|Home | About | Journals | Submit | Contact Us | Français|
Much recent research on sexual minorities has used couples-based samples, which—by construction—provide no information on nonpartnered individuals. We present the first systematic empirical analysis of partnership and cohabitation among self-identified gay men and lesbians using two independent, large, population-wwbased data sources from California. These data indicate that 37%–46% of gay men and 51%–62% of lesbians aged 18–59 are in cohabiting partnerships (compared with 62% of heterosexual individuals in coresidential unions at comparable ages). Unlike previous research, we find that white and highly educated gay men and lesbians are more likely to be partnered, and we confirm that same-sex cohabiting partners in our data have demographic characteristics that are similar to California same-sex couples from Census 2000. We also present the first detailed analysis of officially registered domestic partnerships in California. We find that almost half of partnered lesbians are officially registered with the local or state government, while less than a quarter of partnered gay men are officially registered. We conclude with implications of our findings for couples-based research on gay men and lesbians, as well as recommendations for survey data collection.
The availability of new social science data that allow credible identification of sexual minorities has noticeably increased social science research focusing on gay men and lesbians over the last decade. Perhaps the most widely cited data source used to explore demographic characteristics of the gay and lesbian population is the U.S. decennial census, which allows for the identification of same-sex “unmarried partners,” commonly understood as coupled gay men and lesbians, through descriptions of intrahousehold relationships.1 Black et al. (2000), for example, used samples of same-sex unmarried partner couples from the 1990 census to provide a broad demographic picture of gay and lesbian couples in the United States, and more recently Black, Sanders, and Taylor (2007) used the analogous couples from Census 2000 to describe this unique population.
By construct, however, the decennial census cannot provide evidence on (1) nonpartnered gay men and lesbians and (2) coupled gay men and lesbians who either do not live with each other or who do not choose to describe their relationship as an “unmarried partnership.”2 As such, the census and other data in which sexual minorities are identified only through household relationships (such as the American Community Surveys or the Current Population Surveys) cannot provide evidence on the prevalence of partnership and cohabitation among gay men and lesbians. Moreover, a consistent limitation of couples-based research has been its inability to determine how generalizable any results are to the overall gay and lesbian population, since this question turns explicitly on the degree to which partnered sexual minorities differ from nonpartnered sexual minorities. As Black et al. (2000:141–42) wrote, “understanding partnership is crucial to understanding the sample of gay men and lesbians identified in . . . census data.” Black et al. (2000) and Black et al. (2007) provided some estimates of gay and lesbian partnership prevalence using the General Social Survey (GSS) and the National Health and Social Life Surveys (NHSLS), but these data are limited by small sample sizes and very limited information about the nature of the partnership.3 Importantly, the brief evidence on partnership in previous work was limited to establishing the feasibility of using couples-based data to describe gay men and lesbians.
Our study, in contrast, focuses exclusively on the prevalence of gay and lesbian partnership and cohabitation. Although there is a large and extensive literature on partnership and cohabitation among heterosexual individuals, there is comparatively little research on partnership among gay men and lesbians (see Seltzer 2000 and Smock 2000 for reviews of partnership among heterosexuals). We use data from the State of California’s 2003 Lesbian, Gay, Bisexual, and Transgender (LGBT) Tobacco Survey (henceforth, the “ Tobacco Survey”) and the 2001, 2003, and 2005 waves of the California Health Interview Survey (CHIS). Both are large, independent, population-based telephone surveys that are representative of the state of California. Census 2000 data suggest that more than 15% of all same-sex couples in the United States live in California; as such, California constitutes an appropriate setting for the study of sexual minorities. Our data are distinguished from both the census and the GSS in that both the Tobacco Survey and the CHIS contain individual-level information on partnership status and direct measures of sexual orientation for all respondents. This allows us to estimate the fractions of gay and lesbian individuals that are partnered and to examine individual demographic correlates of partnership. Our data also contain samples of sexual minorities that are 1.7 to 6.1 times larger than those on which previous estimates of partnership prevalence have been based.
In addition to providing detailed evidence on nonpartnered individuals, we offer several other major contributions relative to existing work by Black et al. (2000) and Black et al. (2007). First, we use a more direct and meaningful measure of sexual orientation—adult self-reports—than has been used in previous work. Specifically, our samples are composed of individuals who, when asked a question about their sexual orientation, indicated that they identified as “gay” or “lesbian.” In contrast, the main partnership estimates reported in Black et al. (2000) and Black et al. (2007) are based on GSS data that ask only about sexual behavior; individuals are identified as gay or lesbian in the GSS data if they exhibited recent and/or exclusive same-sex sexual behavior. Such an approach (using behavior to proxy for orientation) necessarily excludes sexual abstainers and incorrectly codes individuals with discordant behavior and orientation. We believe that self-reported sexual orientation— which, as is well-known, is not always concordant with sexual behavior (Laumann et al. 1994)—is more relevant for understanding gay and lesbian partnership and cohabitation, particularly with respect to official domestic partner registrations. This is because the gay men and lesbians who would avail themselves of domestic partner benefits from the government or an employer are probably those who would self-identify as gay or lesbian.
Second, we observe much more direct information on partnership than in previous research. Previous work considers sexual minorities to be partnered if they have a “regular” sex partner and live in a household with at least two adults (Black et al. 2007). Our definition of partnership relies on direct responses to specific questions about living with an unmarried partner (in the CHIS) or cohabiting with a primary romantic partner (in the Tobacco Survey). The Tobacco Survey data also provide detailed information on the nature of the partnership: we can, for example, separately distinguish the presence of a romantic partner from cohabitation with that partner, and we also present the first evidence on relationship duration and cohabitation length among partnered gay men and lesbians.
Third, we present the first estimates of the fraction of gay and lesbian partnerships that are officially registered with the local or state government, as well as the correlates of being officially registered. The prevalence of official registrations in our data is highly relevant for current policy debates: governments can use our estimates—together with the demographic characteristics of gay and lesbian couples in their city or state—to anticipate local demand for official domestic partnership registrations. Finally, we are able to address data quality concerns by comparing the demographic characteristics of gay men and lesbians in partnerships from the CHIS and Tobacco Survey data to the characteristics of same-sex couples in California from Census 2000.
Our study aims to answer five key research questions. First, what fraction of self-identified gay men and lesbians are in a cohabiting partnership? Second, how do the demographic characteristics of partnered lesbians and gay men differ from those of their nonpartnered counterparts? Third, how do the samples of partnered gay men and lesbians from our data sources compare to the much more widely utilized same-sex “unmarried partner” couples from Census 2000? Fourth, what fractions of gay and lesbian cohabiting partnerships are officially registered with the local or state government? And finally, how do the demographic characteristics of those in registered partnerships differ from those not in registered partnerships (and in other relationship states)? To answer these questions we present detailed descriptive statistics on characteristics such as age, race, education, income, and the presence of children from three main data sources: two with individual-level information on partnership and sexual orientation (CHIS and the Tobacco Survey) and one with large samples of same-sex unmarried partner couples (Census 2000).
The 2001, 2003, and 2005 waves of the CHIS were telephone surveys of over 40,000 households in California each year. The CHIS uses a multistage sampling design in which a random adult is selected within each household using random-digit dialing (RDD) methods. When weighted, the sample is representative of the noninstitutionalized population of California. We use confidential versions of these data that contain information on the respondent’s self-reported sexual orientation. Respondents also provide individual information on a variety of health conditions, health behaviors, and demographic characteristics. At the end of the “demographics” section (where age, race, and education information is elicited), adult respondents in the 2001 wave were asked the following: “The next question is about your sexual orientation, and I want to assure you that your answers are completely confidential. Are you gay [lesbian] or bisexual?”4 In 2003 and 2005, individuals were asked, “Do you think of yourself as straight or heterosexual, as gay[lesbian] or homosexual, or bisexual?”5 We use responses to these questions to identify gay men and lesbians.6 The CHIS also includes information on each individual’s partnership status. Specifically, respondents are asked to state their marital status, and one of the choices is “living with a partner.” We identify partnered gay men and lesbians as individuals who reported being gay or lesbian and who concurrently reported living with a partner.
Our approach for identifying partnership among the sample of gay men and lesbians in CHIS has a few drawbacks. Most importantly, we identify partnership on the basis of a question about marital status, and respondents are forced to choose among several categories that need not be mutually exclusive. Another potential problem with our measure is that we do not actually observe the overall sex composition of the household. While it is reasonable to assume that a gay man who reports he is “living with a partner” is, in fact, living with a man (and similarly for lesbians), we cannot verify this to be true. This source of error is likely trivial. In the Tobacco Survey, we find only a single observation of a self-identified lesbian or gay man who reports living with a partner of a different sex.
We note that the measure implies cohabitation with one’s partner and therefore excludes other types of “dating” relationships in which the individuals do not live together.7 For the CHIS and all subsequent samples, we consider adults aged 18–59 (inclusive): this yields 1,306 self-identified gay men and 809 lesbians pooled across the three CHIS waves.
We complement the CHIS with the 2003 California LGBT Tobacco Survey. The Tobacco Survey is a sample of self-identified sexual minorities and individuals reporting same-sex sexual behavior. When weighted, this sample is designed to be representative of California’s lesbian and gay population. The study was commissioned by the California Department of Health and performed by the Field Research Corporation. The telephone-based Tobacco Survey used a disproportionate stratified RDD design and a weighting scheme that explicitly made use of “high-density” gay and lesbian zip codes; importantly, these high-density zip codes were determined by using information on the geographic distribution of same-sex unmarried partners from the 2000 decennial census. This component of the sampling strategy must be kept in mind in the context of our partnership estimates because it is possible that the geographic distribution of sexual minorities varies according to partnership status, and the census identifies only partnered gay men and lesbians. If this is the case, then these data may overstate the proportion of lesbians and gay men who are partnered because the sampling strategy might be biased toward geographic areas where there are more same-sex couples.
All households were first screened using a question that asked whether the respondent was gay, lesbian, or bisexual or had a same-sex sexual experience since age 14. Our analysis sample includes adults aged 18–59 who self-identified as either “gay” or “lesbian,” which includes 770 and 266 individuals, respectively.8
A key advantage of the Tobacco Survey relative to the other data sources is the high level of detail individuals were asked to give about current and previous partnership situations. Specifically, individuals were asked whether they had ever been legally married, as well as their current marital status. Individuals were then asked whether they had a current “primary” partner, which was explained as “someone you love more than anyone else and feel a unique commitment to.”
It is notable that the Tobacco Survey asks about partnership separately from marital status. That is, CHIS identifies individuals who report “living with a partner” as one of the response options to a question about current marital status. Unfortunately, these response options need not be mutually exclusive (e.g., “living with a partner” and “divorced”). The Tobacco Survey, in contrast, asks about marital status and partnership separately. First, the survey asks whether the respondent has ever been legally married. Of those individuals who report having ever been legally married, the survey then asks the respondent’s marital status, intended to elicit current legal marital status. All individuals who did not report that they were currently married were then asked the question about a “primary” partner. In the CHIS, a respondent who is both divorced and currently in a same-sex partnership might not be counted as partnered if he or she chose the “divorced” option on the marital status question. The Tobacco Survey, in contrast, allows respondents to indicate that they are both divorced and in a cohabiting partnership.
Individuals with a current primary partner were then asked whether the partner is same-sex or opposite sex, as well as whether the individual is living with that primary partner. We use combinations of these responses to define “partnered” individuals in the Tobacco Survey as respondents who report living with a same-sex primary partner. Under this definition, individuals who reported a current primary partner but did not report cohabiting with that partner are not considered “partnered” per se. We provide evidence (and prevalence estimates) on these relationships that are likely a mix of those who are seriously dating and those who are “living apart together.” But we impose the general cohabitation requirement to create measures of partnership that are most consistent across data sources. Unfortunately, we do not observe the “has a partner but is not cohabiting with that partner” group in the CHIS. The census data, which we describe later, also require the presence of a same-sex unmarried partner living in the household for identification of gay and lesbian couples.
Individuals in the Tobacco Survey who reported living with a primary partner were also asked about the length of their cohabitation, as well as the length of the overall relationship, and all respondents were asked a battery of standard demographic questions such as race, age, income, education, and the presence of children in the household. Finally, respondents who reported living with a same-sex partner and who also reported not being currently married were asked whether their partnership was registered with the local or state government. We use responses to this question to provide estimates of the prevalence of official domestic partner registrations in California, and we examine the relationship between observable demographic characteristics and domestic partner registration.
Finally, we complement our analyses of California statewide individual-level data with the more well-known Census 2000 data. The Census 2000 5% and 1% Public Use Microdata Samples (PUMS) are drawn from the approximately 20% of households in the United States who received a census “long-form” that asks detailed demographic and economic questions. The PUMS are designed as a 1 in 20 and 1 in 100 sampling of the total U.S. population. We combine the two samples because they are independent draws from the long-form responses. The census does not ask any direct questions about sexual orientation or sexual behavior.9 Rather, census forms include relationship categories that define how individuals in a household are related to the householder. These fall into two broad categories: related persons (e.g., husband/wife, son/daughter), and unrelated persons (e.g., roomer/boarder, unmarried partner). If the householder designates another adult of the same sex as his or her “husband/wife” or “unmarried partner,” Census 2000 enumerates this household as a same-sex unmarried partner couple.10 To accord with our other statewide individual-level data, we present Census 2000 results for California same-sex couples, thus providing an important check on data quality.
Gates and Ost (2004) and Black et al. (2006) suggested a possible serious measurement error problem with census same-sex couple data. Census Bureau coding procedures recode any same-sex “husband” or “wife” from the household roster as an “unmarried partner.” As a result of this procedure, any different-sex married couples that inadvertently miscode the sex of one of the spouses will be coded as same-sex “unmarried partner” couples. Given the 90-to-1 ratio between married and unmarried partners in the census, even rare sex miscodes could significantly contaminate the same-sex couple sample with different-sex married couples. We use the method advanced in Black et al. (2006) and restrict attention to same-sex couples for whom marital status was not allocated for either member of the couple. Census Bureau coding procedures did not permit an “unmarried partner” to have a marital status of “currently married” and allocated any such response. A same-sex “unmarried partner” could be listed as “currently married” for two primary reasons: (1) he or she is part of a same-sex couple in which the partners consider themselves to be married, or (2) he or she is part of a different-sex married couple in which the sex of one of the spouses was miscoded (as described above). By restricting the sample to couples without any marital status allocations, we eliminate the group that is likely to be most prone to this error. Unfortunately, we potentially bias some of the demographic characteristics if same-sex couples who consider themselves to be married differ from those who consider themselves to be “unmarried partners.”
Table 1 presents our main results on the fractions of gay and lesbian individuals who are partnered (again, using our definition that requires cohabitation), as well as demographic characteristics related to partnership. The top row of each panel of Table 1 shows the estimated fraction of each relevant sample in a cohabiting partnership. We find that about 37% of gay men in the CHIS (column 1) and 46% of gay men in the Tobacco survey (column 4) are in a cohabiting partnership.11 For lesbians, we also find a similar partnership estimate for the two California statewide data sources in columns 1 and 4: in the CHIS about 51% of lesbians are in a cohabiting partnership, and the associated estimate for lesbians in the Tobacco Survey is about 61%. For purposes of comparison, the associated partnership estimates for heterosexual individuals in the CHIS (including married individuals) and in the census (California only) is about 62%.
Why are the partnership estimates from the Tobacco Survey slightly higher than the associated estimates from the CHIS? There are several possibilities, though we think a methodological explanation may be particularly important. Recall that the Tobacco Survey recorded partnership status separately from marital status. A problem with the CHIS is that gay and lesbian individuals who are concurrently living with a same-sex partner and who are legally divorced are possibly miscoded as not partnered if they indicate the latter and not the former. Given that 13% of gay men and 28% of lesbians in our Tobacco Survey data reported having ever been legally married, this slippage is potentially substantial. To get a sense of whether this might account for the differences across samples in the fraction partnered, we recoded all Tobacco Survey respondents who reported being divorced, separated, or widowed as nonpartnered. This lowered the partnership estimates for gay and lesbian households to 38% and 46%, respectively, which is much closer to the associated CHIS estimates of 37% and 51%. That the drop in the lesbian partnership estimate is larger than that for gay men is to be expected, since lesbians are much more likely to have been legally married than gay men. Overall, this suggests that surveys combining partnership with marital status in a single question can substantially understate true partnership.
With respect to demographic characteristics in Table 1, both of the gay male samples are largely white, and at least half of the self-identified gay men in both data sources reported having at least a college degree. We find consistent evidence across the two samples that partnered gay men are older, more likely to be white, and more highly educated compared with nonpartnered gay men.12
Like the gay male samples, the majority of lesbians in our California data are white and highly educated. Moreover, the patterns of correlates of partnership for lesbians in our California data are similar to those for gay men: lesbians in partnerships are older, more likely to be white, and more highly educated than nonpartnered lesbians. Overall, the patterns across our two California data sources are very similar, with two exceptions—both related to the presence of children in the household.13 First, the Tobacco Survey data yield higher rates of children present in the household than do the CHIS data, an issue that we revisit below in our census comparisons.14 Second, parenthood rates are very similar for partnered gay males and nonpartnered gay males in the CHIS but are much lower for partnered gay males compared with nonpartnered gay males in the Tobacco Survey data.
How valid are our individual level data on partnership among gay men and lesbians? We are able to assess this question by comparing our data to data from Census 2000. Gay men and lesbians in partnerships from our individual-level data are those that would likely be identified in the sample of same-sex unmarried partner couples from Census 2000; given this, their demographic characteristics should largely accord because the surveys were fielded around the same general time period.15 These comparisons are presented in Table 2. The patterns confirm that the couples from our two probability samples are quite similar to those found in the census. Across a variety of standard demographic characteristics, the Census 2000 sample is very similar to both the CHIS and Tobacco Survey. For example, the average age of partnered gay men is virtually the same in the California Census 2000 and CHIS samples (about 39–40 years) and only slightly higher in the Tobacco Survey (about 41 years). The racial distribution is also very similar: between 72% to 75% of the partnered gay male samples are white. For education, the distributions across the California male samples are also very similar—only 15 to 18% of partnered gay men in the California data have less than a high school diploma—though we find some differences at the higher end of the education distribution. Household income distributions are also quite similar across the California gay male samples in columns 1–3. Finally, we find some evidence that the CHIS underreports the presence of children in partnered gay male households relative to both the Tobacco data and the California Census 2000.
For partnered lesbians in columns 4–6, we also find similar patterns of characteristics across the three data sources, though there are a few more exceptions than in the partnered gay male comparisons. Average age is slightly lower among California Census 2000 same-sex female couple households, but all the California samples are largely white and highly educated. While the low end of the education distribution is similar across columns 1–3, partnered lesbians in the Tobacco Survey are much more likely to have a post-college degree than CHIS partnered lesbians or California Census 2000 same-sex female couples. The lower educational attainment in the California Census 2000 sample in column 6 also translates into lower household incomes. Like the patterns for partnered gay males, we again find a much lower likelihood of children present among the partnered lesbians in the CHIS, though the Tobacco Survey largely accords with California Census 2000 same-sex female couples.
We conclude our investigation by presenting detailed correlates of partnership and “ official” registrations from our Tobacco Survey data in Table 3.16 In doing so, we move the analyses from primarily considering differences in demographic characteristics of individuals based on cohabitation to a comparison of characteristics across four different and potentially distinct groups: those without a partner, those who have a primary partner but who are not cohabiting with that partner (presumably a combination of those who are dating and those who are “living apart together”), those who have a cohabiting primary partner but have not officially registered as domestic partners, and those who have a cohabiting partner and have officially registered as domestic partners.
This marks a conceptual change from a consideration of selection into partnership to a more complex selection process into and out of distinct relationship states. One should be careful not to interpret these different states as steps in a “progression” in relationship formation from single to dating, cohabitation, and formal recognition akin to marriage. The work of Seltzer (2000) and Smock (2000) demonstrates that such a perspective constitutes a naive understanding of the choices and selection into partnership and cohabitation, particularly among those who do not opt to marry. For some, dating and cohabitation represent an intentional progression toward marriage; for others, partnering without cohabiting and cohabitation represent clear alternatives to marriage. In fact, the patterns we document below do not follow a clear gradient across all four relationship statuses.
Table 3 presents the detailed patterns from the Tobacco Survey. A number of patterns for gay men in Panel A are noteworthy. First, only a small fraction of gay men are in officially registered domestic partnerships: while 46% of gay men in the Tobacco Survey are currently cohabiting with a same-sex partner, only 10% of all gay men (about a quarter of those cohabiting) are in partnerships that are officially registered.17 Second, although only about 9% of gay men have ever been legally married, this fraction is much higher (about 20%) for gay males who are currently in a same-sex partnership that is officially registered with the state or local government. Third, gay males in registered partnerships have somewhat longer relationship durations than other partnered gay men whose relationships are not officially registered, and both groups of partnered gay men have been together longer than gay men who have a primary partner but who do not cohabit with that partner. With respect to socioeconomic characteristics, gay men without primary partners are younger, on average, than those with a primary partner, and gay men in officially registered domestic partnerships are substantially older than all other gay men. There is also weaker evidence that the most highly educated partnered gay men are more likely to be officially registered than other gay men of different partnership statuses. Gay men in registered partnerships are more likely to be white than are other gay men and also have the highest household incomes among those who are partnered. Finally, we find that gay men in registered domestic partnerships are substantially less likely to have children in their household, both relative to gay men in cohabiting partnerships that are not registered and especially relative to gay men who report the presence of a primary partner but who do not cohabit.
We perform the same exercise for lesbians in the Tobacco Survey in Panel B of Table 3. Several interesting patterns emerge. First, lesbians are much more likely to be registered with the government than are gay men: fully 28% of all lesbians are in same-sex partnerships that are officially registered. This pattern is consistent with evidence from states that provide some type of formal recognition for same-sex couples: for example, as of April 5, 2006, female couples accounted for 64% of the marriages of same-sex couples performed in Massachusetts (personal correspondence with the Massachusetts Registry of Vital Records and Statistics 2006). Similarly, female couples account for two-thirds of Vermont Civil Unions (The Office of Legislative Council 2002).18
We also find that lesbians are much more likely to have ever been legally married than gay men (consistent with Black et al. 2000); and, similar to gay men, lesbians in registered same-sex domestic partnerships are much more likely to have ever been legally married than are lesbians in cohabiting partnerships that are not registered. With respect to relationship and cohabitation duration, we find somewhat lower durations among lesbians than among partnered gay men. Similar to the patterns for gay men, however, lesbians in officially registered partnerships report longer relationship and cohabitation lengths.19 As was true for black gay men, black lesbians are very unlikely to be in officially registered domestic partnerships. Unlike gay men, lesbian couples’ household incomes are not higher among registered partners than among those who have not registered, though household incomes among both groups of cohabiting lesbians are much higher than among lesbians who have a primary partner but who are not cohabiting. Another dissimilarity relative to the patterns for gay men is that lesbians with a primary partner (regardless of cohabitation or registration status) are much more likely to have children than lesbians without a primary partner.
Are our findings on officially registered domestic partners consistent with a role for economic and legal incentives in partnership registration? Consider that the debates surrounding the legal recognition of same-sex couples, whether through marriage, civil unions, or domestic partnership registries, often suggest economic and legal factors are important for formalizing the partnerships of gay men and lesbians. For example, Bennett and Gates (2004) suggested that marriage could provide a level of economic protection for same-sex couples with children by increasing access to some social programs (like social security) and to health care via employee benefit plans, along with reducing tax burdens for some families. Romero et al. (2007) suggested that same-sex couples with children evidence general economic disadvantage relative to both other same-sex couples and different-sex married couples. Gates, Lau, and Sears (2006) observed higher rates of childrearing among racial and ethnic minority same-sex couples in California along with particular economic disadvantages within this group. Badgett, Gates, and Maisel (forthcoming) provide a detailed treatment of how economic factors might affect the decisions of gay men and lesbians to enter legally recognized relationships.
We find that couples who opt to register tend to have higher socioeconomic status and are more likely to be white. Among men, they are less likely to have children than both single gay men and those in unregistered partnerships. Notably, this pattern is the opposite for lesbians, perhaps somewhat more consistent with the suggestion that those with children gain more from formalizing their relationships. Lesbians in registered partnerships are the most likely to have children present (though they differ only slightly in this regard from those not in registered partnerships).
We find some similarities across the gay male and lesbian samples with respect to official domestic partner registration, however. For both groups, official registration is more prevalent among individuals who had been previously legally married. These individuals likely have more information about the benefits (and costs) of legal recognition of their relationship, as well as more information about the logistics involved with registering their relationship formally. These individuals could also be more likely to have unobserved preferences for commitment in any relationship, whether a same-sex or a different-sex relationship. We also found that among both partnered gay men and lesbians, those whose relationships are officially registered had longer relationship durations than partnered individuals (both cohabiting and noncohabiting) whose relationships are not registered.
Importantly, the association between relationship duration and registration for same-sex couples in our data is conceptually distinct from the association between duration and marriage for different-sex couples. Because state-level partner registration was available for only two years at the time the Tobacco Survey was fielded, the vast majority of same-sex couples in our data did not have the option of registering throughout most of their relationship. This means that the average duration among registered couples is likely to be artificially high because long-duration couples now make up a larger portion of the population of registered couples than they will in the future as couples have the option to register earlier in their relationships. Given this, the decision to register may be less a decision to validate the relationship because it has reached a particular level of commitment and more a decision to register an already marriage-like relationship simply because the option is now available. It may also be true that the association between registered partnership and longer relationship duration indicates that official registration, like marriage, is correlated with unobserved characteristics such as stability and risk aversion and may keep couples together by imposing some of the same responsibilities as traditionally conceived marriage. It may also be that individuals in longer relationships have more common property to protect through legal recognition.
One limitation worth noting with respect to the information on official registration is that the survey question does not distinguish between state and local registration. This could be important because state and local registrations could have very different consequences for couples. In general, state registration brings more comprehensive rights and responsibilities that can affect taxation, parental rights, and formal relationship dissolution compared with local registration that tends to be less comprehensive and often addresses rights like hospital visitation and access to domestic partnership benefits for public employees or employees of public contractors.20 Certainly, neither state nor local registration confers any federal benefits. Also, it is impossible for us to know how many lesbians and gay men are even aware of the state registry. Awareness of the registration procedure and its benefits within the gay and lesbian community could be correlated with higher socioeconomic status (Badgett et al. forthcoming). Another possibility is that awareness could be correlated with location in urban areas, which might also be correlated with higher socioeconomic status and lower rates of childrearing among same-sex couples. Unfortunately, the limitations of our data do not permit us to deeply explore these issues.
We provide new evidence on partnership among gay men and lesbians by using newly available data from California collected using probabilistic sampling techniques. These data include information on partnership status and self-reported sexual orientation and provide sample sizes that are much larger than those used in previous work. More importantly, we base our estimates on direct and detailed measures of both sexual orientation and partnership that are not available in prior studies. Our estimates therefore provide a more accurate measure of the partnership characteristics of self-identified lesbians and gay men in California.21 We find partnership estimates of about 37% to 46% for gay men and 51% to 62% for lesbians. We also document that partnered gay men and lesbians are older, more likely to be white, and more highly educated than their nonpartnered counterparts. Moreover, their demographic characteristics are broadly similar to those from associated samples from Census 2000.
How do these estimates on partnership prevalence and its correlates compare with previous work? Black et al. (2000) used NHSLS and GSS data through 1996 and found that 28% of gay men and 44% of lesbians were in partnerships. More recently, Black et al. (2007) extended the analysis by incorporating GSS data through 2004 and found partnership estimates for gay men and lesbians of about 50% and 63%, respectively. Our data support the pattern that gay men are less likely to be partnered than lesbians, though our actual partnership estimates fall somewhat below the more recent figures.22
Despite the broad agreement that gay men are less likely to be partnered than are lesbians, several specific patterns in our work depart from those in Black et al. (2007) in important ways. First, we estimate that partnership prevalence among gay men is much lower than the associated estimate among heterosexual men (although the estimates for lesbian women are nearly the same as those for heterosexual women). This finding contrasts somewhat with the assessment in Black et al. (2007:56) based on GSS data that “family formation in the gay and lesbian community differs only modestly from the population as a whole.” Second, the correlates of partnership differ substantially between the GSS and our California analyses presented in Table 1. Black et al. (2007:56) found, for example, that “[p]artnered gays and lesbians have levels of education that are similar to their non-partnered counterparts.” In contrast, our California data provide strong evidence that partnered gay men and especially partnered lesbians are more highly educated than their nonpartnered counterparts.
There are several possible explanations for the differences between our findings and those reported in previous work. For example, the GSS and NHSLS data are based on national samples, while our estimates are for California only. Moreover, the GSS and NHSLS data are largely composed of data from the late 1980s and 1990s, while our samples were all collected in 2000 or later. Given that California has long been a visible leader in the LGBT equality movement and the numerous changes in policy and attitudes toward gay men and lesbians nationally since the late 1980s, these spatial and temporal differences could be substantial.
Our findings strongly suggest that researchers should not understate the importance of the apparent selection into and out of partnerships for same-sex couples and the gay and lesbian population more broadly. Given a reasonable distribution of socioeconomic characteristics within the gay and lesbian sample, if individuals with higher socioeconomic status are more likely to find partners (or are less likely to dissolve an existing partnership), then the resulting sample of couples will have higher average socioeconomic status than both the resulting nonpartnered sample and the “true” sample of gay men and lesbians, as is borne out in the CHIS and Tobacco Survey data. Demographic researchers using couples-based samples of gay men and lesbians need to consider the possibility of selection into and out of partnership and its resulting composition, particularly when such selection may be relevant for the research question.
Finally, we conclude with a general recommendation for demographic researchers studying gay men and lesbians: think critically about how the identification of sexual minorities might affect the resulting sample of gay men and lesbians. This recommendation is, of course, not limited to studies of gay and lesbian partnering behaviors but more generally relates to the question of how we identify sexual minorities in large, representative social science and health data. Because of the paucity of surveys that allow identification of sexual orientation, researchers have been creative in thinking about ways to study this important subpopulation. In our opinion, this trend is most welcome but brings with it associated challenges. We have outlined several here with respect to partnership: (1) that couples-based samples may be selected on sociodemographic characteristics such as age, race, education, and childrearing; (2) that identifying partners based on a response option to a traditional marital status question may create problems for gay men and lesbians (and heterosexuals, for that matter) who are living with a partner and at the same time are divorced, separated, widowed, or still legally married to a person of the opposite sex; and (3) that gay and lesbian individuals who report being currently “married” may reflect a heterogeneous group of individuals, some of whom may be married to a same-sex partner.
Correspondingly, these issues we have highlighted translate into recommendations for survey researchers who want to collect information on sexual orientation and partnership. We strongly urge researchers to more routinely include direct measures of sexual orientation identification on surveys, especially on those that might already be measuring sexual behavior. Beyond measuring sexual orientation, our analyses demonstrate the complexities of measuring partnership status among same-sex couples. Some of these complexities revolve around the more general challenges associated with measuring nonmarital cohabitation. In this regard, it is helpful to measure marital status separately from both partnership and cohabitation and to create surveys that allow researchers to distinguish between current and former legal marital status. We also recommend adding a civil union/registered domestic partner response to marital status questions. This would more accurately reflect the legal “marriage” options for same-sex couples (and some different-sex couples who can register in seven states). By 2008, more than 23% of the U.S. population will live in a state that provides a legal status for same-sex couples.23
Finally, we recommend the following: (1) collecting a household sex roster of adults and children in the household as a check on data quality; (2) asking specific questions about the characteristics of the respondent’s partner (e.g., sex, age, current and former legal marital status); (3) collecting a detailed partnership and cohabitation history for each respondent; and (4) collecting samples that are large enough to meaningfully describe gay and lesbian partnership experiences. These suggestions, many of which involve simple modifications of existing response options and/or would involve little extra response burden, would increase the quality and accuracy of social science and demographic data and research on partnership and cohabitation within this increasingly studied population.
We thank the editors, multiple anonymous referees, Dan Black, and seminar participants at the JSM meetings in Toronto for helpful comments. We also thank Larry Bye and the UCLA Data Access Center for data assistance. We thank the Institute for Gay and Lesbian Strategic Studies and the Williams Institute for providing support from a Ford Foundation grant. Gates gratefully acknowledges financial support from the California Center for Population Research (NICHD Grant R24HD41022) and NICHD Grant R03HD046560. Carpenter acknowledges generous support from the Paul Merage School of Business at UC Irvine and a UC Irvine Committee on Research Award. All opinions expressed are those of the authors only.
1.Census data on same-sex unmarried partners are regularly used by government officials and policymakers in major national debates. For example, the Congressional Budget Office (CBO) released a report on the budgetary implications of legalizing same-sex marriage; their analysis relied almost exclusively on assumptions about gay and lesbian people based on data from Census 2000 (CBO 2004).
2.This contrasts directly with census data from countries that explicitly recognize same-sex partnerships. The Canadian census, for example, allows the person filing out the census form to identify another person in the household as a same-sex “common-law partner” or a “same-sex spouse.” See Festy (2007) for a detailed discussion.
3.The GSS identifies gay and lesbian people using measures of same-sex sexual behavior. When pooled, the GSS provides data on 212 gay men and 156 lesbians across the 1989–2004 waves (Black et al. 2007). The NHSLS contains a self-reported measure of sexual orientation identity but has much smaller samples of self-identified gay men and lesbians than the GSS.
4.If the respondent answered “yes” but did not further make clear her sexual orientation, a follow-up question was asked to differentiate between bisexuals and gay men/lesbians. Because of concerns with question wording (Carpenter 2005), we restrict our attention in this paper to adults who reported that they did not have a problem speaking English.
5.If the respondent requested additional information, the interviewer was prompted to say, “Straight or heterosexual people have sex with, or are primarily attracted to people of the opposite sex, gay [lesbian] people have sex with or are primarily attracted to people of the same sex, and bisexuals have sex with or are attracted to people of both sexes.” Unfortunately, there is no way to know which individuals received the additional information.
6.We do not analyze partnership among bisexual-identified individuals in this paper, though it is an important area for future research. One issue is that we do not have a good benchmark data source for comparison for bisexual individuals in partnerships as we do for gay men and lesbians (i.e., the census) because there is no way to identify bisexuals using intrahousehold relationships. A second reason for excluding bisexuals from our analyses is that they are unlikely to be in a same-sex partnership or “at risk” for official domestic partner registrations. Among self-identified bisexual individuals in the Tobacco Survey data, less than 1% of men and just 9% of women are in a same-sex cohabiting partnership. Conversely, more than a third of bisexual men and nearly two-thirds of bisexual women are married or partnered with someone of the opposite sex.
7.In the CHIS, 18 gay men and 19 lesbians report being currently married. These individuals may be in a traditionally conceived heterosexual marriage (i.e., closeted gay men and lesbians), or they may be a member of a same-sex couple in which the partners consider themselves to be married (indeed, they may be legally married in a jurisdiction that permits same-sex couples to do so). Unfortunately, the sex composition of the household is not available in the CHIS. We include these individuals in the full sample across all data sources, but we do not code them as “partnered” in order to provide conservative lower-bound estimates of partnership in California. The broad patterns are little affected by their inclusion, however. Recoding all married gay men and lesbians as partnered—an extreme assumption—increases the estimated fraction partnered by less than 2 percentage points for gay men and by less than 3 percentage points for lesbians (i.e., by less than 5% and 6% of the associated fraction partnered).
8.Specifically, the telephone interviewer asked, “In order to know who can participate in this study, I need to ask you a few questions about the adults who live in your household. We promise to keep all answers confidential. For these interviews, we are interested in speaking with people who are not often studied in public health research: lesbian and bisexual women [gay and bisexual men]. Would you include yourself in one of those groups?” Notably, the screener also included a question about same-sex sexual experience, such that individuals who had experienced same-sex sex were also eligible to participate. If there were multiple eligible individuals in the household, the computer randomly selected a respondent to participate in the phone interview. We did not include 44 individuals who identified themselves as “queer” and five who identified as “questioning” in our analyses because we had no comparable category in the CHIS.
9.Although the census does not ask direct questions about sexual orientation, there is relatively good evidence that the census couples sample is, indeed, gay and lesbian. Black et al. (2000) discussed the reasons why it is unlikely that individuals check the “unmarried partner” option by mistake, and they showed that the spatial distribution of same-sex male couples in the 1990 census closely matches area-specific death rates from AIDS. Carpenter (2004) further documented that same-sex unmarried partner households exhibit sexual and family planning behaviors that are both (1) systematically different from their married and different-sex unmarried partner households and (2) what one would expect if they were, indeed, gay or lesbian.
10.These counts of same-sex couples likely undercount the true population of gay and lesbian couples. Concerns about the confidentiality of their responses may have led many gay and lesbian couples to indicate a status that would not provide evidence of the true nature of their relationship. Other couples may have felt that “unmarried partner” or “husband/wife” does not accurately describe their relationship. A study of the undercount of same-sex unmarried partners in Census 2000 indicates that these were the two most common reasons that gay and lesbian couples chose not to designate themselves as unmarried partners (Badgett and Rogers 2003). Estimates of the undercount range from 15% to 50% (Badgett and Rogers 2003; Gates and Ost 2004).
11.The difference between the reported partnership prevalence estimates and the partnership fraction using the reported sample sizes is due to the fact that we report weighted partnership estimates in the text and tables along with the raw sample sizes on which the weighted means are based.
12.In presenting demographic characteristics by partnership status for gay men and lesbians in California, we are not attempting to identify pathways of causality among partnership, education, and general socioeconomic status measures. We would need richer data—preferably with a longitudinal component—to disentangle whether gay men and lesbians with high socioeconomic status have unobserved characteristics that make them more attractive as partners or whether being in a partnership facilitates improved labor market and educational opportunities. Similarly, socioeconomic characteristics could be related to factors affecting partnership dissolution in addition to partnership formation. All of these possibilities are consistent with the observed patterns in our data that gay men and lesbians with high socioeconomic status are more likely to be in a partnership. Our goal here is to first document these patterns.
13.The parenting outcome is equal to 1 if there are any children under age 18 present in the household at the time of the survey. This measure does not actually require the child to be biologically or legally related to any of the adults in the household. Parenting rates by partnership status are partly mechanically related to the presence of potential parents in the household. As such, the likelihood of parenting is higher for individuals in a couple because there are potentially two parents in the household and our measure makes both of those partners a parent (regardless of the legal parental status of each individual) if a child is present.
14.That the tobacco data yield higher rates of children in the household than the CHIS may be related to the fact that CHIS does not ask partnership separately from marital status (since divorced individuals are more likely to have children from a previous marriage).
15.There have, of course, been numerous changes in attitudes and public policies regarding sexual minorities between 2000 and 2005 that could have changed the incentives to cohabit or otherwise form a partnership— including the 2005 legislation in California (AB205) that gave same-sex domestic partners several of the same rights and responsibilities afforded to married heterosexual couples. As a sensitivity check, we compared findings from our two data sets with same-sex partners from California identified in PUMS from the American Community Surveys (ACS) from 2002 through 2005. The results are qualitatively similar, with the more contemporaneous ACS sample being somewhat more similar to our two data sources than Census 2000 in all characteristics except racial/ethnic composition (the ACS same-sex couples have a higher nonwhite proportion).
16.The samples in columns 2–5 are mutually exclusive. We exclude from those columns the 12 gay men and 13 lesbians who reported that they are currently married because although they reported that they are gay or lesbian, we cannot identify whether their spouse is of the same sex or a different sex. This is because the sex of the partner was asked only to “partnered” (not married) individuals. We include these individuals in the full sample estimates in column 1, however, and they are also included in the denominator when we estimate the fraction of gay men and lesbians in partnerships (as is true in all the data we consider). Columns 2–5 also do not include an additional 5 observations of gay men and 2 observations of lesbians with a missing value that did not allow the classification of partnership status.
17.A common residence is required for official registration in California. See Cal. Fam. Code § 297–297.5.
18.The European experience is notably different in this regard. For example, in the first six months of the availability of civil partnership registration in England and Wales, two-thirds of the registrations were male couples (General Register Office 2006). Similar ratios occurred in Norway and Sweden (Andersson et al. 2006).
19.Patterson (2000) offered a review of some of the literature that explores duration among lesbian and gay couples. Our finding of slightly higher duration among gay male couples than among lesbian couples is consistent with relationship duration estimates made in studies using nonprobabilistic sampling. Blumstein and Schwartz (1983) used a sample of gay men and lesbians solicited from various public appearances by the authors, focusing on specific locations to maximize geographic diversity and draw from areas with different levels of social stigma related to homosexuality. They found longer relationship durations for gay men than for lesbians, with 61% of gay men and 78% of lesbians reporting relationships of less than five years. Kurdek (1988) and Kurdek (1998) used samples drawn from respondents to advertisements in gay periodicals and found average cohabitation lengths of 7.5 and 10.9 years for gay men, respectively, versus 5.0 and 7.1 years for lesbians. Kurdek (2006) used the Blumstein and Schwarz data and reported mean cohabitation lengths of 5.8 years for coupled gay men and 3.9 years for lesbians.
20.At the time of the Tobacco Survey, state registration in California did not confer all of the state-level rights and responsibilities of marriage, though legislation that took effect in 2005 to a large degree equalized marriage and domestic partnership registration.
21.Interestingly, we do not find much evidence that the measurement of sexual orientation in our California data is related to the estimated fraction of gay men and lesbians who are partnered. Specifically, the 2003 and 2005 waves of the CHIS include information on both self-reported sexual orientation and self-reported same-sex sexual behavior for the same respondents, thus allowing a direct comparison of our preferred method of identifying sexual minorities (i.e., self-reports) to the method previously used by Black et al. (2000) and Black et al. (2007) (i.e., same-sex sexual behavior). In results not reported but available upon request, we found behavior-based partnership estimates that were very similar to those based on self-reports. In part, this is because sexual behavior and sexual orientation are highly correlated, particularly among males. In the Tobacco Survey data, we can perform a related exercise by using information on whether the respondent had sex with any same-sex partners in the past year (though we do not observe whether the respondent’s sexual behavior in the past year was exclusively with same-sex partners in these data). As with the CHIS, we find very similar partnership estimates for males using behavior and orientation, though the partnership estimates for females are lower when we use behavior than when we use orientation. Given the slightly different ways that same-sex sexual behavior is assessed in the CHIS and the Tobacco Survey, we do not make too much of these differences. However, a sizable fraction of women who exhibited same-sex sexual behavior in the past year did not self identify as lesbian (44% of women with same-sex sexual behavior in the Tobacco Survey), and only 14% of these women are partnered under our definition compared with 70% of women who exhibited same-sex sexual behavior in the past year and who concurrently identified as lesbian.
22.Carpenter (forthcoming) used large samples of individual level data from Canada in 2003 and 2005 and found partnership estimates of 31.4% and 38.9% for gay men and lesbians, respectively.
23.As of July 2008, same-sex couples can marry in Massachusetts and California. Civil unions and domestic partnership registries in Connecticut, New Hampshire, New Jersey, Oregon, and Vermont are designed to offer the benefits and responsibilities of marriage to same-sex couples. Domestic partner registries in the District of Columbia, Hawaii, Maine, and Washington provide same-sex couples with some rights but are not functionally equivalent to marriage. Couples who are in civil unions or registered partnerships in these states have no way of designating their legal status on a standard marital status question, even though the statutes creating these laws often explicitly equate the status to marriage.