Current Explanations for the Variation in Same-Sex Marriage Policies in Western Countries

Over the last ten years, several western countries have recognized gay marriage either by providing gay couples the same rights as heterosexual couples, or by allowing civil unions. Other western countries have not. What accounts for this variation? This paper reviews and analyzes the key demographic, institutional and cultural arguments found in the literature on the legalization of gay marriage – especially as these pertain to cross-national comparison – and raises questions about assumptions regarding the extent to which there is variation on these variables across western countries. I argue that institutional and cultural explanations are only meaningful in explaining legalization when their combinations are specified in order to shed light on favorable (or unfavorable) circumstances for policy outcomes.


Introduction
Morality policies are usually associated with a great deal of emotion, cultural and political divisions, and conflict. The legalization of gay marriage is no exception. Over the course of the last ten years, several western industrial countries have legalized gay marriage (see Kollman 2007 for a review). On January 8th, 2010, the Portuguese Parliament legalized same-sex marriage (SSM), providing gay couples full legal recognition and the same rights as heterosexual couples (I will continue to refer to this as SSM and to differentiate this from civil unions). 1 At almost the same time, Austria, a relative late-comer to gay marriage policy, granted recognition of gay civil unions. Iceland, which already recognizes civil unions, granted full legal recognition in June 2010. When Portugal legalized gay marriage, the socialist party was in power, allied with several other leftist parties that supported legalization. SSM became legal despite considerable political and public opposition. 2 Similarly, in Austria, the Social Democrats spearheaded the policy, with the help of the Greens, and the government consulted with Lesbian, Gay, Bisexual and Transgender (LGBT) organizations on proposed legislation. Austrian public opinion in favor of gay marriage has historically been among the lowest in the European Union (EU). In Iceland, a coalition of Social Democrats and Greens, and a sympathetic president, led the policy initiative, which faced little formal political opposition. There currently exists significant variation in national responses on the matter of gay marriage. What might explain this variation?
Numerous explanations about the legalization of gay marriage have been offered in the literature, which can be grouped into three general types: demographic, institutional and cultural. As more scholars begin to investigate the question of national policy responses to gay marriage, it is important to address how demographic, institutional and cultural dimensions are supposed to explain variation in policy responses. Large-N studies that employ what Ragin (1987) refers to as a variable-centered strategy have been especially useful in guiding our attention to broad national or regional characteristics that may be correlated with policy outcomes. The advantage of this approach rests in the ability to generalize across many cases. At the same time, they are often at a disadvantage because (1) they tend to focus on maximizing the number of cases more so than on the inherent nature of the cases, and (2) they often cannot account for the more proximate set of circumstances surrounding an outcome. Variablecentered approaches in cross-national work require a reasonably large sample size which often means including a highly heterogeneous group of 1) Full legal recognition refers to policy that treats married gay couples in the same way as heterosexual couples, often referred to as gender-neutral marriage policies, whereas civil unions place certain restriction on marriage. 2) Parliament rejected a petition signed by 90,000 citizens protesting against the policy.
countries. Yet, this does not seem to lend itself well to the case of gay marriage because a vast majority of countries outside the west have no intention of legalizing gay marriage. Including them in a sample truncates the variation. In addition, large-N cross-national studies often use fairly abstract variables to capture institutional and cultural variation which may not be sensitive enough to the more proximate conditions surrounding legalization.
There is thus a tension between the size and composition of the sample and the nature of the variables. Restricting the sample to western countries as a way to control for heterogeneity of cases makes conducting statistical analyses problematic for two main reasons: (1) a loss in degrees of freedom and (2) restricted variation on the independent variables. Comparative case-studies, which are fairly common in the scholarship on gay marriage, have provided valuable insights on the conditions surrounding legalization. Yet, case-oriented approaches often shy away from generalizing beyond their cases which begs the question as to how well we can draw from specific sets of circumstances to make general conclusions. Is there an appropriate middle ground and what can this middle ground reveal about variation in same-sex marriage policies?
The overall objective of this paper is to shed light on the combination of conditions that can account for the variation in national gay marriage policy in the west. First, I provide a brief review of the ways in which scholars have theorized about the effects of specific variables on the enactment of SSM policy and evaluate the purported effects of demographic, institutional and cultural variables on policy. Second, I compare western industrial countries that have not legalized gay marriage, those that recognize civil unions, and those that provide full recognition 3 in order to shed light on this variation. Third, this paper discusses the implications of existing cross-national approaches (both case-and-variable-centered), and raises questions about the evidence used to explain SSM policies in a crossnational context. I draw from Ragin's (1987) "conjunctural analysis" to highlight the ways in which the combination of institutional and cultural characteristics can shed light on policy outcomes. I also discuss the notion of "American exceptionalism" on the issue of gay marriage, as well as the effects of institutionalization of gay marriage on public acceptance.

Demographic Explanations
The singularly most important demographic explanation for why some countries recognize gay marriage while others do not is the declining significance of marriage as a social institution (i.e., "demarriage" among many western countries) as evidenced by declining marriage rates (Festy 2006). Presumably, where marriage rates are low, there is a more favorable context for legalization of gay marriage. The average number of marriages per 1000 in the population (UN, Monthly Bulletin of Statistics, April 2001) for countries with unregistered co-habitation or no recognition of gay marriage is 8.8 (mostly driven by the US, and until recently, Portugal), while it is 6.18 and 5.73 for countries with civil unions and full legal recognition, respectively. However, there are noteworthy exceptions. For instance, Italy and Greece have low marriage rates (5.4 and 5.8 respectively), and do not recognize gay marriage. Alternatively, Belgium and the Netherlands have full recognition but have higher than average marriage rates (6.6 and 6.4 respectively, very similar to Poland at 6.7 which does not recognize gay marriage). This casts doubt on the direction of the relationship. There are two related shortcomings associated with this argument. First, it does not account for attitudes or meanings regarding marriage (which I discuss later in the paper). Second, the relationship between marriage rates and the actual politics surrounding the content of legislation is unclear (e.g., why did Austria adopt civil unions and Portugal SSM?).
In addition to "demarriage," another demographic consideration is "the crisis of fertility" in many western industrialized countries (see Borten 2002). The argument, described by conservative ethicist Margaret Somerville, 4 is that "Reproduction is the fundamental occurrence on which, ultimately, the future of human life depends" and that, for a variety of reasons, gay marriage stands to dismantle an institution meant to transmit human life. In other words, low fertility rates should act as a barrier to legalization. According to the UN World Population Prospects, fertility for 2005-2010 is lowest in Poland (1.41) and highest in the US (2.05; the only western country with above-replacement fertility and nationally does not recognize gay marriage). Average fertility for countries that have SSM is 1.60 whereas countries with civil unions and no recognition is 1.72, and 1.67 respectively. This pattern is fairly weak evidence that a crisis of fertility is related to the legalization of gay marriage. In addition, the crisis of fertility argument ignores the historical trend in western countries of the decoupling of marriage from reproduction.
Although demographic explanations may serve as a basis for understanding how population changes can affect morality policy (perhaps through corresponding changes in public attitudes), they are insufficient at explaining variation in gay marriage policy responses across western countries. It may be for this reason that the majority of explanations about gay marriage politics offered in the literature are either institutional or cultural, or a combination of both.

Institutional Explanations
Institutional explanations broadly refer to the ways in which the structure of political institutions shapes policy outcomes. These explanations focus on whether there is a political opportunity for the legalization of gay marriage. Institutional variables can be operationalized very broadly. For example, democratization (Della Porta and Diani 1999), political systems (Lijphart 1999), the influence of public opinion on policy (Petry and Mendelsohn 2004), the influence of organized challengers such as social movements and labor unions (Rayside 2008;Adam 2003), and a state's capacity to act (see Kitschelt 1986; see also Kriesi, Koopmans, Duyvendak, Giugni 1992). They can also be as specific as the presence of key sympathetic political elites (Amenta, Dunleavy and Bernstein 1994;Tarrow 1998;Cress and Snow 2000), the role of local jurisdictions, political subunits and courts (Smith 2008;Larocque 2006), particular legislative histories and court rulings, and internal party politics (Larocque 2006).

Policy Histories
The literature on gay marriage has adopted a more case-study driven, historical-institutional approach which focuses on the similarities and differences in the history of cross-national policy developments regarding homosexual relations. This has shed important light on the topic. Policy developments appear to be fairly similar in countries that have in some way legalized gay marriage. As Adam (2003) argues, the policy history generally begins with decriminalization, followed by human rights policy expansions, and then gay marriage policies. The European Commission (e.g., the Dudgeon case of 1981) played an important role in striking down policies that still criminalized homosexuality in Northern Ireland, stating that these violated human rights laws (Scotland had decriminalized homosexuality a year earlier). And, as Nardi (1998) argues, the European Parliament has been very progressive on the issue of homosexual relationships. However, this begs the question as to why some countries in the EU recognize SSM while others do not (Kollman 2007 also raises this question).
An important institutional feature that may explain variation in legalization of gay marriage is prior history of gay friendly policies (although it is unclear as to whether this has to do with length of time between an existing policy and recognition of SSM, or the content of the policy; see Soule 2004 on US states). Some countries like France, Belgium, Italy and the Netherlands decriminalized homosexuality in the 18th and 19th centuries while others, like Ireland and the US (in the US, this has been left to the states with the exception of the 2003 Lawrence case) only decriminalized sodomy in the 1990s and 2000s. Italy, although among the earliest to decriminalize homosexuality does not recognize civil unions or SSM. Luxembourg and France recognize civil unions but not SSM. Greece decriminalized homosexuality well before Canada, yet Canada provides full legal recognition and Greece provides none. Similarly, Sweden decriminalized homosexuality before Canada and legalized gay marriage four years after Canada.
It may be that legislative histories beyond decriminalization are important for understanding the timing and content of SSM policies. Large-N cross-national work on policies regulating sexuality and same-sex relations seems to allude to that (Frank and Mceneaney 1999;Frank, Camp and Boutcher 2010). Others have suggested that family policies can shape SSM policy. For instance, both Norway and Sweden (both legalized gay marriage in 2009) introduced the "neutrality principle" on family policy in the early twentieth century (see Bradley 1999) which may have sown the seeds for full recognition of SSM. Finland, which has closely related policies, lagged behind and this may explain why Finland recognizes civil unions (in 2002) and not SSM. This argument also does not speak to Denmark's early recognition of civil unions (1989/1990) and not SSM.
The Scandinavian examples may suggest that early enactment of policies recognizing civil unions may actually stall or close the issue of gay marriage preventing full recognition of SSM. In fact, most countries that fully recognize SSM enacted their policies after 2004, whereas those that recognize civil unions did so before 2004. Two recent exceptions include Iceland, which has moved from recognizing civil unions to full legal recognition (2010), and Austria, which recognizes civil unions and not SSM (2010). The UK's Civil Partnership Act of 2004 which is considered an "all-butmarried" policy may have, at least temporarily, closed the SSM issue in that country, while the Irish Parliament, having recently recognized civil unions in 2011, is already considering constitutional amendments to recognize SSM. This suggests that under particular conditions, prior policy can either close the issue of gay marriage, or it may create political momentum for full legal-recognition, as might be the case in Ireland.

Cross-Sectional Comparisons of Political Institutions
When institutional variables are broadly operationalized, and the sample is limited to western countries, variation on the independent variables is severely reduced. Table 1 shows values from the Database of Political Institutions (DPI) across six variables inferred from the literature on SSM as being influential to legalization. There is no apparent pattern regarding SSM and type of political system or on executive control of houses or coalition of special interest in government, local government, democratic tenure, or polarization. This does not necessarily mean that these do not matter but it does raise questions about the utility of using broad institutional variables in explaining variation in morality policies, especially when comparing western countries.
Comparative work on Canada and the US highlight some proximate institutional differences that allude to these broader operationalizations. These countries differ on two characteristics found in the DPI: executive control of houses, and polarization. They do not, however, differ on the measure for decentralization and independence of local governments. Yet the US and Canada diverge on the role of political subunits in federal SSM policy. I briefly discuss how these countries vary on these three variables and what this means specifically for SSM.
Canada has a highly centralized federal policy-making capacity within a highly decentralized federation. In addition, unlike the US, the definition of marriage in Canada falls under federal jurisdiction, (see Larocque 2006),  although marriage as a civil status is a provincial matter. As provincial governments and lower courts began to recognize SSM (based on their interpretation of the Canadian Charter of Rights and Freedoms), increased pressure was placed on the Canadian Supreme Court and federal government to act. In the US, defining marriage has been a state issue, not a federal one. 5 US States have strong and easily amendable constitutions. Many states 5) Although the Clinton administration did get involved when it passed the Defense of Marriage Act (DOMA) in 1996 and the Obama administration declared it would not amended their constitutions to protect against any redefinition of marriage. The Clinton administration followed suit and passed the Defense of Marriage Act (DOMA). Although courts and local jurisdictions have had some influence on federal US politics (see Stoutenborough, Haider-Markel and Allen 2006), they have not had the same impact as they have had in Canada. Currently, a majority of US states ban SSM while ten states and the District of Columbia have moved in the direction of recognizing gay marriage. Six states (most recently, New York legalized gay marriage although the state already recognized marriages granted by other states) and the District of Columbia allow gay marriage, while four states, most recently Illinois, have something akin to either civil unions or domestic partnerships. Gay marriage in California is conditional on the appeals case that has yet to be heard, and both the Washington state House of Representatives and Senate have recently voted to legalize gay marriage. Unlike the US, pressure from local jurisdictions coupled with a fairly sympathetic government lead to Canada's Civil Marriage Act. The strong equality language of the Charter created a new tool for courts to challenge the government (see Smith 2008;Rayside 2008;Schwartz and Tatalovich 2009). In turn, court rulings have become important not only for politicians and social movement groups (see Smith 2005b), but also in swaying public opinion (Matthews 2005).
Canada and the US also differ on polarization according to the DPI. Canada's four parties do not generally create a fragmented and polarized environment. The Canadian political system has limited entry points for anti-gay marriage challengers, especially the conservative Christian right. In addition, Canadian voters remained uninterested in SSM. For these and other reasons, morality issues like SSM have not historically served as divisive wedge-issues (see Smith 2005a;Rayside 2008;Pettinicchio 2010).
I suggest that the effects of institutions on gay marriage policy outcomes are not likely captured through broad institutional variables -especially when examining western countries. Case studies are particularly useful in shedding light on more specific institutional arrangements. defend the policy in 2011. In addition, many federal agencies rely on some definition of marriage; for instance, the Internal Revenue Service.

Sympathetic Allies and Political Elites
What several cases in the gay marriage debate reveal is the importance of sympathetic elites and allies (e.g., Canada, Iceland, Luxembourg, Spain, and Portugal). Adam (2003) argues that more generally, labor unions (and unionization) and the presence of a labor/social democratic party act as political supporters. But what about labor unions and leftist parties explain legalization? The link between unionization and SSM reveals numerous exceptions. Using Western's (1995) categorization for union membership, the Netherlands, Switzerland and France which are low unionization countries (France's unionization levels are in fact lower than the US) have recognized SSM. On the other hand, Italy and Australia which have not recognized SSM have medium, not low, unionization. Like marriage rates, union membership may be too crude a measure. It may be, as Adam (2003) suggests, that unions have some influence on gay marriage policy (albeit indirect) because they have been champions of antidiscrimination policies particularly in the workplace. Unions may also work at the local level creating broader political opportunities for gay marriage laying the groundwork for national action. However, like those seeking to find direct relationships between social movement mobilization and SSM policy, scholars must contend with a host of mitigating circumstances such as the role of public opinion, political process, party ideology, institutional entrepreneurship, and particularly in the case of labor unions, whether they hold any political clout in a given context. Adam cites Canada as an example of union influence on gay marriage. Yet, work on gay marriage politics in Canada does not suggest that unions have had as prominent of a role as Adam seems to suggest (See Larocque 2006;Smith 2008;Rayside 2008;Pettinicchio 2010). Although some scholars have sought to explain the role of unions in promoting equality and diversity (see for instance, Hunt and Rayside 2007), the precise and direct role of labor unions in the national gay marriage policy debate, particularly in a comparative perspective, is underexplored.
Related to his discussion of unions and "American exceptionalism," Adam (2003) makes the broad assertion that the presence of a labor party facilitates enactment of gay marriage policies. Labor parties are important, and can be a way for SSM to enter into the political arena, but presumably on the condition that (a) these parties make SSM a central issue in their platforms, (b) that they have some power in the legislature, and (c) that they are not faced with great political and public opposition. Spain, Sweden, the Netherlands, Portugal, and Norway all had a version of a labor or social democratic party in power when SSM was legalized in those countries respectively. Poland, Greece, Italy, and Australia all have a labor or social democratic party but their presence has not led to recognition of SSM (although the Australian Labor Party has recently announced its support for legalization). In Italy, for instance, Prodi's center-left coalition tried to pass a bill of rights for all co-habitating couples (including same-sex couples) but failed due to considerable opposition even from within Prodi's coalition. In addition, the presence of a labor party does not determine whether countries adopt civil unions, or full legal recognition (i.e., Switzerland, the UK, Germany, France, Finland, Denmark, and more recently Austria all had labor parties in power when they granted civil unions for homosexual couples, and not full legal recognition). Calvo's (2007) analysis of Spain and Portugal's policy responses demonstrates how leftist parties can matter specifically. In Spain, the socialist party, after facing electoral decay, used SSM as a way to appeal to new voters. They therefore made SSM a central issue. This created a political opportunity for challengers that previously did not exist. In Portugal and other southern European countries, the left has not been as unified around gay marriage equality.
Finally, sympathetic political elites are not limited to socialist or labor parties. In Belgium, it was not the Worker's Party that legalized gay marriage, but rather, the Flemish Liberals and Democrats which is a liberal democratic party appealing to business and property owners. Canada and Luxembourg share a similar political story when it comes to gay marriage. In Luxembourg, the Christian Social People's Party was in power when civil unions were granted. The Christian Social People's Party, which is conservative, was largely against the policy, while the prime minister was strongly in favor. In Canada, the Civil Marriage Act was passed by the center-left Liberal Party although the New Democrats supported the policy (as did most of the Bloc Québecois). Like Luxembourg, there was strong support from Liberal Party cabinet members, while a significant number of members within the Liberal Party sided with the Conservative opposition as was made clear by a free vote in parliament (whereby elected Liberal party members were allowed by the Liberal government to contradict the party's official position in support of gay marriage).
Institutional arguments have provided important insights on the relationship between political arrangements and the adoption of gay marriage policies. However, broad institutional characterizations tend to miss the underlying process of legalization. This may explain why many scholars have turned to more cultural approaches in explaining the adoption of SSM policies, suggesting that cultural, rather than institutional differences, explain policy variations. Schwartz and Tatalovich (2009:78) claim that institutional arrangements may "facilitate or hinder political actions, but they do not dictate their content." The term political culture has been used to refer to a broad range of attitudes and beliefs in a population and as a way to fill in the blanks left by institutional arguments (see Almond and Verba 1963;Inglehart 1988;Pye 1991;Inglehart and Welzel 2005). The key feature of political culture is that it varies across countries, while remaining a fairly stable aspect within a country. An especially pertinent example of such a discussion is in Kollman's (2007) work on SSM policy convergence in western countries. She argues that "in western democracies culture is a more important variable for explaining policy outcomes than the institutional variables highlighted in the literature" (p. 333). She claims this is so because the two key variables that are important in understanding variation -legitimacy of international norms and religiosity -vary across western countries. The presence of these characteristics allows advocates of SSM to use a human rights frame which presumably aids in the adoption of gay marriage policy (an argument also made by Smith 2005b in Canada).

Civil Society, Collective Action and the LGBT Community
Although social movements -broadly defined -are often thought of as necessary for overcoming political obstacles in legalizing gay marriage, the relationship between social movements and the legislative process is often assumed rather than unpacked. Kollman (2007) argues, in part, that the proliferation of SSM policies across western countries is related to a growing link of international LGBT activists and gay rights organizations. There is no doubt that collective action on the part of outside challengers plays an important role in the battle to legalize gay marriage (see Smith 2005b;2008;Hull 2003;Smith 1999;and Stearns and Almeida 2004 for extensive discussion). In Belgium, large marches targeting politicians have been organized by the LGBT community since the late 70s. The Netherlands is home to one of the oldest gay and lesbian rights organizations in the world -COC Nederlands (www.coc.nl) was founded in 1946. In keeping with the civic nature of America, the US has countless LGBT organizations perhaps outnumbering most western countries. In fact, countries that do not recognize SSM all have active, well-established gay communities and LGBT movements including Australia, Greece, Italy, and Poland.
It is problematic to argue that SSM policy responses are directly explained by social movements. First, although a growing body of literature has sought to operationalize the role of movements in the policy process (see for example Johnson, Agnone and McCarthy 2010), this has not been an easy task. Do movements affect policy because of the presence and number of social movement organizations? The membership size of these organizations? (Or, their use of protest and/or institutional tactics?). Frank, Camp and Boutcher (2010) have recently sought to measure social movement involvement as the presence of gay and lesbian INGOs in a given country to determine their influence on criminal regulation of sex. They do not find social movements to be significant. This may not be because movements do not matter, but because their measure is too abstract or as research has shown (see Johnson 2008), because movements have an indirect impact on policy enactment by influencing what the public and government pay attention to (Baumgartner and Mahoney 2005).
Second, it is often difficult to disentangle the direct effects of social movements on policy in light of other mitigating factors like party politics, issue salience and public opinion (Burstein and Linton 2003;Burstein 2003). Kollman (2007) herself argues that the effects of transnational LGBT networks on SSM policy is mediated by such variables as public opinion, elite preferences, and religious communities. In addition, it is even more difficult to link social movement mobilization to policy content, like for instance, whether a country grants civil unions, or whether it provides full legal recognition. Third, it is problematic to argue that the presence of "gay communities" (see Nardi 1998) and social movements explains variation in western countries because by and large, all countries in the west saw the rise of "new social movements" and the growth of identity politics (see Kriesi 1989;Bernstein 2005) and have active LGBT movements. It seems that movement presence is more a constant than a variable. However, what do vary are political opportunities for social movements and the presence and strength of a countermovement, which serves to block policy (see Staggenborg 1991 for detailed discussion of the role of countermovements). The strength of a countermovement (of which the US is a prime example) derives from widely-held beliefs in a population about the nature of sex, marriage and the family (often directly linked to religion). Religious opposition appears to be fairly common in all western contexts where an organized countermovement exists (for example, the Catholic Church in Portugal, some Muslim immigrant groups and leaders, including the Imam Khalil el-Moumni, and to a much lesser extent, certain Protestant churches in the Netherlands). In other countries (e.g., Norway), however, there appears to be no real religious opposition. In Sweden, the Church of Sweden supported SSM (although the Roman Catholic and Pentecostal Churches did not). And, as Smith (2005a) and Pettinicchio (2010) point out, although religion was an important part of the countermovement in Canada, it was not as significant as it is in the US. This also suggests that conservative religious opposition in a population is not enough to create a countermovement; resources and political opportunities for mobilization are also necessary.
If social movements do play a role in SSM policy outcomes, it is important to specify how they interact with countermovements and the extent to which there is a political opportunity for LGBT social movement groups (see Soule 2004). As Bernstein (2002) notes, the goals of LGBT social movement organizations are not confined to policy, but also have cultural goals such as "challenging dominant constructions of masculinity and femininity, homophobia, and the primacy of the gendered heterosexual nuclear family (heteronormativity)." Thus, social movements may be working indirectly to affect policy by influencing public opinion.

Cultural Values and Public Preferences
SSM can be thought of as a morality policy (see Mooney 2000) and it has been noted that on moral policy questions, the public generally is (a) aware of the issue because it is salient and (b) has strong opinions about it (see Norrander and Wilcox 1999;Mooney 2000;Lewis and Oh 2008). Thus, cultural attitudes and beliefs should be fairly important in understanding legalization. For example, Nardi (1998) argues that a culture of repressive tolerance towards homosexuality in Italy as well as the notion of "la famiglia" and "machismo" help explain the ongoing struggle for legal recognition in that country. In a more systematic cross-national comparison of policies regulating same-sex relations, Frank and McEneaney (1999) use "cultural opportunity structure" to describe the ways in which individualism, and by consequence, a growing acceptance of general human rights and gender equality 6 explain why some countries are more open to same-sex relations while others ban them. According to a 2006 Eurobarometer pole, 44 percent of citizens in the EU totally agreed that homosexual marriages should be allowed across Europe. This level of support is about the same as Canada's when gay marriage was legalized (according to a poll conducted the month gay marriage was legalized in Canada, 46 percent supported gay marriage, see Pettinicchio 2010:149). However, it is clear that in countries that do have full legal recognition, public support is very high (e.g., Netherlands at 82 percent and Sweden at 71 percent) whereas European countries with no policy have much lower public acceptance (e.g., Italy 31 percent, Poland 17 percent). Portugal, at 29% in 2006 is an exception. Two interesting questions arise: first, why are countries not enacting gay marriage policy when most appear to be trending upwards in public support? And, alternatively, why do some countries that do not have remarkably high approval enact controversial policy? Second, why have countries with civil unions and with very high public support for homosexual marriage, not yet provided full recognition (e.g. Denmark and Switzerland; Ireland is also a recent case) even when recent polls suggest substantially more than 50 percent are in favor of SSM? This suggests that other variables mitigate the relationship between public preferences and legalization.
There are some clear cultural differences by policy type. The most notable difference is with opinions on homosexuality (see Table 2). Countries that fully recognize SSM are considerably more accepting of homosexuality (e.g., for "Gays as Neighbors" and "Homosexuality Justifiable"). Perhaps the most pervasive cultural variable discussed in the literature on the politics of gay marriage is religion, particularly the ways in which religion and religious organizations create obstacles to legalization. The problem is that religion has been defined and measured in numerous ways (see Table 3 for more details). With the exception of percent Catholic, having confidence in church, and membership in religious organizations, it is clear that 6) However, it is unclear as to how Frank and Mceneaney's approach is cultural, since their measures of individualism/human rights and gender equality are not based on values and beliefs held by the population, but rather, on structural characteristics of countries.  countries that do not recognize gay marriage in any way are more religious on the five other religious dimensions.
In addition, countries with no recognition are less likely to believe that marriage is an outdated institution yet, there are no meaningful differences on importance of family. This may suggest that individuals conceptualize "family" and "marriage" beyond the traditional definition or meaning. Although the link between marriage rates and gay marriage policies is tenuous (see Festy 2006), the meaning of marriage might speak better to gay marriage policy adoption (see Kiernan 2004). Interestingly, there are no differences on political orientation.
Many of the cultural differences become weaker (or non-existent) when comparing countries with civil unions and those with SSM. Countries with civil unions and with full legal recognition do not differ on questions of the family, or on religion, with the exception of religious organizational membership which is higher for countries with SSM than those with civil unions. This data suggests that religion has a much more complicated role in the gay marriage debate, and the way one conceptualizes religion can produce different interpretations.
Following Frank and McEneaney's (1999) argument that a cultural focus on the individual and individual rights facilitates liberal policy on same-sex relations, we can draw parallels to SSM and expect citizens in those countries that legalized gay marriage to hold individualistic values. The variable in the WVS that comes closest to this understanding has many missing countries. 7 But for the countries where data is available, it appears that all countries overwhelmingly have a high proportion of individuals saying that an emphasis on the individual is a good thing. In fact, the countries with the lowest percent of individuals saying it is a good thing are Portugal and Luxembourg (both at 78 percent); Portugal granted full legal recognition. Greece (93 percent) has a very high percent of individuals saying it is a good thing, and has no gay marriage rights. Thus, although I do not claim to be formally testing their hypothesis, it appears that extending the individualization argument to SSM is problematic. In addition, the US seems to be a counterintuitive case since it is always thought of as more individualistic and individual rights-oriented than "collectivist" Europe, yet does not recognize SSM (I discuss American exceptionalism in greater detail in the discussion). Frank and McEneaney (1999) also suggest that high levels of individualization are associated with increasing focus on gender and human rights. 8 Presumably, attitudes about gender and rights should be positively related to legalization. The values in Table 2 for human rights show no difference between SSM and civil unions, and that slightly more people in countries with no SSM policy say that that there is less respect for human rights 7) I use the World Values Survey (WVS) question from the times closest to enactment of a gay marriage policy (or more recent wave if none was enacted) for those countries for which data is available (I do not include this variable in the table) asking if it is a good thing, don't mind, or a bad thing if future changes in a country means more emphasis on the individual. 8) The WVS question I use asks individuals about whether there is respect for human rights nowadays. Though not quite like the measure used by Frank and McEneaney, it still captures the nature of individual attitudes about rights within the context in which they live. See Table 3 for variable information. nowadays. There are greater differences on attitudes about gender equality. Fewer people in countries that have SSM believe that when jobs are scarce, priority should be given to men when it comes to employment (Frank et al. 2010 do not find significant relationships between status of women and criminalization of sex).

Discussion
The last two years saw important changes in gay marriage policy (including changes in Iceland, Ireland, Austria and Portugal). A major difficulty in drafting this manuscript was keeping apace with policy change. If countries are enacting policies, are they doing so because cultural values and attitudes are in line with such a policy? Iceland, and Austria (but not Portugal), tend to look more like other countries that recognize gay marriage, (compared to, for instance, Italy, Greece and Poland) especially on their attitudes on gender equality, marriage, and religion. Or, are policies enacted in these countries because they have favorable institutional environments, or both? This is inherently a set of questions that requires comparative analysis. However, when demographic, institutional and cultural variables are operationalized too broadly, as is frequently the case in variable-centered large-N studies, there is not enough variation across western countries to make comparative claims about policy responses nor are these broad measures sensitive enough to meaningfully explain outcomes. Elaborating upon these variables is an important first step in shedding light on cross-national gay marriage politics.
In the introduction, I raise the question as to whether a middle ground is appropriate in understanding the legalization of gay marriage. I do not seek to provide a systematic outline of such an approach here, but my analysis lends itself to that possibility. In fact, Ragin's (1987:71) "synthesis" seeks to bring together variable-and-case-oriented methods. The advantage of this approach is that it allows for an analysis of combinatorial conditions while using a fairly large sample. It is more considerate of cases and allows for control over heterogeneity which facilitates the process of locating meaningful conjunctural conditions as they pertain to particular outcomes. As Amenta (2003:102) argues, conjunctural analyses provide a middle range approach in explaining social policy outcomes. Figure 1 shows the intersection of institutional and cultural dimensions. When the conjuncture of institutional and cultural dimensions is taken into account, it reveals cases that fit expectations as well as the significant number of countries that do not fit the cultural and institutional stories told about SSM policy enactment.
Quadrant 1 represents both favorable cultural and institutional environments. Belgium, Sweden and Norway, all in Quadrant 1, most cleanly fit with cultural and institutional explanations. Quadrant 4 represents the least favorable scenario for gay marriage where, not surprisingly, the US is located. But what Figure 1 also reveals is that a majority of cases are somewhere in the middle, having either favorable institutional conditions or unfavorable cultural ones, or vice versa. This suggests that it is not necessary to have both favorable institutional and cultural conditions for policy responses. Four countries that recognize SSM are in either Quadrant 2 or 3, and Portugal is in Quadrant 4. Canada fully recognizes SSM, but is in the right-hand column, suggesting that it does not have a favorable cultural climate. In Canada overall, the majority of the public did not favor SSM, 9 9) For instance, a poll taken in July 2005, the month gay marriage was legalized, showed that 46 percent of Canadians approved, and 51 percent opposed.

Figure 1
Cross-National Cultural and Institutional but institutional activists or entrepreneurs pursued it anyway -a situation that may not yet be present in other countries in that quadrant like Italy, Poland and Greece. Spain and the Netherlands, both of which recognize SSM, have a favorable cultural environment but unfavorable institutional arrangements. As Calvo (2007) argues, Spain's transition to democracy and the political and cultural opportunity to embrace post-materialist values aided in overcoming some of these obstacles. Portugal, despite its location in Quadrant 4, fully recognizes gay marriage, and Austria which recently recognized civil unions is also in that quadrant. Unfavorable cultural environments may also partially explain why Finland and Denmark do not have full legal recognition.

A Note on American Exceptionalism
A common theme in the literature on gay marriage (whether cross-national or case-study) suggests that the US stands alone both institutionally and culturally on the issue of gay marriage -playing into the idea of American exceptionalism (see Lipset 1996). However, this is somewhat of an overstatement. Demographically, it is true that the US has the highest marriage rate among countries included in this sample (although New Zealand and Australia are fairly high as well). At the same time, the US has by far the highest divorce rate in the sample, but the UK, Australia, New Zealand, Denmark, Canada, and Finland are also fairly high. On the institutional dimension, the US is very much like Australia, the Netherlands, France, and Spain. The US also does not have the lowest trade union density (France is lower, and Spain is just slightly higher than the US). Fewer people in Poland and Iceland think marriage is outdated than the US; the US is not the most religious country (Portugal, Poland, Italy, Iceland, Ireland, Greece, and Austria) are either the same or more religious (although it can be argued that it is not religiosity but the kind of religiosity that matters); it ranks fairly high (in fact much higher than most countries) on attitudes about gender equality; and the US is not the least tolerant of homosexual relationships (Denmark actually is slightly less tolerant). This does not mean that the US does not have any unique characteristics. In their study of the role of religion in political decision making on gay marriage in the US, Oldmixon and Calfano (2007) find that despite ideology and partisanship, legislators were highly influenced by conservative religious Protestants and Catholics in their districts. Thus, one major "exception" in the US is the work of a well-endowed, long enduring countermovement with political clout.

Coming to Terms with the Legalization of Gay Marriage
There has been a convergence on favorable attitudes towards homosexuality (not relationships specifically) across western countries, no doubt creating a favorable climate (necessary but perhaps insufficient) for SSM policy. Every country in the sample has experienced increasingly favorable attitudes towards homosexuality since 1980. 10 In 1981, the US, Italy, Spain and Ireland had the least favorable attitudes towards homosexuality, and Denmark and the Netherlands had by far the highest (the difference between the Netherlands and the US and Italy is almost three-fold). In 1990, the Netherlands still had the most favorable attitudes at 78.7 percent, followed by Iceland. Some countries, although had increasingly favorable attitudes, increased at a slow pace (i.e., the UK, Sweden, and Denmark, all had less than a 10 percent increase in a ten-year period). Other countries experienced a more substantial increase including Canada, France, Germany, Belgium, and Spain where in 1981, about 25 percent had favorable attitudes towards homosexuality which doubled to 50 percent in 1990. By 2000, of the thirteen countries for which I have data, only Poland had less than 50 percent favorable attitudes towards homosexuality. Ireland, which thirty years ago had very low approval, is another example of significant changes in public preferences on gay marriage. According to a 2008 poll, 11 84% of Irish were in favor of civil unions and 58% in favor of SSM. Ireland recognized civil unions in 2011 and is now contemplating full legal recognition.
However, favorable attitudes about homosexuality do not necessarily translate into support for gay marriage. According to a poll taken in response to Prodi's proposed rights for cohabitating couples, a great majority of Italians supported the bill, although they did not have the same level of support for SSM. Most publics have less favorable attitudes on SSM then on homosexuality more generally. This is perhaps most clearly evidenced by American public support for employment equality for gays 10) Using the WVS question about whether homosexuality is justifiable, in 1980 for countries where data was available. 11) http://www.breakingnews.ie/ireland/mhojojeyauid/ and lesbians, and the most recent support for repealing the "Don't ask, Don't tell" policy of the US military. These favorable attitudes have not translated into support for equal marriage rights in that country.
An ongoing debate in social science is whether attitudes change policy or whether policy changes attitudes, or a combination of both. Sociologists and political scientists have noted a feedback effect from policy to public opinion such that individuals come to accept the policy (even controversial policy) once enacted (Brooks and Manza 2006;Pierson 1994). According to public opinion polls 12 taken before the enactment of legislation and after the legislation, Canada and Portugal by far had the most impressive shifts in favor of SSM. In 2006, only 29 percent of Portuguese in the sample were in favor of SSM, but in a poll conducted between January 7th and January 12th (Parliament legalized gay marriage on January 8), 52 percent were in favor of SSM. Although the percent in favor in Canada was considerably higher immediately before SSM, it was still below 50 percent, whereas a poll conducted one year later in 2006, 58 percent of Canadians were in favor. Similarly, support for SSM in Spain remained at about 61 percent before and after legalization (only Belgium and Luxembourg appear to be exceptions). Polls conducted in both France and Finland years after their recognition of civil unions show that individuals are still below 50 percent in favor of SSM, and to a lesser extent, polls conducted in Germany and the UK show that individuals are split on the issue. This may explain why these countries have not yet provided full legal recognition. It may also be the case that individuals, whether originally in favor of the policy or not, simply see enactment of a policy as final, thereby closing the issue. For instance, in Canada where fewer than 50 percent of individuals favored SSM, most (62 percent according to a 2006 Environics poll) believed after legalization that the issue was settled.

Conclusion
This paper sought to evaluate the most widely used explanations for crossnational variation in SSM policy. The main contribution of this paper lies in addressing some basic assumptions found in the literature on gay marriage policy upon which cultural and institutional theories rely: that there are fundamental cross-national institutional and cultural differences that help explain the legalization of gay marriage. But what exactly are these differences? I have argued that when examining the west (in this case, 22 western countries), institutional variation is seriously compromised suggesting broad institutional factors cannot account for legalization of SSM, and we must then turn to more specific institutional arrangements within countries to help explain legalization.
An important assumption in the literature is that there is more variation on cultural dimensions (see Kollman 2007). I find that by and large, there are some important cultural differences across countries given their policy on gay marriage -for instance, on attitudes about homosexuality and religiosity. However, differences tend to be considerably smaller when comparing countries with SSM and those with civil unions.
The idea behind comparison is to be able to speak of commensurable differences and similarities in the hopes of uncovering underlying social processes and causal mechanisms. In order to successfully do this, there must be, at a minimum, some cases that are similar to each other on one dimension, and different from each other on some other dimension. An underlying problem that Skocpol (1979) alludes to, and which appears to be especially salient when examining gay marriage policy in the west, is not so much that there are few cases (or small-N), but that the cases are not fundamentally different from each other on the independent variable.
One way to increase variation is by including countries outside the west. This would allow more degrees of freedom and also increase variation across independent variables. However, so few non-western countries recognize SSM in any degree (some exceptions include South Africa, Hungary, Ecudaor, Columbia and most recently Argentina), 13 that this would not necessarily increase variation in a meaningful way, but rather truncate the variation. In other words, comparing (in statistical terms) western countries (no matter what their policy) to non-western countries where for the most part, there is no recognition of SSM and no signs of that happening any time soon, would be like comparing apples and 13) As of 2010, Mexico City allows gay marriage and gay adoption, while Mexico does not. Indeed, local jurisdictions sometimes deviate from national law and policy. For instance, some local jurisdictions in Italy provide civil partnership registries even though the Italian civil code does not recognize same-sex unions.
oranges. Is the question of interest why Sweden recognizes SSM and Iran does not, or is it why Sweden does, and Italy does not? The second question seems to be the more interesting one because it sheds light on how robust or weak mechanisms for explaining the enactment of "moral policies" might be given comparable cultural and institutional contexts.
Case studies suggest two important features of legalization that are missed in broader comparative analyses: (1) support from elite actors and specific favorable political circumstances, and (2) the timing of legalization. In both Canada and Luxembourg, key elites facilitated legalization, and in some cases actively pursued legislation despite considerable opposition from both within and outside their respective parties. A similar situation existed in Spain. In both Canada and Portugal, the parties that legalized gay marriage were originally opposed to it whereas in Iceland, there was no formal political opposition. Courts also played an important role. Yet, the judiciaries in both Canada and Spain did not legislate from the bench. Unlike prior provincial court decisions, the Canadian Supreme Court, when asked by the Liberal government, opted not to rule on the constitutionality of the definition of marriage, and in Spain, the judiciary believed that granting SSM was not required constitutionally (and actually recommended civil unions). In Portugal, the court ruled in 2006 that it is not unconstitutional to allow only heterosexual marriages, but the court also did not oppose SSM. Following the passage of SSM policy, the court ruled the SSM bill to be constitutionally valid (which is far more concrete than the court rulings in Spain and Canada).
In all the cases mentioned above, timing appears to be especially important. Cross-sectional, large-N studies do not explain why, for example, Canada and Spain legalized gay marriage in 2005, Sweden and Norway in 2009, and Portugal in 2010. Nor do they shed light on why Iceland, which granted civil unions, is now fully legalizing gay marriage, while Denmark, which has for twenty years recognized civil unions, has not. I do not claim to systematically answer this question using this data, but it does seem that broad institutional environments (that are fairly constant over time), demographic trends, and public preferences do not do a very good job at explaining the timing of SSM policy. They act as necessary contextual characteristics, but without examining their conjuncture, it is not possible to make comparative claims about timing. Although I address some of the more process-driven explanations such as political and judicial histories, the role of local jurisdictions, and changing public values, future work should seek to more systematically link these processes in a longitudinal analysis.
However, given the combination of cultural and institutional conditions, it is possible to draw certain modest general conclusions. First, demographic trends offer little insight on the adoption of gay marriage policy, unless they are tied to the meaning of marriage and fertility. Second, sympathetic political allies and entrepreneurs are crucial for legalization. Third, early forerunners (i.e., countries that recognized civil unions early) may have difficulty passing SSM policies if granting civil unions effectively closed the issue of gay marriage (especially if they face cultural obstacles). Fourth, high public approval may not be necessary for legalization, although it does facilitate the process. In addition, when gay marriage policies become institutionalized, the public may come to terms with the legislation which reduces the political costs in adopting controversial policy. Fifth, religiosity does explain variation but there are two caveats: 1) differences are considerably weakened when comparing countries with SSM and civil unions and 2) not all conceptualizations of religion matter in the same way. Finally, organized support of legalization (i.e., social movements and interest organizations), does not directly explain variation in gay marriage policies. On the other hand, the presence of active countermovements with political clout and resources does help explain the inability to enact gay marriage policies.