Since the 2015 U.S. Supreme Court ruling that marriages of same-sex couples are legal in all states in the union, federal surveys have adapted to the shifting legal climate and included new measures that more directly identify same-sex and different-sex cohabiting and married couples. In this research note comparing the largest and most recent federal surveys—the 2019 American Community Survey and Current Population Survey—we find consistent levels of cohabitation and marriage across surveys. While the vast majority (90%) of different-sex couples were married, we report a more even split in cohabitation and marriage among same-sex couples. Our evaluation of sociodemographic characteristics of married and cohabiting couples indicates that differences were less prominent among same-sex couples than among different-sex couples, suggesting weaker sociodemographic selection into marriage among the former. However, factors affecting same-sex and different-sex couples' decisions to live together and marry may differ because of legal and social climates that still present unique obstacles for same-sex couples. Researchers need to acknowledge these differences in assessments of the implications of marriage for health and well-being.
The U.S. Supreme Court ruled in 2015 that marriages of same-sex couples were legal in all states, and concurrently the U.S. Census Bureau revised their data collections to directly count same-sex and different-sex married and cohabiting couples. These legal and data developments provided new prospects for studying cohabitation and marriage patterns. Given the new legal context of marriage, selection into marriage for same-sex couples may be quite distinct from that for different-sex couples. Generally, married couples are more economically advantaged than cohabiting couples, in part because of the rising expectations that couples be economically settled prior to marrying (Addo 2014; Gibson-Davis et al. 2018; Ishizuka 2018; Vespa and Painter 2011). Further, cohabiting couples with more positive economic prospects more often marry than those with weaker economic circumstances (Guzzo 2014; Lamidi et al. 2019). As the economic bar for marriage has increased and is more difficult to attain (Ishizuka 2018), marriage has increasingly become the terrain of the most advantaged in terms of education and income. This selection is used to explain differentials in the well-being of cohabiting and married couples (e.g., Brown 2017; James and Beattie 2012; Musick and Bumpass 2012; Perelli-Harris and Styrc 2018; Williams et al. 2011). These arguments and findings are centered around different-sex couples and may not apply to same-sex couples. Although same-sex and different-sex couples share similar motivations for marriage in terms of love and commitment, the high levels of discrimination and stigma may make the legal protections offered by marriage critically important for same-sex couples (Drabble et al. 2020; Gates 2015; National Academies of Sciences, Engineering, and Medicine (NASEM) 2020; Pew Research Center 2013). Thus, the economic bar for marriage could be lower for same-sex than for different-sex couples.
The data changes that make analysis of same-sex couples possible are based on the Census Bureau's modification to items on respondents' relationship to householder rosters, which now include direct questions about whether respondents were living with an “opposite-sex” or “same-sex” spouse/husband/wife or unmarried partner (Kreider and Gurrentz 2019; Manning and Payne 2021). Prior to this change, there were major methodological challenges associated with the accurate measurement of same-sex couples, which relied on an indirect approach based on responses to questions about the sex of the householder and their spouse or unmarried partner (Black et al. 2007; Cohn 2011; DiBennardo and Gates 2014; Gates and Steinberger 2009; Kreider and Lofquist 2015; O'Connell and Feliz 2011; O'Connell and Gooding 2006).1 The revised items on the household roster were first applied to all households in the 2017 Current Population Survey (CPS) and the 2019 American Community Survey (ACS). We compare estimates of cohabitation and marriage among same-sex couples in the 2019 CPS and ACS, updating similar work by the Census Bureau a decade ago (Lofquist and Ellis 2011).2 The improved measurement of cohabitation and marriage among same-sex couples provides new opportunities for assessing the characteristics of same-sex couples using large-scale, population-based data.
ACS and CPS are the two largest data sets available that include direct estimates of same-sex marriage and cohabitation, and we compare estimates across surveys and examine whether the marriage–cohabitation differences in sociodemographic characteristics evident for different-sex couples also exist for same-sex couples. We provide an update of early evidence of differentials according to marital status that were noted using unedited decennial 2000 census data (Fields and Clark 1999). If the marriage–cohabitation differences are comparable, this could be considered evidence for similar selection into marriage and cohabitation for same-sex and different-sex couples. Given the important legal protections offered by marriage, we expect that marriage–cohabitation sociodemographic differences may not be as great for same-sex as for different-sex couples. These results have ramifications for our understanding of the levels, correlates, and implications of cohabitation and marriage among same-sex couples. A secondary goal is to determine if similar marriage–cohabitation differentials exist among male and female same-sex couples. Because of sample size constraints, we focus on the ACS for these contrasts.
Data and Methods
We rely on the largest demographic surveys in the United States, the recently released 2019 American Community Survey and the 2019 Current Population Survey. Sociodemographic indicators were employed to reflect couple-level characteristics. The categorization of the resulting measures ensured couple-level detail without compromising statistical power. Identification of same-sex cohabiting and married couples was based on the question describing the relationship to the householder of all individuals in the household. Response options included opposite-sex spouse (husband/wife), opposite-sex unmarried partner, same-sex spouse (husband/wife), and same-sex unmarried partner; these were coded into married and cohabiting same-sex and different-sex (opposite-sex) couples.
The ACS is conducted annually and provides up-to-date information that was formerly available on the “long-form” of the decennial census. Participation in the ACS is required by law. The one-year estimates are representative of areas with populations of at least 65,000 people and sampled 3.54 million households. We rely on the 2019 ACS Public Use Microdata Sample (PUMS) downloaded from the U.S. Census Bureau website (U.S. Census Bureau 2020). The 2019 ACS PUMS, representing approximately 1% of all households, includes a sample of 3,239,553 individuals from 1,276,716 households. The ACS conducts surveys over the entire year. The data are largely (70%) collected by mail or internet with follow-ups via phone and in-person interviews. All results are weighted using the household replicate weights provided by the Census Bureau (person-level weights were used to calculate average household income per person). The revised relationship to householder roster was first administered to respondents in the 2019 ACS.
The CPS is jointly sponsored by the Bureau of Labor Statistics and the Census Bureau, and we rely on the March Annual Social and Economic Supplement (ASEC) of the Integrated Public Use Microdata Series (Flood et al. 2018). Unlike for the ACS, most of these data are collected in March. The CPS is nationally representative, and all data are weighted (household replicate weights are applied to generate empirically derived standard errors). The CPS is based on a sampling frame of about 100,000 housing units conducted by computer-assisted phone or in-person interviews, and in 2019 it included 180,101 respondents. Participation in the CPS is not required by law. While both surveys share the same questions to identify couples' relationship status and sex composition, the CPS also includes a “cohabitation pointer” to identify cohabiting couples who do not include the householder. But that pointer was not used in these analyses because there is no direct indicator of whether the cohabiting partner is “same-sex” or “different-sex.” Starting in 2017, all respondents in the CPS received the new relationship to householder response options (Kreider and Gurrentz 2019).
Relationship status was derived from the relationship to householder question, with “spouse/husband/wife” responses coded as married, and “unmarried partner” responses coded as cohabiting.
Household composition of the couple was coded into three categories: (1) couple-only households, (2) households with the couple and at least one biological/step/adopted child (and possibly others), and (3) households with the couple and others who were not biological/step/adopted children.
Sex, age, race/ethnicity, nativity status, and residence of the couple were included as demographic indicators. Couple sex was coded as different-sex couple, male couple, or female couple. While this is a flawed conceptualization, we are limited to the binary measures included in the surveys. Age of the younger partner was coded categorically into four groups: 18–29, 30–39, 40–49, and 50+. The couple's age gap was calculated by subtracting their ages. The racial/ethnic composition of the couple was coded into a four-category variable: (1) both non-Hispanic Black, (2) both non-Hispanic White, (3) both Hispanic, and (4) interracial/ethnic or non-Hispanic other (Asian, American Indian, or two or more racial/ethnic groups). Nativity of the couple was coded as 1 if at least one member of the couple was foreign-born, with all others coded 0. Residential history identifies couples in which at least one member had moved in the previous year. Region of current residence was coded into categories based on the census regions Northeast, Midwest, South, and West.
We coded four socioeconomic status measures. Educational attainment of the couple was coded into three categories: (1) couples in which both members had a high school diploma or less, (2) couples in which only one member had at least a bachelor's degree, and (3) couples in which both members had at least a bachelor's degree. Couples' employment status was coded as a three-category variable: (1) both members worked full-time, (2) one member worked full-time, or (3) neither member worked full-time. Housing tenure distinguished between couples who owned their home or rented. The mean household income per person was a continuous indicator measuring income in the last year.
The shares of same-sex and different-sex couples who were cohabiting and married are presented in Table 1. Nearly 90% of different-sex couples were married in the ACS and CPS—88% and 89%, respectively. In contrast, just over half (and statistically similar percentages) of same-sex couples were married in these surveys—58% and 54%, respectively. Thus, the ACS and CPS estimates of cohabitation and marriage are quite similar.3
Table 1 also shows the characteristics of married and cohabiting couples within each survey, denoting differentials in characteristics. Across the surveys, same-sex married and cohabiting couples were more similar than were different-sex married and cohabiting couples. While the point estimates associated with significant differences occasionally differ across the data sets, the levels typically trend in the same direction.4
Married couples more often lived with children than did cohabiting couples (same-sex couples in the CPS trend in this direction, but not significantly). Married couples were older than cohabiting couples among both same-sex and different-sex couples—the modal age category of the younger partner among married couples was 50+, whereas the modal category among cohabiting couples was 18–29 regardless of couple gender composition. Within-couple age gaps were similar across marital status for same-sex couples, but among different-sex couples, the age gap among married individuals was smaller than that among cohabitors. While same-sex couples had few differentials in their racial/ethnic composition according to marital status, different-sex married and cohabiting couples had disparate racial/ethnic composition. Married couples more often included someone who was foreign-born, except in the CPS among same-sex couples. Cohabiting couples were more likely to have moved in the last year than were married couples, except for same-sex couples in the CPS. Although there were regional residence differences, the data did not reflect a singular pattern.
Married couples had higher levels of educational attainment than cohabiting couples among both same-sex and different-sex couples, but the marital status differential was greater among different-sex couples. The marital status employment differential exists primarily for different-sex couples, with few differences for same-sex couples. Home ownership was substantially greater for married couples, but the marital status differential was larger for different-sex couples. Across surveys, respondents in married couple households reported higher incomes per person, and this difference was greater for different-sex couples in the ACS and for same-sex couples in the CPS.
A final set of comparisons considers the cohabitation–marriage differentials for male and female same-sex couples using the ACS. In Table 2, the pattern of findings is similar for male and female same-sex couples: greater shares of married couples than of cohabiting couples live with children, are aged 50 or older, are foreign-born, are residentially stable, are highly educated, are homeowners, and earn higher incomes. Among male couples, the racial/ethnic composition of married and cohabiting individuals is similar, but among female couples, greater shares of married than cohabiting individuals identify as White.
It is important that surveys capture the relationship status of same-sex and different-sex couples, and particularly critical given the recent passage of marriage equality. Contrasting levels and patterns across surveys matter for newly developed questions and research involving relatively small groups, such as sexually diverse and gender-diverse populations (NASEM 2020). Our analysis yields three key sets of findings. First, although the sampling, sample sizes, and interview mode are different across surveys, the estimates of marriage and cohabitation were similar, with marriage being more common among different-sex than among same-sex couples. This finding aligns with earlier indirect evidence and is consistent with recent Gallup results (Jones 2021). A few differentials existed in the sociodemographic composition of married and cohabiting same-sex couples across surveys, and we speculate that these were largely due to interview mode and smaller sample sizes in the CPS. Second, answers to questions about whether the same marriage advantage observed among different-sex couples exists among same-sex couples are consistent across surveys. While differentials exist for married and cohabiting couples, many of the differentials are consistently greater among different-sex than among same-sex couples, specifically educational attainment, employment, homeownership, age, and presence of children. Third, a focus on the sex composition of the couple shows differences in their sociodemographic profiles, but a similar marriage–cohabitation differential is evident among same-sex male and female couples.
Comparisons among couples based on sex composition and marital status can be complicated because of social and legal climates that can affect couples' decisions to marry. For same-sex couples, the recent widespread availability of legal marriage is a factor in who decides to marry. Over time, new patterns of selection into marriage may be observed among same-sex couples. Regardless of legal conditions, same-sex couples frequently face a social climate that resists acceptance of their relationships. As a result, some may avoid marriage because of potential hostility from family and employment discrimination associated with the requisite outing that accompanies marriage (Drabble et al. 2020; Ocobock 2013). Hence, who opts to couple and marry differs in important ways for same-sex and different-sex couples. Research on the implications of marriage for the health and well-being of same-sex couples requires a nuanced approach.
While the new relationship options in federal surveys represent an important step forward, there remain critical measurement issues. First, the measurement of “sex” rather than gender is problematic, and treating gender as a binary construct provides a limited view of how individuals identify and ignores the transgender population. Further, respondents' sexual identity is omitted, and gender composition of the couple is not directly analogous to sexual identity. Another constraint is that the relationship items in the roster are not available to identify married and cohabiting couples who are “doubling up” and living in households where one member of the couple is not the householder. Given sociodemographic patterns of doubling-up (Mykyta and Macartney 2011), findings based on householders are skewed toward more advantaged couples. Finally, these data are limited to current relationship status and do not reflect transitions into and out of marriage, cohabitation, and singlehood. Attention to relationship transitions will be a critical component of new data collections and an important next step in assessments of cohabitation and marriage premiums among sexual minorities. The federal data infrastructure must continue to keep pace with the country's shifting family landscape and include measurement of all family forms.
This research was supported in part by the Center for Family and Demographic Research, Bowling Green State University, which receives core funding from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (P2CHD050959).
The language used to refer to sexually diverse and gender-diverse populations is emerging (NASEM 2020). We use “sex” rather than “gender” to mirror the language used in the data collections, rather than the preferred language used in the literature. Certainly, sex and gender are not interchangeable and “opposite” denotes an outmoded binary understanding of gender.
Further contrasts across the CPS and ACS estimates according to sociodemographic characteristics are presented in the online Supplemental Table 1. Same-sex married couples are similar in both surveys across most indicators (except household composition and income), as are same-sex cohabiting couples (except couples’ marital history and residential moves).
The larger samples of different-sex couples may result in more significant differences (owing to smaller standard errors), but the magnitude of the difference is typically greater among different-sex couples.