The Meaning of Marriage to Same-Sex Families: Formal Partnership, Parenthood, Gender, and the Welfare State in International Perspective

Lesbian and gay male couples formalize romantic partnerships at different rates. These rates differ differently depending on national context, with female partnerships exceeding male partnerships in U.S. contexts, while gay male partnerships exceed lesbian partnerships in some European contexts, and rough parity is reached in Canada. Through a cross-national analysis of welfare regimes, partnership forms, and patterns of formal same-sex partnership in eight nations and three U.S. states over time, we explain these gendered differences in formal partnership rates. We conclude that forms of relationship recognition interact with parental rights, either to facilitate or impede the achievement of social and legal parental ties between same-sex couples and their children. As formal partnership forms for same-sex couples change over time, their relative appeal to male and female couples changes as well.

State recognition of same-sex relationships, whether through marriage, civil unions, de facto schemes or other official forms of relationship recognition have become increasingly common globally (Paternotte and Kollman 2013). Yet researchers have consistently noted a "gender gap" in same-sex formal partnership rates (Ramos, Goldberg, and Badgett 2009;Rothblum, Balsam, and Solomon 2008;Rydström 2011;Solomon, Rothblum, and Balsam 2004). In industrialized democratic welfare states, female partnerships exceed male couple partnerships 1 in U.S. contexts, while male partnerships exceed female partnerships in some European contexts, and rough parity is reached in Canada. In this paper, we explain cross-national differences in the gendered patterns of relationship recognition over time.
Scholars posit three explanations for same-sex couples' decisions to formalize their relationships: differences in the benefits that attach to formal relationship recognition; differences in the rate of parenthood between male and female couples; and the structure and provisions of the social safety net. Each of these explanations sheds some light on how inequality, gender, and national context shape the individual conditions and experience of lesbian and gay life. However, these explanations cannot sufficiently explain cross-national differences or changes in formal gendered partnership rates over time. We argue that whether or not relationship-recognition schemes facilitate or impede, the establishment of ties between parents and children explain initial differences in the gendered rates of partner recognition. Over time, as relationship-recognition policies become more facilitative of parental ties, we see shifts toward gender parity or female majorities in partner recognition rates even in countries with early male majorities.
In the next section, we discuss the three existing theoretical explanations for gender differences in same-sex marriage rates and their limitations. We develop a more complete explanation that illustrates that shifts in gendered rates of partnerships over time are highly correlated with changes in the content of relationship recognition policies, specifically as they influence parental ties with children, across eight countries and three U.S. states. We highlight how changes to local partnership forms influence their relative appeal to male and female couples. This study has implications for LGBT activism, illustrating that relationship recognition alone may not sufficiently protect families with children and, in some cases, may actually make having children and establishing legal ties with them more difficult.

Marriage Benefits for Couples
In addition to its symbolic value, the material benefits of the marital contract generally encompass three main areas: (i) laws that build on assumed emotional relationships between spouses, such as the ability to make financial and medical decisions for one's partner should he or she become incapacitated; (ii) regulation of the financial relationship between married people and the state, such as taxes, social security, student stipends, or child benefits; and (iii) assumptions about paternity and parental status, regulations addressing parenting, such as custody, visitation, and child support (e.g., Chambers 2001). Forms of relationship recognition that are available to same-sex couples, however, vary considerably. Thus, many scholars have argued that variation in benefits offered by different relationship-recognition forms explains differences in the rates by which different-sex and same-sex couples access formal recognition of their relationships and may account for the gendered relationship recognition gap between male and female couples. We separate material benefits and protections for the couple from benefits related to parenting since, as we will demonstrate, they are often delinked in law and policy.

Social Discrimination and Limits of State Recognition
Research on the United States, Europe, and Canada illustrates that same-sex couples are much less likely to seek formal recognition of their relationships M. Bernstein et al. than heterosexuals because of the second-class status of formal relationshiprecognition policies. For example, in the United States, as of this writing, states can still refuse to recognize same-sex marriages performed in other states (Badgett, Gates, and Maisel 2008). Furthermore, inequalities persist throughout the United States in such areas as healthcare and education even following adoption of various same-sex relationship recognition schemes (Killian 2010). In their cross-national comparison of relationship recognition schemes in Europe, the United States, and Canada, Kitzinger and Wilkinson (2004) illustrate that not only do varied civil partnership forms generally not provide the same range of rights and protections that marriage does (their shortcomings include lack of spousal immigration rights, spousal pension provisions after death, full adoption rights, and tax benefits), but that even where civil partnerships are claimed to be fully equivalent legally, the symbolic denial of marriage status renders them a second-class form of recognition.
The inadequacy of relationship-recognition schemes for same-sex couples and the fact that relationship recognition in itself will not alleviate all the forms of discrimination faced by lesbians and gay men and their families help to explain why same-sex couples do not access formal recognition of their relationships at the same rates as different-sex couples. While these studies shed light on differences between same-sex couples on the one hand and differentsex couples on the other, they do not explain differences between same sex female and male couples.

Class as a Factor
It is possible that gendered differences in income and wealth between men and women and the range of benefits offered by formalized partnerships in the United States could explain the gender gap. However, not only do many committed same-sex couples choose not to formalize their relationships even where allowed, there is little evidence that the material benefits of marriage explain the gender gap in partnership rates within countries. Badgett, Gates, and Maisel (2008) hypothesized that higher income should either increase the likelihood of participation in formal same-sex partnerships among all same-sex couples, with individuals seeking to jointly protect assets, or that higher income would decrease likelihood of formal partnership, with individuals reluctant to risk dividing assets in case of relationship dissolution. However, drawing on survey data from California, they found that income affected the likelihood of formalized partnership only among gay men, with higher income increasing the likelihood that gay men would be in a registered domestic partnership. However, lesbians in this sample formalized their partnerships at a higher rate overall (with 29.1 percent of lesbian participants being in a registered partnership vs. 10.8 percent of gay male participants). Thus while the material benefits of formal partnerships for couples might increase the appeal of formal partnerships among some gay men, it cannot explain differences among lesbians or between lesbians and gay men. If gay men by virtue The Meaning of Marriage to Same-Sex Families of being men generally have more income and wealth than lesbians, we would expect that gay male couples would seek official recognition of their relationships at higher rates than female couples, but that is not the case crossnationally or over time as we will demonstrate. Others suggest that differences in parenthood rates between male and female couples may explain the gendered relationship-recognition gap.

The Parenthood Gap
Noting that material benefits granted to couples through formal partnerships cannot explain gender differences in same-sex marriage rates, Kimport (2013) draws on interviews with same-sex couples marrying in San Francisco in 2004 to argue that the U.S. "gender gap" in rates of formal same-sex marriage is better understood as a "parenthood gap." Kimport asserts that though both lesbians and gay men face similar impediments to parenthood as a result of their sexuality, lesbians, by virtue of being able to give birth, are less physically constrained around procreation. Kimport notes that in California, 34.4 percent of female couple households have children, compared to 20.2 percent of male couple households (Simmons and O'Connell 2003) and it follows that lesbians would be more likely to see marriage as a path, not to gay liberation, but to social validation, tolerance, and safety for their families.
The simultaneous trend of higher rates of lesbian parenthood and marriage are consistently evident in U.S. data. Rothblum, Balsam, and Solomon (2008) found that across all three states surveyed, Vermont, Massachusetts, and California, approximately 30 percent of women had children, compared with 14 percent of men. Solomon, Rothblum, and Balsam (2004) found that within their representative sample of same-sex civil union recipients, roughly one-third of all lesbian subjects reported having children, meanwhile 17.9 percent of gay men with civil unions and 9.7 percent of gay men without civil unions reported having children (see also Schecter et al. 2008).
Interviews with U.S. same-sex couples find that parent couples are eager to marry in order to provide stability and validation to their children (Porche and Purvin 2008) and lesbian couples with children emphasize that formalized partnerships could eliminate the costs engendered by a burdensome secondparent adoption process and reduce the vulnerability of non-biological parental rights in cases of relationship dissolution or partner death (Wall 2011).
Furthermore, lesbian parents are concerned about how social devaluation of their relationships and families affects the psychological and physical wellbeing of their children (Wall 2011). Lesbian mothers' concern that their children would be teased, bullied, or simply made to feel inferior and ostracized as a result of their awareness of homophobia and heterosexism provides theoretical support for lesbian parents (and perhaps same-sex parents in general) being more apt to opt for same-sex partnership recognition even where it conveys no tangible benefits, since it provides symbolic affirmation of the strength and validity of one's family.
The parenthood gap explanation for gendered patterns of relationship recognition would lead us to expect lesbians to formalize their relationships in greater numbers than their gay male counterparts to ensure a formal legal relationship with their family, protect the parental rights of the non-biological or non-adoptive parent in the case of relationship dissolution or provide social support to their children through public recognition of their parents' relationships. However, the parenthood gap cannot explain why, in many European countries, male couples formalize their relationships in greater numbers than female couples. Rydström (2011) examines rates of same-sex formal partnership in Scandinavian contexts and notes both the initial male majority and the move to female majority over time. He explains these trends largely in terms of initial lesbian and lesbian feminist skepticism of marriage as an institution. 2 The author explains that these attitudes shifted as the availability of reproductive technologies such as in vitro fertilization (IVF) became more commonplace and social visibility and tolerance around lesbian motherhood increased, "discursive and legal changes that have weakened the feminist marriage resistance and increasingly associate same-sex couples with child-rearing" (Rydström 2011, 144). Overall, Rydström suggests that increasing rates of lesbian parenthood reduced lesbian opposition to formal partnerships, accounting for the increase in the proportion of lesbians in Scandinavia formalizing their relationships over time.
While lesbian opposition to marriage may indeed have subsided over time, as same-sex adoption became more prevalent, Rydström provides no systematic examination of the interaction between formal partnership policies and the ability of same-sex couples to establish legal ties with children.
The broader limit of the parenthood gap explanation is that it assumes that marriage and other forms of relationship recognition necessarily foster greater parental rights for same-sex couples when, in practice, relationship recognition policies vary considerably in their impact on the ability of same-sex couples to form families with children. The parenthood gap explanation also assumes that the symbolic value of marriage is the same cross-nationally and that marrying will grant cultural legitimacy to same-sex families. We will argue that, cross-nationally, this is not the case.

Social Safety Net and the Meaning of Marriage
Cross-national differences in the social safety net are related to both the symbolic and material value of marriage which might influence gendered patterns of relationship recognition. In order to explain why gay male couples sought official recognition of their relationships in greater numbers in Northern Europe than their lesbian counterparts, Rothblum (2005) suggests that given the relatively robust social safety net in Northern European nations, Northern European lesbians may feel less pressure to get married out of The Meaning of Marriage to Same-Sex Families material need, and thus more inclined to abstain from marriage as a political statement. In contrast to Rydström (2011), Rothblum (2005) suggests that radical lesbian opposition to marriage continues to be more prevalent generally in Northern European countries; however this explanation cannot explain shifts from male majorities to female majorities in those countries over time. Nonetheless, based on Rothblum's analysis, we would expect there to be a relationship between gendered patterns of marriage, the symbolic meaning of marriage, and the strength of the social safety net.
Left out of the discussion of safety net and the symbolic meaning of marriage, is whether or not states, regardless of the strength of their social safety nets, make it harder for same-sex couples to create families with children. Waaldijk (2004) asserts that while marriage and other kinds of formal partnership available in Northern European countries provide social and legal benefits that might be appealing to couples, the available relationship recognition forms do not grant the full range of parental rights guaranteed by marriage. For instance, initially, in the Netherlands, same-sex marriage was equivalent to heterosexual civil marriage in most respects, except that same-sex marriage did not extend presumption of paternity to the non-biological parent of a child born into a same-sex partnership. Thus married female partners did not automatically obtain the status of legal parent to children born to their partners, and consequently certain legal aspects of parenthood, such as inheritance rules on intestacy, did not apply. In order to obtain legal parental status, the married partner of the birth mother had to go through a formal adoption process. Same-sex parents also could not participate in inter-country adoption, a rule attributed to fear of negative backlash from other countries concerned about children living with same-sex parents. Single individuals and heterosexual married couples could adopt, just not married same-sex individuals. Thus there were no options for married same-sex partners to gain joint parental authority over children adopted from abroad. Thus in countries where formal relationship recognition reduces or makes parental rights more difficult to achieve, we expect male couples to seek recognition of their relationships at higher rates than female couples.
The literature on the gendered relationship recognition gap shows that gender differences in same-sex marriage rates can be explained at least partially by the extent to which marriage makes available desirable legal and social benefits to couples (Rothblum 2005;Wall 2011), by gender differences in same-sex couple parenting rates and the desire for family benefits and protections (Kimport 2013;Wall 2011), and by the relative strength of a nation's social safety net independent of marriage (Rothblum 2005;Waaldijk 2004). However, we argue that these explanations themselves do not sufficiently explain gendered variation in rates of marriage among male and female couples. In order to understand these differences, we must also focus on the extent to which various relationship recognition forms including marriage facilitate or impede the achievement of binding legal ties between parents and children.

Methods
In this study, we examine data for eight nations and three U.S. states that have adopted same-sex partnership forms since 1989 when Denmark became the first nation to legalize registered partnerships. We examine the extent to which each of the proposed explanatory models-couple benefits and the symbolic meaning of marriage, the parenting gap, and social safety net-explains gender differences in rates of formal partnership cross-nationally. Data are drawn from the census, relevant government reports, and survey data identified from multiple sources including secondary analyses, government web sites, research centers and advocacy organizations including GLAD and ILGA Europe. The diversity of data does present a challenge for a comparative study such as this. In addition, as Maks Banens (personal communication, 2014) points out, mobility across the U.S. states is greater than that between European states and this does influence the stability of the number of same-sex marriages in each jurisdiction and the gender of the marriage partners. While acknowledging this limitation, we believe that given the small window of time we are considering, this factor would not alter our overall analysis. Another dimension for the United States is that, until the part of the Defense of Marriage Act [DOMA] which defines marriage as limited to different-sex couples for the purposes of receiving federal recognition was overturned in 2013, state recognition did not translate into access to federal social benefits that are available to different-sex couples. 3 This factor might temper the number of same-sex couples who would avail themselves of civil marriage at the state level, although we cannot determine the extent to which this factor would influence the gender of same-sex partners who do choose to marry under these complicated circumstances.
Our choice of cases was guided largely by practicality and represents the majority of industrialized welfare democracies where some form of same-sex partnership union has been legalized by the prevailing government authority. A number of additional Western European countries legalized some form of relationship recognition during that time frame, including, for example, Portugal, Germany, and Switzerland. 4 While these cases would allow us to further test and refine our model, analysis was circumscribed by both language barriers and the simple lack of available data on gendered rates of same-sex partnership in these countries. Nonetheless, the present analysis represents the majority of cases where same-sex unions have been legalized, and allows us to create a robust theoretical model presenting the main explanatory factors for changes in gendered patterns of same-sex marriage-a model which can then be tested and refined through the collection of additional data.
We compare national data to data from three individual U.S. states. While U.S. states share sovereignty with the federal government and are linked together in a relatively weak welfare state, they also possess sufficient political autonomy to justify state-by state policy comparisons (Amenta, Carruthers, The Meaning of Marriage to Same-Sex Families and Zylan 1992). Indeed, Soule (2004) demonstrates that state-level political factors, such as the openness of the polity, electoral competitiveness, and policy history, explain adoption of same-sex marriage bans in U.S. states. Because we are interested in the interaction between couple rights, parental rights, the relative strength of the welfare state, and gendered rates of same-sex marriage, and because U.S. states have jurisdiction over marriage, they can be usefully compared to cases where such rights are determined at the national level.
We analyze three of the thirty-six U.S. states that as this article goes to press allow same-sex marriage, Vermont, Massachusetts, and California, because they have been extensively examined and were among the first states in the United States to grant such rights (Bernstein and Taylor 2013). Vermont enacted "civil unions" in 2000 (Bernstein and Burke 2013) and Massachusetts became the first U.S. state to legalize same-sex marriage in 2004. California legalized registered domestic partnerships in 2004 and briefly granted same-sex couples the right to marry in 2008, until Proposition 8 instituted a state constitutional amendment banning same-sex marriage (Rothblum, Balsam, and Solomon 2008) which was overturned by the U.S. Supreme Court in 2013 (Hollingsworth vs. Perry). Thus the three states we have selected provide the most data with which we might examine how changes in policy affect partnership rates over time.
Our data does not contain a systematic breakdown of formal same-sex partnerships by race or ethnicity and thus risks performing "race blind" or "race neutral" analysis, which ignores the way that racial inequality is intertwined with the reproduction of inequality in a larger "matrix of domination" (Collins 1990). However, because our primary goal involves extensive cross-national comparison, we are also attentive to the ways that racial meanings do not fully translate cross-nationally (Purkayastha 2012). Because hierarchies of racial inequality and racialization processes differ around the globe, it is "not always clear when and how we are to conceptualize 'race' within the intersectionality matrix" (Purkayastha 2012, 58) when we perform cross-national comparison. Thus, even with data on rates of formal same-sex partnership by gender and race, rigorous attention to the local categories of racial meaning and histories of racial inequality and stratification would be necessary to avoid the risk of imposing U.S.-centric definitions and racializing processes. While the limitations of our data simply make us unable to incorporate such analysis, this is a fruitful avenue for future research.
In order to operationalize the symbolic meaning of marriage, we adapt the analysis of Kiernan (2001) and Heuveline and Timberlake (2004), who equate the significance of marriage within nations to the extent to which cohabitation is equated with marriage, understood through the length of time couples spend cohabiting, the amount of time children spend in cohabiting vs. married couples, and the extent of childbearing outside of marriage. Based on these criteria, we divide the countries that we analyze into four groups: marriage is least important in the Northern European countries, somewhat more important in France and Canada, more important in the United States, and the most important in Spain. We examine the extent to which these crossnational differences in the relative value of marriage vs. cohabitation influence gendered patterns of relationship recognition for same-sex couples.
More specifically, in the Northern European countries, there is very little difference between cohabitation and marriage (Heuveline and Timberlake 2004;Kiernan 2001). Marriage is less important in France, where Pacte Civil de Solidarité (PACS) provide an alternative to marriage for heterosexual and same-sex couples, and in Canada (Banens and Leita 2008;Banens 2013;Heuveline and Timberlake 2004;Kiernan 2001), where common-law marriage exists as a viable alternative to marriage (Young and Boyd 2006). Additionally, heterosexual marriage in the United States is heavily intertwined in constructions of deserving citizenship, bolstering its power to act as a gatekeeping apparatus to a host of social rights and privileges. The language of "marriage" and "family values" is used to justify extensive cutbacks in social services and denigrate alternative family forms, especially poor black single mothers, as inherently unstable, deviant, and threatening to the social order. Thus only heterosexual married families are constructed as deserving of rights and resources (Kandaswamy 2008;Polikoff 2008). The relatively weak social safety net coupled with the centrality of marriage to social constructions of "strong families" in the United States imbues marriage with significant material and symbolic appeal. In Spain, marriage, as opposed to cohabitation, is still the primary route into first union for the overwhelming majority of people underscoring its symbolic and material importance. Childbearing and child-rearing outside marriage is relatively rare in Spain compared to the other countries.
The strength of the social safety net, understood in terms of welfare regimes, could also explain gendered patterns of relationship recognition for same-sex couples. Esping-Andersen's (1990) three-part model of welfare capitalism, the most commonly used typology, broadly differentiates between liberal regimes which "only react to market or family 'failures' and limit assistance to marginal or especially 'deserving' groups; social-democratic regimes which are universalistic and egalitarian, and conservative-corporatist regimes that preserve status and class differentiations" (Orloff 1993, 310). This typology has been criticized for failing to incorporate attention to the dimensions of gender and sexuality in the assessment of different welfare state regimes (Bernstein and Naples 2010;Orloff 1993) as well as for not adequately categorizing nations or capturing additional types of welfare regimes or how they have changed over time (Arts and Gelissen 2002). In our analysis, we consider whether or not welfare state frameworks, both Esping-Andersen's and alternatives, are useful in explaining approaches to relationship recognition for same-sex couples or otherwise influence gendered patterns of relationship recognition.
In the next section, we present our general model which centers the relationships between forms of relationship recognition and the ability to establish binding legal ties with children. We analyze in detail gendered variation in marriage rates between lesbian and gay male couples over time and crossnationally and link those to changes in the forms of relationship recognition policies. We find some support for regional variations in the enactment of relationship recognition policies and that welfare regime types help explain changes in the content of relationship recognition policies for same-sex couples over time. We then examine the impact of marriage's symbolic importance on gendered patterns of relationship recognition, concluding that such differences cannot explain the patterns we see.

Relationship Recognition Policies and Parental Rights
In order to understand gendered variation in rates of relationship recognition, we must look at the interaction between forms of same-sex relationship recognition and how they affect parental rights. Figures 1 and 2 illustrate the configuration of factors at time 1, when some type of relationship recognition scheme was enacted, how that scheme affected the ability of same-sex couples to establish legal ties with their children, and the resulting gendered patterns of formal recognition. Time 2 illustrates whether or not relationship recognition policies remained the same or were altered to remove parenting disincentives or even to create parenting incentives in relationship recognition laws and how this influenced the gendered distribution of relationship-recognition (for a more detailed breakdown by country, see Appendix A). We discuss these three clusters of countries in detail and then return later to figures 1 and 2 to explain the different approaches taken by these countries over time.

Anglophone
The United States and Canada can be considered liberal welfare regimes although the United States has a weaker social safety net than Canada and many legal rights adhere to married couples. Canada, in contrast, provides a legal status that is virtually equivalent to marriage and is available to different-sex and same-sex couples. Despite these differences, figure 1 illustrates that each took similar approaches to relationship recognition and parental rights. Prior to relationship-recognition, the United States had strong parental rights in the three states we examine while parental rights for same-sex couples varied across the Canadian provinces (see figure 1). Early relationship recognition schemes in the United States left these rights intact. As time passed, same-sex marriage in U.S. states facilitated legal ties between parents and their children, leading to either gender parity or female majorities in the United States. In Canada, same-sex marriage did not interact with parental rights in the same way and variability across the Canadian provinces remained. Thus lesbians in Canada had less incentive to marry than lesbians in the United States and we see continued gender parity in Canada. Out of the eight countries we examine, the United States had the strongest initial support for lesbian and gay parental rights. In Vermont, Massachusetts, and California, the enactment of legal same-sex partnerships did not necessarily extend automatic dual-parent recognition to same-sex parents, but legal provisions that allowed for dual-parent recognition, usually in the form of second-parent adoption of an existing child or joint adoption of a new child, existed in some form prior to state adoption of formalized partnerships (see Appendix A).
Tables 1-3 present the number of gay male and female couples in the 2000 and 2010 census in Vermont, Massachusetts, and California, 5 as well as the number of couples who are recognized as spouses or partners in 2010. Tables 1-3 also include data from a survey conducted by Rothblum, Balsam, and Solomon (2008) of same-sex couples one year after civil unions (Vermont, 2001), marriage (Massachusetts, 2004), and registered partnerships (California, 2004) were passed. The Vermont data shows that women initially constituted 71 percent of the couples who received civil unions, 10 percent greater than their percentage in the population, based on the 2000 census data. By 2010, lesbian couples still constituted the majority of married same-sex couples (two-thirds are female; one-third are male). However, the rates by which same-sex couples marry in Vermont is nearly equal, with 36 percent of male couples marrying and 35 percent of female couples marrying. The larger percentage of married couples that are female (66 percent) is a function of the significantly larger lesbian than gay male presence in Vermont as shown in the 2010 Census.
In Massachusetts, in 2005, female couples as a percent of married same-sex couples were 4 percent higher than their numbers in the population based on the 2000 census and in 2010, they were 5 percent more than their numbers in the population would predict, indicating that married female couples in Massachusetts continue to outnumber married male couples. In fact, female couples are choosing to formalize their relationships at a rate that is 9 percent higher than the male rate (48 percent vs. 39 percent, respectively). Finally, the pattern in California (see table 3) closely resembles Vermont, with an initial overrepresentation of female couples as a percent of married same-sex couples by 9 percent. But by 2010, female couples represent only 2 percent more of the married same-sex couples than male married couples and lesbian couples are only marrying at a 3 percent higher rate than male couples. In short, the movement in Vermont and California has been from female majorities toward parity, while lesbians are still in the majority of married same-sex couples in Massachusetts. Solomon, Rothblum, and Balsam (2004) also found that 80 percent of couples who formed civil unions in the year after they were enacted came from out of state and therefore their civil unions conferred no material benefits on them. It is likely that these couples entered civil unions in order to gain social legitimacy for their families. For in-state couples, formal partnerships do provide some degree of material benefits in all three U.S. states. Thus the relatively weak social safety net of the United States coupled with the tangible benefits and symbolic value that accrue to married couples make formal relationship recognition more desirable to female couples. Over time, the trend in Vermont and California is toward gender parity in rates of formal relationship recognition. However, the gender gap remains in Massachusetts. While the initial female majority and subsequent move toward gender parity can be explained by an initial rush to marry among female couples because of issues relating to benefits, security for one's family, and children, it is unclear why the gender gap has not narrowed in Massachusetts.

Liberal Welfare Regimes: Canada
Canada provides an interesting intermediary to the extremes of the United States and Europe. Canada is a liberal welfare regime and has a stronger social safety net than the United States, making marriage less important for gaining benefits. Canada also provides a longstanding culturally acceptable alternative to marriage for heterosexual couples that confers nearly all the same benefits that marriage does (Young and Boyd 2006). And, unlike registered partnerships or civil unions which were developed specifically for same-sex couples in the United States, common-law relationships have been developed for and utilized by heterosexual couples.
In Canada, the ease of access to same-sex parental rights varies tremendously by province or state and territory (CBC News 2012b; Edmonton Journal 2007). Because formal relationship recognition did not help or impede the establishment of legal ties with children, we see rough gender parity in the same-sex partnership rates in Canada (see table 4). We also see Canadian same-sex couples opting for marriage (16 percent for male couples and 17 percent for female couples) at much lower rates than in the United States, ranging from a low of 27 percent for gay male couples in California to a high of 48 percent for lesbian couples in Massachusetts, possibly because of the stronger symbolic value of marriage in the United States and the lack of legally recognized alternatives to marriage such as common-law recognition.
Thus, Canada occupies a middle ground between the United States and European countries from the strength of the social safety net to the available protections for same-sex partners and families, and this middle ground is made manifest by the relative parity in formal partnership rates.

Europe
European countries, whether designated as social democratic (Finland, Denmark, Sweden, Norway), corporatist (France), hybrid corporatist/social democratic or a subcategory of corporatist (Netherlands) (Esping-Andersen 1990) or a separate category of Mediterranean states (Spain) (Arts and Gelissen 2002), all took a similar approach to initial forms of relationship recognition (see figure 2). These early relationship recognition schemes made it more rather than less difficult for same-sex couples to create families with children and to secure legal ties to those children. As a result, regardless of regime type, we see male majorities in who accesses the highest forms of relationship recognition. While the European countries took a similar approach in their initial enactment of relationship recognition and its relationship to parenting rights, a modified version of Esping-Andersen's (1990) welfare regime typology is related to when these countries removed parenting disincentives from their relationship recognition policies. The social democratic countries of Northern Europe, including the hybrid case of the Netherlands, are quicker to remove parenting disincentives for same-sex couples, resulting in a shift to female majorities. France, a corporatist country, has been slower, only recently passing same-sex marriage and removing impediments to lesbian and gay parenting. So at the time of this writing, French gender patterns have not changed, but we expect that over time they will. Finally, Spain, has not yet removed parenting disincentives from its relationship recognition laws, giving some credence to welfare state typologies that view Spain as part of a fourth "Mediterranean" welfare regime.

Social Democratic
In Northern European countries such as Denmark, Norway, Sweden, Iceland, and the Netherlands, initial rates of formal same-sex partnership find male couples outnumbering female couples (see tables 5 -9). In Denmark (table 5) and Norway (table 6), we see male married or partnered couples outnumbering lesbian married or partnered couples by a ratio of 3 to 1. While there are no data regarding how many male vs. how many female couples there are in the population, it is unlikely that a population imbalance could explain away such a large male advantage. In Finland (table 7) and the Netherlands   (table 8), males are overrepresented as a percent of same-sex registered partners or spouses by 8 and 10 percent, respectively. The numbers in 1996 in Iceland (table 9) are too small to make any meaningful conclusions. So in contrast to all three U.S. states where female couples initially sought formal recognition of their relationships at higher numbers than male couples, we see the opposite in Northern Europe.
A social democratic welfare model applies to all the Northern European countries, 6 so while these nations also have adopted varied formal partnership forms since 1989, they are linked by the presence of strong social safety nets. Benefits apply more generally to individual citizens, reducing the need for formal partnership as a means of accessing social benefits. More importantly, the initial relationship recognition schemes in these countries made it more rather than less difficult for same-sex couples to create families with children and to secure legal ties to those children. Parents in these countries have little need of the financial benefits of marriage because of their nations' robust social safety nets, and thus it follows that if they cannot achieve parental status through marriage, why bother? Indeed, it makes sense that some same-sex parents (or potential parents) in these countries would consider marriage unnecessary, or that the presence or desire for children would actually act as a disincentive to formalize partnerships (Waaldijk 2004).
In each case, legal provisions for establishing same-sex parental ties either took place well after the enactment of legal same-sex partnerships, have not yet taken place at all, or the formalization of legal same-sex partnerships actually came packaged with additional barriers to parenthood for same-sex couples (see Appendix A and figure 2 for a summary of this argument). Thus, not only did formal partnerships in these contexts not ease the establishment of parental ties, but in some cases actually made them more difficult. For instance, though Denmark established registered partnerships for same-sex couples in 1989, second-parent adoption was not available to same-sex parents until 1999, while joint adoption and inter-country adoption were not extended until  If we look at the gendered partnership gap over time, we see that as their policies have changed to grant increased parental rights to same-sex couples, Denmark, Finland, and Norway have all moved from male majorities to strong female majorities, varying from 6 percent in Denmark to 14 percent more female married or registered couples than male married or registered couples in Finland. In the Netherlands, marriage has gone from male dominance (10 percent) to a slight female advantage (2 percent). In Norway, the initial male advantage (74 percent male vs. 26 percent female) gave way to female majorities in marriage (37 percent male vs. 63 percent female) and in conversions to marriage (45 percent male vs. 55 percent female), though when registered partnerships are included, male couples still have a higher relationship recognition rate (52 percent male vs. 42 percent female).
In Iceland, the same 2006 act of parliament that gave same-sex couples full adoption rights also made available an additional partnership form previously only available to heterosexual couples, the consensual union. The consensual union, which operates much like the French PACS, provides a partnership form more casual than full civil marriage but that nonetheless conveys a number of social and financial benefits (Gudmundsdottir 2010). Indeed, many heterosexual couples in Iceland marry after having cohabitated with a consensual union for a few years (Statistics Iceland 2007), suggesting that it acts as a test of compatibility and a precursor to full marriage. Despite the availability of registered partnerships for same-sex couples, which became full civil marriages with the legalization of same-sex marriage in 2010 (Arce 2010), we can see from table 9 that female same-sex couples engage in civil marriage at approximately equal rates with men, but participate in consensual unions at much higher rates, leading to an overall female majority in Icelandic formal partnerships. So we see a shift either toward parity or to female dominance in these Northern European countries. In order to explain these shifts over time, we argue that changes in public policy that either linked marriage and registered partnerships with increased parental recognition or removed constraints from becoming parents facilitated these changes, thus making marriage a more attractive option for female couples. In all of these countries, parental recognition occurred after the adoption of formal partnership policies. Furthermore, many of these early partnership schemes actually made it more difficult to form families with children as same-sex couples were banned from adopting children from abroad. As a result, we suggest that these impediments to parental ties that came with early relationship recognition schemes made formal state recognition less attractive to female couples and that the presence of a strong safety net provided no further incentive to seek formal relationship recognition. As these impediments were removed and formal recognition became a way to create parental ties and families with children, the move is toward parity or lesbian dominance. Thus where the general social safety net is strong and where law governing parental rights and recognition actually creates disincentives for parents or potential parents to formalize their partnerships, the highest available form of same-sex partnership is more often favored by men, since their lower rates of parenthood render them less affected by such disincentives and female couples have little to gain in the form of benefits and much to lose in terms of parenting. As impediments to establishing legal ties to children disappear and are replaced by relationship recognition schemes that enhance parental ties with children, increasing numbers of female couples seek formal recognition of their relationships.

Conservative Corporatist
France, a conservative corporatist welfare regime, legalized a formal partnership option available to same-sex couples in 1999 in the form of the PACS. While PACS confer benefits and responsibilities similar to marriage, such as lowered income tax through joint filing and partner property inheritance, they are not fully equivalent to civil marriage, and are easier to both execute and The Meaning of Marriage to Same-Sex Families dissolve (Cody 2009). Most importantly, since only married couples can adopt in France, PACS provide no provisions for legally recognized same-sex parenthood and thus confers no advantages in terms of parenthood (BBC News 2012).
While gender data was not collected on same-sex couple PACS for the first few years after legalization, male couples in France, like their Northern European counterparts, have participated in PACS in greater numbers since 2007, when approximately 60 percent of same-sex PACS were male couples. In 2011, male couples represented 63 percent of same-sex PACS (see table 10;  INED 2013; INSEE 2011). PACS are also available as an alternative to marriage and are becoming increasingly popular for heterosexual partners who participate in PACS in far greater numbers than same-sex couples, representing 76 percent of PACS in 2000 and 96 percent of PACS in 2010 (INSEE 2011). France passed same-sex marriage in 2013, which includes adoption (Euronews 2013), thus eliminating some impediments to the achievement of binding legal ties between parents and children. As a result, we expect an increase in the number of female partnerships and a move toward greater gender parity in France.

Mediterranean
Spain is sometimes considered a conservative corporatist welfare regime (Esping-Andersen 1990) and other times part of a separate welfare regime of Mediterranean countries (Arts and Gelissen 2002). In Spain, male couples comprise 72 percent of same-sex married couples in 2005 and 67 percent in 2010. Pichardo Galán (2011) notes that the current incarnation of same-sex marriage in Spain does not provide adequate protections for parental rights. For instance, married female couples are forced to present additional documentation to ensure filiation rights of non-biological parents and to pay additional fees at fertility clinics because they don't qualify as "infertile." They cannot use insurance for reproductive services and surrogacy performed abroad provides no avenue for non-biological parent recognition (except where lesbians use the ovum of one partner and the other gives birth). Furthermore, all female and male couples are potentially subject to the rulings of homophobic judges when it comes to asserting family ties in court. Because same-sex marriage may actually act as an impediment to inter-country adoption from some regions, those who wish to become parents in this manner actively refuse marriage. Unless Spain removes impediments to same-sex couples' forming legal ties with their children, we expect male majorities to continue (table 11). In sum, same-sex formal partnership rates fluctuate greatly over time. As available partnership forms change, usually growing closer to full equivalence to civil marriage and guaranteeing more rights for same-sex couples and families, the gendered appeal of marriage changes. In particular, as same-sex parental recognition and adoption laws become less restrictive, disincentives for female formal partnerships begin to disappear. These changes suggest that as formal partnerships become less of an impediment to same-sex family formation, we will see a gradual move toward gender parity in contexts with strong social safety nets, and eventual female majority where marriage is central to ensuring legal ties between same-sex parents and children. This may explain the apparent anomaly of Massachusetts where the initial female majority has remained strong.

Discussion: Gendered Patterns of Formal Relationship Recognition
It is possible that the intertwined symbolic meaning and material consequences of marriage influence gendered patterns of relationship recognition. Table 12 illustrates that countries that place low, moderate, and high values on marriage all have initial male majorities of same-sex couples who choose to formalize their relationships. Over time, we see male majorities give way to female majorities in Time 2 for countries that place a low value on marriage, but no change in those countries with moderate and high values on marriage. Similarly, other countries that place a moderate value on marriage have gender parity which is maintained over time, while those that place a high value on marriage begin with female majorities and either remain female majority or shift to gender parity. Thus table 12 illustrates no discernible relationship between the value placed on marriage, as indicated by cohabitation, marriage and fertility patterns and gendered patterns of relationship recognition for same-sex couples. Thus the symbolic value of marriage becomes important only when analyzed in conjunction with the type of welfare state and whether or not relationship recognition policies facilitate or impede forming legal ties with children.
The type of welfare regime is correlated with how relationship-recognition policies have changed over time and their relationship to gendered patterns of relationship recognition. Figure 2 illustrates that the European countries, including the Northern European countries as well as France and Spain, all initially provide weak parental rights and marriage disincentives for same-sex couples. However, over time, as the Northern European countries removed disincentives toward parenting, we see a shift toward female majorities. Although we see no change in France or Spain, the recent removal of some disincentives to marry in France suggests that France may ultimately move in the direction of gender parity or even female majorities to the extent that marriage facilitates legal ties to children. These findings suggest that the European countries more generally, whether categorized as Mediterranean, social democratic or corporatist or both (as in the case of the Netherlands), initially provided forms of relationship recognition that produced active impediments to parental recognition resulting in male majorities. It appears that the social democratic countries (including the Netherlands) were the first to remove those impediments, followed by the corporatist country of France. Given that Spain is even more of a laggard in removing those incentives, it appears that Spain may indeed be a part of a fourth type of welfare regime as many theorists suggest (Arts and Gelissen 2002). As figure 1 illustrates, when the liberal welfare states enacted relationship recognition policies, these countries either had strong parental or varied parental rights, but, either way, relationship recognition policies did not make achieving legal ties with children more difficult. This resulted in either a female majority in the United States which attaches more material and symbolic importance to marriage or gender parity in Canada which provides legally recognized and socially acceptable alternatives to marriage. With no change in parental rights linked to relationship recognition over time, gender parity in Canada is maintained. In the United States, female majorities are maintained with evidence of some movement toward parity. Conversely, where the social safety net is strong or moderate and relationship recognition policies for same-sex couples come linked with disincentives toward parenting, we see male majorities. In the Northern European countries, early relationship recognition schemes came coupled with disincentives to parenting. Given that these countries also had strong safety nets that did not depend on marriage, lesbian couples had little incentive to marry and thus gay male couples formalized their relationships at higher rates than lesbian couples. Over time, as these countries removed the disincentives to parenting, lesbian and gay male couples formalize their relationships at relatively equal rates and even shift to female majorities. Similarly, the conservative corporatist and Mediterranean countries bundled their initial relationship recognition schemes with parental disincentives, resulting in male majorities. These parental disincentives remained intact, resulting in continued male majorities, though as those disincentives to parenting are removed, we expect to see a shift to gender parity or female majorities.

Conclusions
We have synthesized existing theoretical explanations for gender differences in same-sex formal partnerships, both within and across nations. To do so, we examined welfare regimes, available partnership forms, and the rights and benefits accorded to same-sex couples by these forms in eight nations and three The Meaning of Marriage to Same-Sex Families U.S. states, and compared differences in these factors to differences in gender rates of same-sex formal partnerships. We then extended our analysis over time to examine how changes in partnership forms continually affect the gendered appeal of marriage. Overall, we conclude that cross-national gender differences in rates of formal same-sex partnership can be attributed to the same-sex parenthood gap working in concert with local welfare regime type, especially the relative strengths of a given social safety net, and, most importantly, the extent to which relationship-recognition policies impede or facilitate the achievement of binding legal ties between parents and children. Thus, gender and parenthood alone are only partial explanatory mechanisms for these phenomena. Indeed, a fuller explanation must account not only for the ways that physical and social constraints create disparate rates of lesbian and gay parenthood, but the ways that different legal and economic conditions shape family needs and forms of relationship recognition, including the institution of marriage itself, cross-nationally and over time.
While we are not able to examine the impact of class on gendered marriage rates, the literature that we reviewed as well as our findings regarding changes over time suggest that class differences between male and female couples cannot account for the gendered patterns of difference. Nor can our data speak to the ways that gendered patterns of relationship recognition intersect with race. The impact of the intersection of race, class, and gender on the patterns of relationship recognition should be investigated in future studies.
There is nothing inherent in gender that makes marriage more appealing to male or female couples. Given the gendered history of marriage and its negative consequences for women, we might expect lesbian couples to opt out of marriage at greater rates than male couples. But while there is a widespread critique of marriage and the marriage equality movement among feminists and queer activists (Bernstein and Taylor 2013), the option to act on that critique may carry with it disparate consequences depending on public policies. For example, the strong symbolic meaning attached to marriage in the United States coupled with the security which marriage provides families may make it more costly for those lesbians who are politically opposed to marriage to opt out. The strength of the social safety net, the cultural status of marriage, and whether or not relationship recognition policies enhance or impede parents' legal ties to their children, explain the gendered gap in relationship recognition rates and the costs of opting out. These explanations are further related in that welfare systems and benefits likely reflect a country's cultural and symbolic values regarding marriage. The political, economic, social and cultural context in which marriage is defined, supported, regulated, and celebrated as a family form shapes gendered decision-making. In order to understand this multidimensional context, both relationship recognition and parental rights need to be brought into view in future analyses of same-sex marriage. Male 15 Full equivalence to civil marriage. 9 Extension of parental rights over children born to married lesbian couples and second-parent adoption with marriage. Twenty-seven 3-year probation period for step-parent adoption and inter-country adoption ban end in 2005. 28 Sources: (1) Esping-Andersen (1990) 1. We generally use the term "male" and "female" couples rather than "gay male" and "lesbian" couples when referring to married same-sex couples so as to avoid making the possibly erroneous assumption that married or coupled female partners necessarily identify as "lesbians" or that married or coupled male partners necessarily identify as "gay." 2. Lesbian feminist critiques of marriage center on the ways marriage systematically advantages men at the expense of women. Historical criticism emphasizes husbands' control of wives' property and earnings, over children, and over women's bodies (Echols 1989;Stein 1997;Kerber 1998). Additional critiques highlight the gendered structure of marital life, wherein wives and husbands have different roles and duties to perform predicated on essentialist notions of biological sex differences. Even as the aforementioned legally-authorized marital inequalities have gradually been remedied, this gendered structure of marriage continues to reproduce systematic disadvantages for women. Women, for instance, perform more housework and childcare than men, even when employed full-time; men's salaries and careers are advantaged by marriage and wives' domestic labor, while women's salaries and careers typically decline after marriage; and married women suffer 3. There are over 1,000 federal benefits attached to marital status. These benefits include Social Security survivor benefits, federal tax, health, and retirement benefits for federal employees.
5. There are concerns that the U.S. Census undercounted same-sex couples (e.g., O'Connell and Lofquist 2009; Cohn 2011), but no evidence to suggest that these have led to a systematic gender bias.
6. We include the Netherlands here because its timing resembles those of the other social democratic countries, although many argue that it can be understood as a hybrid regime between the social democratic and corporatist types (Arts and Gelissen 2002).