Introduction

On June 26th, 2015, the U.S. Supreme Court’s decision in Obergefell v. Hodges legalized same-sex marriage nationwide. To further protect these rights, Congress passed the Respect for Marriage Act, which President Biden signed into law on December 13, 2022, ensuring federal recognition of same-sex marriages. Despite these significant legal advancements, demographic research on marriage and cohabitation among sexually minoritized populations remains stalled because national data on marriage and cohabitation by sexual identity remains fragmented. There is currently no single publicly available dataset to provide up-to-date estimates of marriage rates and cohabitation among sexually diverse populations. Marriage rates are an important demographic tool for assessing marriage patterns across populations, but there are no marriage rate estimates based on sexual identity. Given that, until recently same-sex couples could only cohabit, and nearly half still do—compared to just 10% of different-sex couples (Manning et al., 2022)—cohabitation remains a significant part of the relationship landscape, highlighting the importance of analyzing cohabiting and marital relationships separately.

Using both the American Community Survey (ACS) and the Household Pulse Survey (HPS) we have two goals (1) estimate refined marriage rates for straight, lesbian, and gay-identifying individuals aged 18–64 and (2) examine the union status distributions according to sexual identity, sex, and age. Although the ACS collects data on same-sex marital and cohabiting unions via their household roster, it does not contain questions about sexual identity, making it impossible to rely on just the ACS to calculate rates of marriage or estimate the union status for gay and lesbian individuals. Conversely, the HPS, which does include questions about sexual identity, lacks critical data on the timing of marriages, the sex of spouses, or cohabitation status. By leveraging these two data sources our study provides the first national estimates of marriage rates for lesbian and gay identifying populations. We assess variations by age because of age-based marital patterns (Hemez, 2020; US Census Bureau, 2023), and the higher average age at marriage among gay and lesbian individuals (Westrick-Payne & Manning, 2023a) along with the tendency of younger birth cohorts to endorse lesbian and gay identities more often (Jones, 2024; Julian et al., 2024). We provide new insights into marriage and cohabitation patterns providing the groundwork for future scholarship on union formation and stability of lesbian and gay populations.

Data and Methods

This study utilizes data from two U.S. Census Bureau sources—the American Community Survey (ACS) and the Household Pulse Survey (HPS)—to estimate refined marriage rates [(number of marriages/population aged 15 and older who were single, widowed, or divorced)*1,000] and the share of individuals in cohabiting and marital unions according to sexual identity. Both datasets include large sample sizes permitting national-level single year estimates of marriage and cohabitation. Due to data limitations, the numerators in these analyses are derived from the ACS and the denominators come from the HPS. Although relying on two datasets is not ideal, this method aligns with traditional practices by the National Center for Health Statistics (NCHS), which uses administrative counts of marriages as numerators and census or survey data as denominators.Footnote 1

Data

Administered by the U.S. Census Bureau, the ACS is the largest nationally representative survey fielded by the federal government. The questionnaire is dispatched to a sample of addresses across the U.S., totaling around 3.5 million solicitations a year. The 1-year estimates encapsulate 12 months of collected data with the most recent available data reflecting the year 2022.Footnote 2

The HPS was launched in 2020 by the U.S. Census Bureau as a part of their Experimental Data Series to measure the impacts of the COVID-19 pandemic (U.S. Census Bureau, 2022). To produce estimates quickly and efficiently, they employ a weekly phase sampling strategy. In July of 2021, the HPS began asking respondents about their sexual and gender identities, making it the largest national survey to do so (Anderson et al., 2021). HPS response rates are lower than in the ACS (Julian et al., 2024; U.S. Census Bureau, 2022), and although weights are used to help adjust for some biases, others likely remain (Peterson et al., 2021). As such, the Census Bureau advises caution when using HPS data as standalone estimates for the prevalence of LGBT adults (Fields, 2024).Footnote 3 Building on prior comparisons (Deng & Watson, 2023; Julian et al., 2024) we provide contrasts of HPS estimates to 5 national data sources (Appendix Table 1) and show the HPS contains higher overall LGB estimates.Footnote 4 For consistency with the 2022 ACS, we use HPS data covering weeks 41 to 52 (all weeks of data collected in 2022) for a total of 753,217 respondents. Missing HPS data is not generally imputed, but some imputation was necessary to generate survey weights. The Census used hot-deck imputation for age, race, Hispanic origin, sex at birth, education, and total number of adults derived from ACS estimates and the Census Bureau’s Population Estimates Program (PEP), with person weights adjusted to be representative of the average population through an iterative raking ratio (File & Scherer, 2023; U.S. Census Bureau, 2022).

Measures

To estimate refined marriage rates by sexual identity the ACS is used for the numerators because the HPS does not ask about recent marriages. ACS respondents are asked whether they got married in the past 12 months. Respondents are also asked their relationship to each household member in which they can identify an “opposite-sex” or “same-sex” husband/wife/spouse. Therefore, the numerator for the refined same-sex marriage rate is the number of currently married individuals aged 18 to 64 in a same-sex marriage (disaggregated by sex) who married in the past 12 months. Similarly, the numerator for the refined opposite-sex marriage rate is the number of currently married individuals aged 18 to 64 in an opposite-sex marriage who married in the past 12 months.Footnote 5

The denominators for these refined marriage rates—the population “at risk” of marriage—come from the HPS, as the ACS does not estimate the unmarried population by sexual identity. HPS respondents are asked “Which of the following best represents how you think of yourself? (1) Gay or lesbian, (2) Straight, that is not gay or lesbian, (3) Bisexual, (4) Something else, please specify, or (5) Don’t know.”Footnote 6 Combined with the question on respondents’ current marital status we estimate the number who were currently unmarried and aged 18 to 64 by sexual identity.Footnote 7 Figure 1 provides a detailed visualization of the computation of the refined marriage rates by sexual identity.Footnote 8

Fig. 1
figure 1

Estimates and computations for refined marriage rates by sexual identity. Sources Numerator, U.S. Census Bureau, 2022 American Community Survey, 1-year PUMS; Denominator, U.S. Census Bureau, Household Pulse Survey, Weeks 41–52, PUF

Full size image

To estimate coresidential union type (cohabitation, marriage) we again use the ACS for the numerators and the HPS for the denominators. The ACS provides information on cohabitation status based on the household roster, whereas the HPS does not have any measure of cohabitation. In the ACS, respondents can identify their relationship to each household member as an “opposite-sex” or “same-sex” husband/wife/spouse as well as an “opposite-sex” or “same-sex” unmarried partner.

The numerators in our analysis of union type consist of (1) the number of individuals currently in a cohabiting relationship with a same-sex partner or an opposite-sex partner, and (2) the number of individuals currently in a married relationship with a same-sex partner or an opposite-sex partner, and both are disaggregated by sex and age. The denominator, again estimated from the HPS, because the ACS does not ask about sexual identity, represents the total population aged 18 to 64 who identify as gay males, lesbian females, straight males, and straight females.

Regarding sex, the ACS asks about sex at interview, with respondents providing answers for all household members, and the HPS asks about the respondents’ sex at birth.Footnote 9 The ACS provides age at interview, whereas age in the HPS is calculated by subtracting respondents’ birth year from 2022. All counts and percentages are weighted, and replicate weights were used to derive appropriate standard errors for the estimation of confidence intervals.

Results

2022 Marriage Rates

The 2022 refined marriage rates reveal distinct patterns by sex and sexual identity (Fig. 2). Gay males had a significantly lower refined marriage rate (18.2 per 1000 unmarried) compared to straight males (54.7 per 1000). This disparity may reflect lingering social, cultural, or economic barriers that influence marriage decisions among gay males. In contrast, the refined marriage rates for lesbian and straight females were similar (48.2 vs. 50.4 per 1000 unmarried), that may reflect similar expectations for marriage and potentially reflecting changing societal norms or greater acceptance of female same-sex relationships (Drabble et al., 2021, McCarthy, 2023; Pew, 2019).

Fig. 2
figure 2

Refined marriage rates by sex and sexual identity per 1000 unmarried population. Sources Numerator, U.S. Census Bureau (2022) American Community Survey, 1-year PUMS; Denominator, U.S. Census Bureau, Household Pulse Survey, Weeks 41–52, PUF. Notes Refined marriage rates of males and females are shown by sexual identity (error bars show 90% confidence intervals). Numerators, denominators, refined marriage rates, and 90% confidence intervals are available in Appendix Table 2

Full size image

2022 Marriage Rates by Age Group

Given the distributions in the age at marriage and later ages at marriage observed among gay and lesbian couples (Westrick-Payne & Manning, 2023a, 2023b), we investigate marriage rates by age. Marriage rates vary substantially by age, with notable differences across sexual identities (see Fig. 3).Footnote 10 Among younger individuals (18–29), gay males had the lowest marriage rate (17.9 marriages per 1000 unmarried), reflecting a delayed entry into marriage compared to their straight counterparts. This delay is consistent with prior research showing that gay males tend to marry at older ages (Westrick-Payne & Manning, 2023a). Similarly, lesbian females in this age group had a relatively low marriage rate (38.8 per 1000 unmarried), and straight females had the highest rate (85.3 per 1000 unmarried), emphasizing pronounced age-related gaps in marriage patterns across sexual identities.

Fig. 3
figure 3

Marriage rates by sex, sexual identity, and age group per 1000 unmarried population. Sources Numerator, U.S. Census Bureau (2022) American Community Survey, 1-year PUMS; Denominator, U.S. Census Bureau, Household Pulse Survey, Weeks 41–52, PUF. Notes Refined marriage rates by age, sex, and sexual identity (error bars show 90% confidence intervals). Numerators, denominators, refined marriage rates, and 90% confidence intervals are available in Appendix Table 2

Full size image

As individuals age into their 30 s, marriage rates increase, but disparities persist gay males still have the lowest rates (21.1 per 1000 unmarried), underscoring their continued lower likelihood of marrying. Lesbian females, however, close the gap with straight females, both showing marriage rates near 75 per 1000 unmarried, suggesting that marriage becomes more appealing or accessible to lesbian females as they age. In the 40 s and 50 s age groups, the pattern shifts again. Lesbian females surpass both straight females and males in marriage rates, particularly in their 40 s (70.4 per 1000 unmarried), indicating a possible convergence in marital behavior among older females.

Coresidential Union Type by Age Group

The analysis of coresidential union types (Figs. 4 and 5) further highlights the complexity of relationships across sexual identities and age groups. Gay males consistently have lower shares in both marital and cohabiting unions across all age groups, with a slight increase in those in marital unions among older cohorts. This trend may suggest that gay males face unique challenges in forming and maintaining coresidential unions, or that alternative relationships forms, such as non-coresidential partnerships are preferred.

Fig. 4
figure 4

Share of male (sex at birth) population in a coresidential union by union type, sexual identity, and age group. Sources Numerator, U.S. Census Bureau (2022) American Community Survey, 1-year PUMS; Denominator, U.S. Census Bureau, Household Pulse Survey, Weeks 41–52, PUF. Notes Numerators, denominators, percentages married and percentages cohabiting, and 90% confidence intervals are available in Appendix Tables 3 and 4

Full size image
Fig. 5
figure 5

Share of female (sex at birth) population in a coresidential union by union type, sexual identity, and age group. Sources Numerator, U.S. Census Bureau, 2022) American Community Survey, 1-year PUMS; Denominator, U.S. Census Bureau, Household Pulse Survey, Weeks 41–52, PUF. Notes Numerators, denominators, percentages married and percentages cohabiting, and 90% confidence intervals are available in Appendix Tables 3 and 4

Full size image

For females, the patterns are different. Lesbian females exhibit higher shares in cohabiting unions across all age groups except among the youngest group (those aged 18–29) compared to their straight counterparts, particularly among those aged 50–64, where 14% of lesbian cohabit compared to only 5% of straight females. This suggests that cohabitation remains an important relationship form for older lesbian females, even as marriage rates increase with age. We find that marriage is more common among straight females, particularly at younger ages, but lesbian females catch up by their 40 s, narrowing the gap significantly.

Discussion

This research note provides up-to-date refined marriage rates that adjusted for the unmarried population and distributions across coresidential relationships among gay, lesbian, and straight populations in the United States. Utilizing two sources of data from the U.S. Census Bureau collected in 2022, our findings underscore the importance of considering age when analyzing marital behavior alongside sexual identity. Notably, we observed that marriages and coresidential relationships were less common among gay identifying males across all age groups and relationship types. Conversely, marriage rates among lesbian identifying females were comparable to those among straight identifying males and females, albeit with differing age patterns. These patterns may shift as new generations of lesbian and gay individuals in the U.S. grow up in a nation with marriage equality.

Although we provide new estimates of refined marriage rates and union status distinguishing sexual identity, there are some limitations. First, the HPS was designed as a demonstration project and released as part of the Census Bureau’s Experimental Statistical Products Series (U.S. Census Bureau, 2022). As such, the Census did not perform edits or statistical imputation for the majority of survey results (File & Sherer, 2023). Although the Census used the ACS to compute person weights and measure potential sampling error, nonsampling errors such as measurement, coverage, nonresponse, and processing errors may still occur and are not reflected in the standard errors of the HPS (U.S. Census Bureau, 2022). As noted earlier, the Census advises caution when using HPS questions to estimate prevalence of LGBT adults in the U.S and consistent with prior work we found HPS estimates of gay and lesbian individuals are higher than other data sources (Deng & Watson, 2023; Julian et al., 2024) potentially biasing our same-sex marriage rate estimates downward. Certainly, sample size and precision are crucial when estimating smaller subpopulations, like sexual minorities, as their margins of error can be large. Second, due to differences in data collection processes and editing procedures the HPS is not entirely comparable with other data sources. In short, the HPS and ACS are unique with different modes and response rates. Because estimating sexual minority populations is still evolving, it is important to assess each survey's ability to accurately estimate this subpopulation. Third, we cannot distinguish bisexual individuals in the ACS which may bias estimates. We cannot determine the full extent of the bias but most married and cohabiting bisexual individuals (81–93%) are in opposite-sex unions (Hsieh & Liu, 2019; Jones, 2022; Parker, 2015;), but represent a small share of all individuals in opposite-sex unions. Further, most same-sex couples include individuals with gay or lesbian identities. Calculations based on estimates from NHIS indicate 95% of married same-sex couples are gay or lesbian (Hsieh & Liu, 2019) and drawing on Gallup results about 85% (Jones, 2022). As a result, our counts of marriages to same-sex couples are slightly overestimating marriages to lesbian and gay couples. When we reduce the number of same-sex couples by an average of these two reports (90%) we obtain similar findings. Demographers need to move beyond traditional approaches to understand union formation for diverse populations. One way to advance research on marriage rates and cohabitation is population-based data that includes sexual identity, recent marriages and cohabitation, and current marital and cohabitation status measures.

These analyses highlight the nuanced ways in which sex, age and sexual identity intersect to shape marriage and coresidential union patterns. Although straight individuals, especially males, marry at higher rates across most age groups, lesbian females demonstrate a unique age trajectory, with marriage becoming more prevalent in midlife. The reason for these differing patterns by age among females warrant further investigation. One theory is that it may be driven by the fact females have longer life expectancies, generally throughout the lifecourse (Feraldi & Zarulli, 2022). However, work on the health and wellbeing of sexual minorities does not indicate they fare better than their straight counterparts (Choi & Meyer, 2016; Hoy-Ellis & Fredriksen-Goldsen, 2016). Specific to mortality, current scholarship indicates a significantly elevated risk among sexual minorities compared to their non-sexual-minority counterparts (Cochran et al., 2015; Laughney & Eliason, 2022). Accounting for differentials in mortality was beyond the scope of this project. Gay males, on the other hand, across age groups continue to have smaller shares in marital and cohabiting unions, suggesting that traditional coresidential unions may not fully capture their relationships experiences. Thus, sex differentials in marriage are masked when lesbian and gay populations are combined together. We underscore the need for new data infrastructure and continued exploration of how sexual identity disaggregated by sex/gender influences relationship formation, particularly as legal and social landscapes evolve.