hypotheses. Among this population, young adults were at risk for entering into a first union beginning at age 16; failure occurred at the age at which young adults entered a first union, regardless of whether the union was with a different- or a same-sex partner. Young adults were censored at their age at the Wave V interview if they did not form a union during our period of observation. We use Kaplan–Meier survival estimates to examine timing (in age) of a first union, either with a same- or a different-sex partner, for lesbian/gay, bisexual, and heterosexual young adults (Hypothesis 1). This estimate was supplemented with a multivariate Cox proportional hazards model predicting timing (in age) of a first union.
Gender differences in timing of a first union (Hypothesis 2) were estimated using Kaplan–Meier survival estimates and Cox proportional hazards models stratified by gender to predict timing among men and women; Wald tests were used to identify significant gender differences in union timing. To explore differences by partner composition among those who formed a union (Hypothesis 3), we estimated Cox regression models predicting timing of a first different-sex union or same-sex union, and Wald tests were used to identify significant differences between different-sex and same-sex unions.
Results
Descriptive Statistics
Descriptive statistics for the full sample of young adults and by sexual orientation are presented in Table 1. Youth were predominantly heterosexual and White, and were more likely to have had an intact family in adolescence and parents with some college education. The sample was split fairly evenly by gender. By adulthood, most had at least a high school degree, were currently employed, and were more likely to live in the South than in other U.S. regions. On average, young adults first entered a union around age 22, were more likely to cohabit than marry, and were more likely to partner with a different-sex partner than a same-sex partner in their first union. Among sexual minority young adults, lesbian/gay youth entered a (primarily cohabiting) union around age 23, whereas bisexual youth entered a (primarily cohabiting) union around age 21. Lesbian/gay young adults primarily partnered with a same-sex partner, while bisexual young adults primarily partnered with a different-sex partner.
We used chi-square and ANOVA tests to examine descriptive differences by sexual orientation (Table 1) and gender (Table 2). Lesbian/gay young adults were older at the age of their first union and were also more likely to have a same-sex partner in that union than all other sexual orientations. Lesbian/gay and bisexual youth were more likely to cohabit in their first union compared with heterosexual youth. Gender differences suggested that women were more likely to report a bisexual orientation than men; there were no other gender differences by sexual orientation type. Men were also older at their first union (by one year) and were more likely to enter cohabitation as their first union type than were women.
Testing of Hypothesis 1
We used Kaplan–Meier survival estimates to examine timing of a first union for young adults. Consistent with Hypothesis 1, estimates for timing of a first union (Figure 1) suggested that, overall, lesbian and gay young adults formed a union at later ages than heterosexual youth. Inconsistent with Hypothesis 1, however, bisexual young adults formed a union earlier than heterosexual youth. Life table survival estimates are also presented in online appendix Table A1.
We used Cox proportional hazards regression models to confirm these survival estimates after controlling for demographic, socioeconomic, and contextual characteristics (Table 3). Compared with heterosexual youth, lesbian/gay young adults delayed a first union, whereas bisexual youth were more likely to enter a union at younger ages. Women entered a union at younger ages than men; young adults living in the South or Midwest also entered unions at younger ages than those living in the Northeast. Being a racial or ethnic minority, being older, and having greater educational attainment (except for some college) were associated with delayed first union formation. Young adults who grew up in an intact family in adolescence or who had parents with at least a college degree also delayed union formation.
Testing of Hypothesis 2
Kaplan–Meier survival estimates (Figure 2) suggested that, overall, men formed unions at later ages than women, particularly gay men. Bisexual women formed unions earlier than heterosexual and lesbian woman. Results from Cox proportional hazards regression models echoed these results: gay men delayed a first union compared with heterosexual men, whereas lesbian women did not significantly delay a union compared with heterosexual women (inconsistent with Hypothesis 2; see Table 3). Wald tests indicated that this gender difference was significant (z = 16.05, p < .001), suggesting that the delay in union formation for lesbian/gay young adults is driven by men. Yet, consistent with Hypothesis 2, bisexual women entered a union at younger ages than heterosexual women; the association between sexual orientation and union timing was not significant for bisexual men. Wald tests indicated that this difference was marginally significant (z = 3.67, p < .05), suggesting that early union entrance for bisexual young adults is likely driven by women.
For both genders, being older, Black, or Hispanic, or having at least a college degree, was associated with delayed union formation; having parents with a college degree or more was also associated with delayed union formation. Living in the Midwest or the South was associated with early union entrance for both men and women compared with living in the Northeast. Additionally, for women only, being from an intact family in adolescence was associated with delayed union formation. For men only, being Asian, not being employed, and having a high school education or some college were also associated with delayed union formation.
Testing of Hypothesis 3
Results from Cox proportional hazards regression models were partially consistent with Hypothesis 3, as they indicated that lesbian/gay young adults who entered a union with a different-sex partner did so at younger ages than heterosexual youth who entered a union with a different-sex partner (Table 4). Wald tests confirmed that lesbian/gay young adults also entered a different-sex union at younger ages than lesbian/gay young adults who entered a same-sex union (z = 5.76, p < .05). Inconsistent with Hypothesis 3, bisexual young adults did not differ in timing for unions with a different-sex or a same-sex partner.
Results also indicated that young adults who were not currently employed, who were living anywhere other than the Northeast, or who were women entered a first different-sex union at younger ages. Black and Hispanic young adults or those who reported an “other” racial category, older young adults, and those with more educational attainment entered a different-sex union at older ages. Young adults who grew up in an intact family in adolescence and whose parents had at least a college degree also delayed different-sex unions. For same-sex unions, women and those who were not currently employed entered a union at younger ages than men and those currently employed, respectively; no other covariates were significantly associated with same-sex unions.
Discussion
For different-sex couples, cohabitation and marital union formation has changed rapidly over time: most young adults now cohabit, often as their first union, and delay marriage until their late 20s or early 30s (Furstenberg 2011; Mernitz 2018; U.S. Census Bureau 2023a). Estimates suggest that there are more than 1.2 million same-sex couples in the United States (U.S. Census Bureau 2023b), but little is known about same-sex union formation. Further, sexual minority youth commonly enter same- and different-sex dating relationships (Bauermeister et al. 2010; Russell and Consolacion 2003), yet many studies do not consider the timing of unions and whether this timing differs if the union is with a different- or a same-sex partner. Consistent with our first hypothesis, we found that lesbian/gay young adults formed a union at older ages than heterosexual young adults; bisexual young adults formed a union earlier than heterosexual young adults (inconsistent with Hypothesis 1). Our findings suggest that lesbian/gay young adults might enter unions at older ages to solidify a sexual identity (Furman 2002; Morgan 2013) or to disguise same-sex attraction or reduce stigma surrounding a sexual minority identity (Glover et al. 2009). Further, our results showed that, for our smaller sample of lesbian/gay young adults who formed a first union with a different-sex partner, they did so at younger ages than heterosexual young adults. Taken together, these findings suggest that lesbian/gay young adults could be using different-sex unions for the same reasons they are delaying unions broadly—namely, for sexual identity development or to reduce societal stigma surrounding an identity.
Similarly, bisexual young adults entered a union at younger ages relative to heterosexual youth (inconsistent with Hypothesis 1), and bisexual young adults did not differ in timing for unions with a different-sex or a same-sex partner (inconsistent with Hypothesis 3). Bisexual young adults might form unions at younger ages than heterosexual youth for the same reasons that lesbian/gay young adults form different-sex unions; early union formation among bisexual youth could aid in sexual identity development (Furman 2002; Morgan 2013) or mask any same-sex attraction from others (Glover et al. 2009). Further, although different-sex partners may be more accessible than same-sex partners (Diamond et al. 1999), bisexual young adults do not differ in union timing by whether they had a different- or a same-sex partner. This finding might suggest that partner availability could be a factor in union timing for bisexual young adults. Yet, it could also be that bisexual youth who desire same-sex partners might have limited options for these partnerships, forming unions with different-sex partners instead.
Consistent with research on different-sex union formation (Manning et al 2014), union formation for sexual minority and heterosexual young adults was also gendered. Indeed, gay men entered a union at later ages than heterosexual men and lesbian women, but lesbian women did not significantly differ from heterosexual women in union timing (inconsistent with Hypothesis 2). Men, who are socialized to align with strict masculine gender and sexual norms, are subjected to greater stigma and stressors if they violate these norms (Ferree 2010). Further, because men are more likely than women to exhibit sexually exclusive attraction and behavior (Katz-Wise 2015), they might be unlikely to form a union with a different-sex partner to mask any same-sex attraction or behavior. Taken together, gay men could delay same-sex unions, and not enter different-sex unions, to disguise same-sex attraction or behavior from others (Glover et al. 2009) and, thus, avoid societal stigma and repercussions from entering that union.
Conversely, lesbian women may not differ from heterosexual women in union timing owing to economic gender inequality. Although women are more educated than men on average, they continue to earn less than men in the workforce (Blau and Kahn 2017). Because young adults are often motivated to enter romantic unions for economic purposes, such as to reduce the cost of living (Huang et al. 2011), women may have more economic incentive to enter a union, regardless of sexual identity. Although lesbian women experience stigma and minority stressors from their same-sex union, which are not experienced by heterosexual women (Frost 2020), the economic incentive to enter a union may offset any incentive to delay a union to avoid stigma or minority stress.
However, bisexual women entered a union at younger ages than heterosexual women and marginally earlier than bisexual men; union timing did not differ between bisexual men and heterosexual men (consistent with Hypothesis 2). Prior research on sexual minority youth milestones suggests that women are less likely to report exclusive same-sex attraction and behavior and are more likely than men to identify as sexually fluid (Galupo et al. 2015; Katz-Wise 2015), which was reflected in our bisexual group being predominantly women. Women are also more likely to engage in different-sex sexual experiences prior to having same-sex sexual experiences than are men (Maguen et al. 2002), and these sexual experiences often occur in the context of romantic dating relationships (Savin-Williams and Diamond 2000). Moreover, research has shown that women who experience sexual fluidity do not report same-sex attraction or begin to identify as bisexual until they are in their late 20s (Kaestle 2019), so entering a union at a younger age than their heterosexual counterparts might help inform one's sexual identity. Indeed, relationships often provide insights that shape a young adult's sense of self, including sexual orientation (Morgan 2013), and perhaps unions serve similar purposes.
Limitations
Although this study provides one of the first examinations of same-sex union formation, we acknowledge some limitations. First, we are unable to ascertain the motive for union formation, particularly with different-sex partners, among sexual minority young adults. These young adults may be delaying same-sex unions (but not different-sex unions) relative to heterosexual youth because doing so benefits them in some way, such as by aiding in the development of a sexual identity (Furman 2002; Morgan 2013), or for reasons that might be harmful to them, such as to reduce stigma regarding having same-sex partners (Glover et al. 2009) or because they do not have supportive contexts in which to establish these unions (Prince et al. 2020). Conversely, young adults may have competing motives to enter same-sex unions as all these reasons could apply to them, which could have implications for their unions and their personal development. Future qualitative research could disentangle these motivations for entering romantic unions and examine how these motivations differ by gender.
Second, there was not enough variability to distinguish between cohabitation and marriage. Like different-sex young adult couples who often enter cohabitation prior to marriage (Manning et al. 2014), young adults who enter first same-sex cohabitations might do so earlier than those who enter direct same-sex marriage not preceded by cohabitation. Because of the unequal access to same-sex marriage until recently, timing of same-sex marriage versus cohabitation could also be different from patterns found among young adults who form only different-sex unions. Because certain union types also have different implications for young adult health and well-being (Mernitz and Kamp Dush 2016), distinguishing between union types is a critical next step for future research. Third, we were unable to ascertain when these unions first began as relationships and, thus, do not know if observed delays in union formation were due to delayed relationship formation generally or to delayed progression within a relationship. Given that partners are significant sources of support (e.g., Umberson et al. 2013), understanding the nuances behind relationship and union formation is an important avenue for future research. Additionally, as scholars have documented interesting gender differences in relationship formation (i.e., men enter same-sex relationships earlier than women, but they do not differ in the progression into cohabitation; Orth and Rosenfield 2018 ), these formation processes are likely to be gendered.
Further, we were unable to capture unions that occurred between Waves IV and V if they were not a current or a most recent union at Wave V. Although most unions occurred prior to Wave V (93%), we recognize that we might be overestimating delays in union formation for certain groups if they experience shorter union durations than other groups. However, results were broadly replicated when omitting Wave V relationship information. Lastly, certain cell sizes, such as those for bisexual men and those who formed a first same-sex union, are small and results should be interpreted with caution. Add Health was designed to be nationally representative of the population at the time of data collection, including the relatively smaller number of sexual minority young adults at that time, and is the only population-based dataset with longitudinal relationship data among self-identified adolescent and young adult sexual minorities. Nonetheless, we acknowledge that interpreting the results within subgroups of sexual minorities (e.g., Black youth who formed a same-sex union) might have limited statistical power to detect differences, and an absence of a significant effect may be due to sample size. Focused convenience samples with specific subgroups, such as bisexual men or sexual minorities of color, may provide better insight into relational dynamics particular to these understudied populations.
Conclusions
Despite these limitations, our study is the first to examine timing of a first cohabitation or marital union among sexual minority youth, while also considering that sexual minority young adults might form unions with different-sex partners. Although union formation during the transition to adulthood has received significant scholarly attention (Furstenberg 2011), with implications for health and relationships (Beckmeyer 2015; Collins et al. 2009), union formation among sexual minority young adults has received significantly less, especially during the transition to adulthood. We found that lesbian/gay young adults formed a first union at later ages than heterosexual or bisexual young adults, whereas bisexual young adults formed a first union at younger ages than heterosexual youth. However, lesbian/gay youth who formed a union with a different-sex partner did so at younger ages relative to heterosexual youth and lesbian/gay youth who formed a same-sex union. Union timing was also gendered, such that gay men were driving the observed delays in union formation and bisexual women were driving earlier union entrance. Taken together, these findings highlight the importance of considering variation in union formation by partner gender and suggest that there might be different motivations for the timing of unions among sexual minority young adults. Future research should consider identifying the motivations for same-sex or different-sex union formation among sexual minority young adults as well as exploring the consequences of union formation.
Acknowledgments
This research was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) grants R03HD099417 (awarded to S. Mernitz) and P2CHD042849 (awarded to the Population Research Center at The University of Texas at Austin). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Add Health is directed by Robert A. Hummer and funded by the National Institute on Aging cooperative agreements U01AG071448 (R. Hummer) and U01AG071450 (A. Aiello and R. Hummer) at the University of North Carolina at Chapel Hill. Waves I–V data are from the Add Health Program Project, grant P01HD31921 (K. M. Harris) from NICHD, with cooperative funding from 23 other federal agencies and foundations. Add Health was designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris at the University of North Carolina at Chapel Hill.
