The States and the Differing Impetus for Divergent Paths on Same-Sex Marriage, 1990-2001
Barclay, Scott, Fisher, Shauna, Policy Studies Journal
Since July 1, 2000, when the law allowing civil unions went into effect in Vermont, until January 4, 2002, 3,471 civil unions involving parties from 48 U.S. states and several foreign countries were issued by the state of Vermont (Vermont Civil Union Review Commission, 2002). Civil Unions extend to gay and lesbian couples "all the same benefits, protections and responsibilities under law, whether they derive from statute, administrative or court rule, policy, common law or any other source of civil law, as are granted to spouses in a marriage" (15 V.S.A. [section] 1204(a)). While differing in nomenclature--and subsequently in social status--from marriage, civil unions represent the most comprehensive effort by a U.S. state to date to incorporate committed same-sex romantic relationships into an institution that parallels heterosexual marriage. Thus, it is the closest legal representation to same-sex marriage.
The fact that Vermont considered same-sex marriage at this time was neither original nor unusual. Between 1990 and 2001, state courts in Hawaii, Alaska, and Vermont introduced the possibility of judicially ordering their respective state to sanction same-sex marriage,(1) and Nebraska, Minnesota, Illinois, Hawaii, and Rhode Island all introduced bills before the legislature in support of same-sex marriage. (2) Simultaneously, 6 states held referendums or initiatives, and 29 separate states passed laws precluding same-sex marriage from being performed within their respective states.
To date, research on the politics surrounding sexual-orientation issues has focused predominantly on antidiscrimination policies at the local, state, and national levels. Based on the findings in regards to antidiscrimination policies, it has previously been argued (e.g., Haider-Markel & Meier, 1996, Wald, Button, & Rienzo, 1996) that states and localities are more likely to be supportive of lesbian and gay rights when there is high urban density, greater racial diversity, partisan alignment towards the Democrats, and reduced Christian religious activity. Yet as the Vermont case demonstrates, such findings in relation to antidiscrimination policies may not effectively capture the factors that generate and influence state laws in relation to same-sex marriage. Vermont lacks the urban density, racial diversity, partisan alignment, and religious configuration that have been previously correlated with the increased likelihood of a state or locality supporting gay and lesbian rights through policies and statutes.
Notwithstanding Vermont's introduction of civil unions, the vast majority of state legislation on same-sex marriage during this period actually involved proscribing such activity. Therefore, in this article, we focus on the impetus for the enactment of state laws that bar the celebration of marriages involving lesbian and gay couples. Using a regression model of pooled time series data from the 50 states in the period 1990 to 2001, we consider the validity of existing political, demographic, and social movement factors that have been previously associated with the introduction of laws concerning sexual orientation at the state and local level. In addition, we introduce new factors that might apply uniquely to explaining legislative hostility to same-sex marriage.
The Difference that Marriage Makes
While much of the existing empirical research has focused on the factors that provide the impetus for introducing (or precluding) antidiscrimination policies in various localities and states (e.g., Haider-Markel & Meier, 1996, Wald et al., 1996), there are two reasons to believe that the same factors may not be key to understanding the impetus for the proscription of same-sex marriages. First, same-sex marriage policies are potentially more prone to invoke a morality-based response than are antidiscrimination laws involving sexual orientation. Antidiscrimination laws are based on an extension of the logic of the civil rights movement (Wald et al., 1996). Such laws introduce sexual orientation into the category of existing social characteristics, such as race and gender, which are legally restricted from invidious discrimination in the allocation of employment, public accommodation, and housing, as well as access to governmental benefits and financial institutions. These laws recognize that homosexuality has been historically utilized as a basis for unequal political and legal treatment of individuals and attempt to preclude the continued enforcement of such discriminatory practices.
In contrast, same-sex marriage laws require positive action by a state to (a) publicly sanction the committed, romantic, and sexual relationships of gay and lesbian couples, and (b) offer to those couples the same benefits, obligations, and social status accorded to their heterosexual counterparts. Such activity requires each state to engage in redefining the parameters of romantic and sexual relationships. Historically, religious institutions have defined these relationships (Herman, 1997). This raises the possibility that a greater percentage of each state's population will make hostile judgments concerning same-sex marriage based on existing religious proscriptions.
Second, the stakes for state recognition and endorsement of same-sex marriage are considered greater, both by supporters and opponents, than in the area of antidiscrimination laws. In this way, it is more of a lightning-rod issue. Eskridge (2002, p. 2) has argued that marriage is "the ultimate goal, the crowning achievement" for gay and lesbian groups. State recognition of same-sex marriage accords equality with heterosexual couples in the area of romantic relationships of same-sex couples. Such equality includes both the legal obligations and rights currently associated solely with heterosexual marriage in most states, (3) but it also includes the social status that accompanies such legal rights.
Unlike antidiscrimination laws that are restricted to particular acts and are triggered by selected contexts related directly to the sexual orientation of the party, same-sex marriage offers gay and lesbian couples access to a comprehensive package of legal rights and predefined social status related only to their romantic relationship and independent of their sexual orientation. For this reason, same-sex marriage has consistently been a fundamental demand of lesbian and gay rights' groups since the late 1960s (see Eskridge, 1996, pp. 73-74; Eskridge, 2002, pp. 6-7), and same-sex marriage legal claims have a history that extends back to 1971. As early as 1972, "when the National Coalition of Gay Organizations drew up a list of demands for law reform, it included a key item" (Eskridge, 2002, p. 5)--same-sex marriage. A similar definition of the impact of same-sex marriage has made it a particularly targeted area for groups opposed to all forms of public recognition of homosexuality. For example, Green (2000, pp. 128-129) notes these latter groups have successfully used their opposition to the possibility of positive sexual-orientation laws, such as same-sex marriage, as one of the primary means through which to define their organizations as well as recruit new members.
The more visceral public and interest group reaction to same-sex marriage highlights the possibility that different and potentially additional factors may be found in the consideration of the impetus for policies in this area than to antidiscrimination laws. Such differences may alter the dominant factors that have traditionally been found to influence the likelihood of jurisdictions introducing sexual-orientation laws by bringing to the front new factors previously not identified as dominant in shaping such laws.
A Short Note on States, Laws, and Same-Sex Marriage
It is important to recognize for our purposes that states have several institutional means through which to express a policy regarding same-sex marriage. Our consideration in this article will be on the creation of promulgated laws or published court opinions that are recognized as establishing formal state law. In this regard, the law or court decision must apply throughout the whole state and definitively articulate the sanction of the state in supporting or proscribing same-sex marriage. Using this definition, we consider (a) traditional state laws passed by a legislature and allowed to become law by the action or inaction of the state governor, (b) the decisions of a state court that created binding law in this area, and (c) the legal incorporation of the results of a referendum or initiative.
For our purposes, a formal opinion of the state's highest legal officer, normally the attorney general, will not be treated as equivalent to a law, because it is not immediately legally binding in any of the states under consideration. Similarly, although the policies of the Department of Health (which oversees the criteria for marriage in many states) or the similarly designated bureaucratic agency create the initial governmental hurdle to being issued a marriage license for same-sex couples (along with requisite numeric, cosanguinity, residency, mental competency, and age requirements for all parties seeking marriage), we treat such decisions as only a bureaucratic statement (even if it involves an administrative rule) on this issue until it is formalized by a state court or by legislative action directly on this topic. (4) Our definition also proposes that a state must definitively proscribe or endorse the recognition of same-sex marriages. Unlike Eskridge (1999), we propose that riders to antidiscrimination laws that state that the extension of coverage to sexual orientation cannot be construed as an endorsement of same-sex marriage are not legal statements that definitively preclude state action on same-sex marriage in general.
Finally, it should be noted that there are actually two state actions (even if they are often, but not always, enacted as part of a single law by state legislatures) involved in the issue of same-sex marriage. First, a state can sanction or proscribe the performing of a same-sex marriage and the issuing of licenses to a gay or lesbian couple within its own jurisdiction. Second, a state can recognize the validity of a same-sex marriage performed in another state, whether or not such a marriage could have been legally performed in the first state. Traditionally, rather than same-sex considerations, this latter distinction has generated court cases related to age and cosanguinity requirements of new state residents seeking to transfer rights associated with existing marriages performed in other jurisdictions. In all of these areas, states generally have a more complex legal relationship to the issue of recognition of another jurisdiction's parameters for marriage than exists for their own in-state criteria. In this article, we focus on state laws concerning in-state issuance of marriage licenses.
In our analysis, we treat the states as being independent actors unconstrained by the 1996 Federal Defense of Marriage Act (28 USCS [section] 1738C--hereafter known as DOMA). First, DOMA does not constitutionally preclude states from choosing to perform or recognize same-sex marriages in their own jurisdiction; instead it focuses on the conditions for recognition of such marriages performed in other jurisdictions. While clearly of important symbolic value, the federal law does not change the well-established state power over the conditions for recognizing a marriage performed in a different jurisdiction--an area in which most states can point to extensive existing state court precedent related to numeric, cosanguinity, residency, mental competency, and age requirements. Second, the variation among the states in terms of the years that they introduce laws proscribing same-sex marriage and the fact that over one-fifth of all states did not proscribe same-sex marriages indicates that DOMA was not a definitive statement capturing the sentiment of many individual states. Finally, the legislative impact of DOMA on the 50 states is conflated temporally with the eminent possibility of a court-ordered decision (resulting from Baehr v. Lewin [74 Haw. 530]) allowing same-sex marriage in Hawaii.
The Impetus for Same-Sex Marriage in the States
To explore the impetus for state laws on the issue of same-sex marriage, we created a database that incorporated political, demographic, and social factors for each of the 50 states for the period 1990 to 2001. The chosen period is not problematic because although 11 states--California, Idaho, Indiana, Louisiana, Maryland, Nevada, New Hampshire, Oklahoma, Virginia, West Virginia, and Wyoming--passed laws to preclude same-sex marriage within their own jurisdictions in the years prior to 1990, many of these same states reconsidered or passed new laws on this issue between 1990 and 2001 in response to events in Hawaii, Alaska, and Vermont. Therefore, we consider that the major legislative, judicial, and gubernatorial activities in most states around this issue began after 1990.
We consider whether each state passed a law during this period that proscribed state recognition of same-sex marriages performed within its jurisdiction. This created a simple dichotomous variable, Rejection of Same-Sex Marriage, which measured whether each state passed a law proscribing such recognition during this period. (5) This measure was compiled directly from a Lexis search of each state's codes and statutes for each of the 12 years in the period 1990 to 2001. (6) This variable was coded 1 in the year of introduction of the law and 0 in all other years. This created 600 cases based on 50 states, each tracked for 12 years on this issue. Table 1 documents whether a state enacted a law restricting same-sex marriage during this period and the year that such a law was first enacted.
In total, 29 states enacted laws (through the legislative process rather than by referendum or initiative) to preclude same-sex marriage within their own jurisdictions during this period--1 in 1993, 1 in 1994, 13 in 1996, 7 in 1997, 6 in 1998 (including Alaska for a second time), and 2 in 2000. It should be noted that during this same period, 32 states passed laws proscribing recognition of such marriages when performed in other jurisdictions, such that not all states that enacted laws on same-sex marriage saw the need to alter their existing state law on the issue of performing same-sex marriages within the state.
Based on the existing literature on the factors that influence the presence of sexual orientation policies at the local and state level, as well as some assumptions about factors likely to impact upon same-sex marriage policies, we considered six categories of variables containing a total of 21 variables. The idea of this comprehensive model is to capture all of the possible forces that might influence states to enact laws concerning same-sex marriage. It should also be noted that all variables, except the number of localities with domestic-partnership coverage, are controlled in relation to the respective state's population in order to avoid inappropriately weighting the values from the more populous states.
A. Institutional Interactions
The passage of a law in any time period is clearly influenced by the interplay of other institutional actors. We identified five factors that might influence the decisions of a legislature in this regard.
(i) Prior Decisions of the State Courts
As mentioned previously, state court decisions on same-sex marriage began appearing as early as 1971, and all were in the negative. Most such decisions were resolved negatively at the lowest level of the courts without appellate intervention. We assumed that any state court decision (at any level of the state judicial hierarchy) prior to 1990 that directly addressed any aspect of same-sex marriage could alter the need for a law in this area during the period under consideration. Therefore, we coded all state court decisions as 1 for all years between 1990 and 2001 in which a state court decision on this issue was binding and 0 for all other years. This data was obtained by a search in Lexis for published court opinions in each state (7) and the findings were consistent with Eskridge (1999, Appendix B3).
(ii) Prior Laws passed by the Legislature
New York (8) and Minnesota (9) heard their first same-sex marriage cases as early as 1971, and these and similar state court cases in the next 5 years (10) encouraged California to alter its marriage laws in 1977 to preclude same-sex marriage being performed within the state (Eskridge, 2002, pp. 10-11) However, most of the other 10 states that passed such laws simply had originally used gendered references in defining marriage. Some of these states passed their specific statutes on this issue in the 19th century to deal with numeric requirements raised by polygamous heterosexual marriages associated predominantly with the Church of the Latter Day Saints. A state that altered its marriage laws to preclude same-sex marriage prior to 1990 should have slightly less incentive to alter its laws during the current period of consideration. We coded any year from 1990 to 2001 as 1 if a state law passed before 1990 involving same-sex marriage was still active. We coded 0 for all other years. This measure was compiled from a review of each state's marriage laws as well as information from Eskridge (1999, 2002).
(iii) Prior Attorney General's Holding
While clearly not enforceable in a state court, we assumed that a holding of the attorney general made prior to 1990 and that directly addressed the state's legal position on same-sex marriage could carry substantial weight with the legislature and governor in relation to the need to alter existing state laws on this issue. We coded any year from 1990 to 2001 in which a state attorney general's holding made prior to 1990 was still valid as 1 and 0 for all other years. This data was obtained from information from Eskridge (1999) and by a search in Lexis-Nexis for published state attorney general opinions. (11)
(iv) State-wide Referendum or Popular Initiative
During the period from 1990 to 2001, six states--Alaska, California, Colorado, Hawaii, Nebraska, and Nevada-amended their state constitutions or altered existing state laws through a statewide referendum or a statewide initiative that directly addressed same--sex marriage. We assumed that such popular activity would act to diminish the necessity for related legislative activity, except where such legislative activity was associated with enacting the outcome of such an event. We coded 1 for every year that the holding of the referendum or initiative was in force in the state (for example, Colorado's Amendment 2 can only be considered as active for the period 1992 when it passed until 1996 when it was struck down by the U.S. Supreme Court) and 0 for each remaining year. This data was obtained by a search in Lexis-Nexis of state constitutions and state statutes with information from Eskridge (1999). (12)
(v) Current State Court Decisions
Obviously, a current state court decision on the validity of same-sex marriage should have a resultant impact on legislative action in this issue. For example, the legislatures in Hawaii, Alaska, and Vermont all responded directly to state court decisions involving same-sex marriage during this period. We coded as 1 every year that a post-1990 opinion of the court represented the law of the state and 0 for all other years. Data was obtained directly from a search in Lexis of published state court opinions for the period, 1990-2001. (13)
B. Partisan Politics and Elections
We used four variables to capture the expected influence of partisan politics on the likelihood of a state passing a law proscribing the recognition of same-sex marriage.
(i) Percentage of Democrats in the State House
In the period under consideration, the Democratic Party at the national, state, and local levels was more likely to support progay and lesbian policies, including the introduction of antidiscrimination policies and hate crimes laws. We coded the percentage of the legislators in a state's lower chamber that belonged to the Democratic Party for each year. This data was obtained from The Book of the States for the years 1990-2001. It should be noted that Nebraska has a unicameral legislature and a nonpartisan electoral system. For Nebraska only, the percent of Democrats in the lower chamber was coded using U.S. senatorial representation for the state on the basis that the U.S. senators represented the best statewide indicator of Democratic political influence translated into direct office holding.
(ii) Party of Governor
We also expect the partisanship of the state's governor to matter in the current context. We coded a 1 for all Democratic governors or Independents with closer ideological allegiance on issues to the Democratic Party and 0 for all Republican governors or Independents with closer ideological allegiance on issues to the Republican Party. The partisan affiliation of all governors for this period was based on data from The Book of the States for the years 1990-2001.
(iii) Election Year--Governor
Unlike earlier studies (e.g., Haider-Markel & Meier, 1996), the longitudinal nature of the current research database allows us to consider the influence of other effects on the partisan actors. In this regard, we assume that gubernatorial actions on this issue might differ in election years, especially given high interest group activity surrounding this issue. Based on the date of gubernatorial elections obtained from The Book of the States for the years 1990--2001, we coded as 1 every year that each state held an election for governor. All other years were coded 0.
(iv) Election Year--Legislature
In a corollary to the logic of the effects of gubernatorial elections, we expected that legislatures were more likely to consider this issue during an election year. Based on the date of legislative elections for the lower chamber obtained from The Book of the States for the years 1990-2001, we coded as 1 every year that each state held an election for any or all of the representatives of the lower chamber. All other years were coded 0.
C. Characteristics of each State's Population
Prior research has highlighted four key demographic elements that influence the likelihood of a jurisdiction's population being more amenable to extending gay and lesbian rights. We introduced these four demographic indicators into the equation, but we also introduced two additional independent variables that we assumed might be important in the current context. These two variables sought to give an indication of the attitude of the population of each state to both marriage and political issues in general.
(i) Level of Racial Diversity
Research by Dorris (1999) has indicated that the level of racial diversity in a city is an important indicator of the likelihood of such jurisdiction's actually introducing sexual-orientation policies. Jurisdictions with greater racial diversity are more likely to be supportive of the introduction of such policies. The assumption is that such groups are more amenable to a claim of new rights introduced by other historically oppressed groups. To capture the racial diversity of each respective state, we coded the percent of each state's population that self-reported their race in the 1990 and 2000 Census as being any racial category other than Caucasian. We used linear interpolation of percentages for the intervening years. (14)
(ii) Urban Density
Similarly, urban populations have previously been found (Klawitter & Hammer, 1999) to be more likely to be amenable to government policies that expand gay and lesbian rights than are more rural locations. It is assumed that urban environments breed greater tolerance for alternative lifestyles either through greater likelihood of encountering gay, lesbian, and bisexual residents in everyday existence or through the fact that urban environments lead residents to develop more expanded definitions of acceptable social behavior. We coded the percent of each state's population that the 1990 and 2000 Census reports as residing in an urban area. We used linear interpolation for the intervening years.
(iii) College-Educated Population
There is also strong evidence that the greater the number of persons in a jurisdiction with higher education, the more likely that jurisdiction is to support extending government recognition of sexual-orientation issues (Dorris, 1999; Haider-Markel & Meier, 1996). Like urban living, higher education is correlated with greater acceptance of social difference. We coded the percent of each state's population that the 1990 and 2000 Census reports as both over the age of 25 and in possession of a bachelor's degree. We used linear interpolation for intervening years.
(iv) The Wealth of a State
In addition to level of racial diversity, percent of college-educated population, and the level of urban density of each state, it has previously been argued that more affluent cities and counties were more likely to be supportive of gay and lesbian rights. Although Wald, Button, and Rienzo (1996) used per capita income as an indicator of wealth for cities and counties, we felt that this and similar measures, such as the median income of a state's population, were more obtuse indicators when applied to the state level. Instead, we believed that a more accurate indication of the financial position of both the state as an entity and its resident population is found in the Gross State Product per capita. We might expect wealthy states with greater largesse would be less prone to divisive fights over the allocation of resources and might be more amenable or be in a better position to respond to arguments concerning potentially divisive social issues. We coded the Gross State Product per capita of each state for each year, with the values for 2000 and 2001 generated as the mean of the surrounding series. The data were obtained from State Politics and Policy Quarterly Data Resource. (15)
(v) The Breakdown of Marriage
Judicial opinions supportive of recognition of gay and lesbian relationships in recent adoption cases involving same-sex couples have noted that "most children today do not live in so-called 'traditional' 1950 television situation comedy type families." (In the Matter of the Adoption of Caitlin, 1008) In acknowledging the potential virtues in lesbian and gay relationships, state judges tend to reference the current high divorce rates among heterosexual couples to challenge the myth that heterosexual relationships are by definition inherently stable environments. High divorce rates appear to challenge traditional definitions of family structure, and by consequence, the existing gender parameters for a committed, romantic relationship. Given that state judges highlight this factor in their discussion of the value of same-sex romantic relationships, we assumed that it was a potentially important factor for consideration by legislators. Consequently, we coded the percentage of divorced persons per state as reported in the 1990 and 2000 Census. We used linear interpolation for the intervening years.
(vi) The Ideology of the Citizenry
We also assumed that the aggregated ideology of each state's citizenry might be an important measure of their likelihood to support the preclusion of same-sex marriage, based on the logic that more liberal states might be less amenable to introducing laws that proscribe same-sex marriage. Using the citizen ideology scores generated by Berry, Rindquist, Fording, and Hanson (1998, pp. 330-331) and updated by Fording, we coded the ideology of the combined citizenry of each state on a scale between 0 and 100 where 9.25 was the most conservative value reported (for Oklahoma in 1996) and 88.3 was the most liberal value reported (for Massachusetts in 1990). The mean ideology score across all states for the listed time period was 48.3. Since citizen ideology scores have not yet been updated through 2002, values for 2000 and 2001 were generated as the mean of the surrounding series.
D. Religious Factors
Prior research has clearly shown the key role that religious affiliation plays in defining the debate in relation to sexual-orientation issues. Schroedel's (1999) study of political elites and Lewis and Roger's (1999) study of Gallup and CNN polls demonstrated the role of religion in influencing an individual's position on gay and lesbian issues. Similarly, Wald, Button, and Rienzo (1996) found a negative correlation between certain religious groups and the adoption of antidiscrimination ordinances in cities and counties. Religious groups play two related roles in this regard. First, they are important leaders in defining opinion on such "morality" issues. Second, they often act as interest groups, lobbying legislators and governors and filing amicus curiae (friend of the court) briefs with the state courts on issues such as same-sex marriage.
Consistent with Wald, Button, and Rienzo (1996), we focus on the effect on legislation of the percentage of religious adherents in each state affiliated with specific religious groups. Similar to their approach, we consider religious groups that have recently taken a public stance condemning homosexuality and same-sex marriage, such as Catholics, Mormons, and Baptists, as well as groups that have offered public support for same-sex marriage, such as Jewish groups. While religious groups hostile to same-sex marriage have received much attention (e.g., Herman, 1997), supportive religious groups were a prime factor in shaping popular and legislative opinion in the controversy over civil unions in Vermont in the years 1996 to 2001 (Barclay & Marshall, 2002). All values for percentage of religious adherents associated with a specific religious group are taken from the values as reported in Bradley et al. (1992) Church and Church Membership in the United States, 1990. Values are reported for 1990 and treated as fixed through the whole period under consideration. While there may be internal division in some member churches, we used the expressed position of the overarching organization as the basis for inclusion in our research.
(i) Percent of Hostile Religious Adherents in a State
We began with the assumption that the greater the percentage of Southern Baptist and Catholic adherents as well as members of the Church of Latter Day Saints in a state, the less controversy that might be expected concerning legislative consideration of same-sex marriage. We coded the percentage of total population of each state in 1990 who were adherents of these three religious groupings.
(ii) Percent of Supportive Religious Groups
In contrast to many other religious groups, Jewish groups have demonstrated strong support for extending the rights of lesbian, gay, and bisexual residents, including in the area of same-sex marriage. We coded the percentage of total population of each state in 1990 who were active adherents of Judaism.
E. The Presence of Lesbian, Gay, and Bisexual Residents
Haider-Markel and Meier (1996, p. 336) highlight the important role played in shaping state government policies concerning sexual orientation by lesbian, gay, and bisexual residents in a state. Similarly, Haider-Markel and Meier (1996) note the impact of interest group activity by lesbian and gay rights groups, such as Lambda Legal Defense Fund, Inc. and National Gay and Lesbian Task Force (NGLTF), on government policy. Accordingly, we introduced one measure of the number of lesbian and gay couples in a state as well as one for the interest group activity of lesbian and gay rights groups.
(i) Reported Same-Sex Couple Households
Haider-Markel and Meier (1996, p. 336) were unable to find a direct measure of the gay and lesbian population in a state and instead relied on substitute indicators, such as the number of gay bars and gay newspapers in a state. We were more successful in the pursuit of such a variable because of changes in the format of the Census in 2000, which documented for the first time the number of households that self-reported same-sex couples living together. The National Gay and Lesbian Taskforce also reported updated figures from the 1990 Census in relation to same-sex couples. (16) These 1990 figures have been used previously in similar studies (e.g., Wald et al., 1996). We combined these updated 1990 figures and the more accurate 2000 Census figures and used linear interpolation to generate the figures for intervening years. (17) Using these figures, we coded the percentage of total households in each state that self-reported as being occupied by a same-sex couple. Such a self-reported figure potentially drastically underreports the total presence of lesbian, gay, and bisexual individuals in a state since it only counts same-sex couples, not individuals. Further, this data only captures people who are willing to openly report their sexuality to the government in connection with a specific type of household living arrangement. However, it is the most accurate figure available, and it seems particularly apt in considering same-sex marriage.
(ii) Gay Interest Group Membership
To represent the level of activity within each state by interest groups working for the legal rights of lesbian, gay, and bisexual residents, we coded the percentage of the state's total population involved as members in Lambda Legal Defense Fund, Inc. (18) This variable, of which a single year version was previously used by Haider-Markel and Meier (1996), has the value of incorporating both a measure of the level of activity of the interest group in the state as well as the level of active endorsement offered by supporters (both gay and heterosexual) of lesbian and gay legal rights.
F. Prior State Action on Sexual Orientation in General
Clearly, the level of controversy associated with state action in relation to same-sex marriage is related in part to prior actions of a state or its localities on issues involving sexual orientation. In effect, these prior actions reflect the willingness of the residents of a state to engage positively with issues associated with sexual orientation. It should be noted that we do not include consideration of state sodomy laws (or their removal), as it was highly correlated with the two current measures, and a dichotomous variable associated with the presence of sodomy laws in a state lacked the nuance inherent in these two current measures.
(i) Population with Sexual-Orientation Antidiscrimination Coverage
In their study of city and county antidiscrimination laws that protect individuals based upon their sexual orientation, Wald et al. (1996) argue that the "political opportunity structure" of a jurisdiction influences whether or not it is able to adopt and maintain an antidiscrimination law. This group of factors includes previous local government action on sexual-orientation issues such as the presence or absence of an antisodomy law or a hate crimes statute that includes sexual orientation. Using a similar logic, we assume that the presence of local government and/or statewide antidiscrimination laws will promote an environment more conducive to extending rights to gays and lesbians. Our measure of the percentage of a state's population covered by antidiscrimination laws or orders protecting individuals based on sexual orientation was compiled using city, county, and state information obtained from Lambda Legal Defense Fund, Inc., in combination with city, county, and state populations obtained from the 1990 and 2000 Census data.
(ii) Number of Localities in State with Domestic-Partnership Coverage
We also assumed a connection between the number of localities within a state covered by domestic-partnership laws and a state's willingness to consider legally recognizing same-sex marriages. We felt that the ability to register one's relationship had symbolic power similar to marriage and therefore considered only those localities with domestic-partnership registries. The number of cities and counties within a state for each year with domestic-partnership registries was obtained from Lambda Legal Defense Fund, Inc. (19) This measure was not controlled for population because we believed the important aspect was the number of localities that demonstrated a willingness to engage in the political process required to successfully pass such a law, independent of the size of the city. For example, in Washington, Seattle's domestic-partnership coverage incorporates a much larger percent of the state's population, but the political efficacy of the small city of Lacey's decision to create a domestic partnership registry should not be overlooked given the fact that Lacey is the home of many state government bureaucrats working in the nearby state capital of Olympia. We believed that each state contained such a large number of localities that might potentially pass such laws that each state could be treated as similar on this measure, independent of population differences.
The Enactment of Anti-Same-sex Laws
To consider the impetus for same-sex marriage in the 50 states, we created a population-averaged logit regression model clustered on each state. Such General Estimation Equation (GEE) models, with an assumption of equal correlation structure, (20) have been shown to provide consistent estimates for pooled cross-sectional time-series data of the type used in the current research (e.g., Zorn, 2001). Although duration models have been utilized traditionally in relation to longitudinal studies, as well as applied successfully to the role of diffusion between states in relation to the introduction of sexual-orientation policies (see Berry & Berry, 1999), the presence of 10 states with marriage laws at the start of the period (and long predating any logical starting point for a logical duration model) restricted the effective use of duration models. The impact of such early legislation in these states is accounted for in the current model by the use of the dichotomous variable, prior laws on sanction, which notes the presence of a prior state law on this issue that is likely (but not necessarily in the case of a state such as West Virginia) to eliminate the need for new law on this issue to be introduced.
Using whether a state had introduced a law proscribing same-sex marriage in its jurisdiction in the period from 1990 through 2001 as the dependent variable, we introduced the 21 listed independent variables into the population-averaged logistic regression model using robust (Huber White sandwich estimator of variance) standard errors. (21) The results are reported in Table 2.
The resultant equation produces some interesting, as well as some expected, results. Obviously, we were not surprised to find that prior laws on this issue were significant in the model and decreased the likelihood of introducing an anti-same-sex marriage law between 1990 and 2001. Within this model, percent of educated persons over the age of 25 per state was also significant. This finding confirmed earlier studies about the role of education on a jurisdiction's sexual-orientation policies by demonstrating that states with a higher percentage of its population college educated were less likely to introduce such laws.
Of more interest, but consistent with our assumption concerning this variable, was the finding that there was a greater likelihood that anti-same-sex marriage laws would be passed in a year that coincided with legislative elections by the lower house of the state legislature. At first glance, one may assume a simple and spurious overlap between the timing of the activities in Hawaii that prompted some states to respond and the 1996 election year. However, it is important to remember that, as noted before, many states did not pass anti-same-sex marriage laws in 1996 and that the election pattern for representatives to the lower house in many states certainly does not mirror in any way the election cycle of the U.S. House of Representatives. Instead, this finding within the current model supports a notion that the more intensive scrutiny placed on this issue by supporters and opponents, along with the related negative public opinion, has made it an area that is especially likely to be raised in ways that coincide with legislative elections.
While it was not significant at the 0.05 level, the sign for the gubernatorial elections was negative. This may indicate that in statewide races, such as for state governor, candidates are less willing to raise this issue for fear of alienating voters in some sections of the state. In addition, in some states, governors have been openly dismissive of state restrictions on the state's lesbian, gay, and bisexual population. For example, in Washington, the governor in 1996 and 1997 vetoed a bill similar to the one that was eventually passed into law in 1998 over his veto. It appears that the electoral effects of hostility to same-sex marriage are better translated at the local level where individual representatives can play to different constituencies than at the statewide level where candidates must consider differing electoral combinations for success.
The model also produced a number of puzzling findings. Unlike earlier studies, the combined role of the two variables designed to capture the divergent impact between the states of differing religious configurations were clearly not significant in the model. One possible explanation for this finding is that contrary to our original assumption, as well as earlier research on the impetus for other forms of sexual-orientation policies, religious configurations are not as important as previously suspected. This may occur because most individuals already have adopted a predefined view of same-sex marriage, largely independent of the particular religious configuration apparent in each state. Conversely, it is possible that the real influence of religious configuration within the model is being captured through the combination of demographic and political values for each state's population.
There is also the conundrum in the current equation that the greater the number of reported same-sex couple households, the more likely a state was to pass an anti-same-sex marriage law. On first glance this appears to be contradictory--a larger presence of gay and lesbian couples leads to a state being more likely to introduce laws hostile to sanctioning these couples' relationships. To add to the enigma is the finding that states with higher numbers of cities with domestic-partnership coverage for same-sex couples were also more likely to pass an anti-same-sex marriage law. Interestingly, the sign was in a more logical direction for membership in a gay and lesbian rights organization, such that a larger number of members associated with interest groups organized to support lesbian, gay, and bisexual rights led to a state being less likely to pass a hostile antimarriage law.
One explanation is a backlash theory that plays off the difference between jurisdictional levels in each state. If we imagined that the majority of lesbian and gay couples are concentrated in a few supportive counties, cities, and towns within each state, these locations are likely to respond politically to their presence by creating local policies extending domestic-partnership coverage. In such locations, lesbian, gay, and bisexual residents would in turn be more likely to feel comfortable enough to identify themselves as part of a same-sex couple household for the purposes of the Census as well as part of a larger community. In direct contrast to supportive local communities, it is possible to posit that some state legislatures respond negatively to the presence of an active lesbian, gay, and bisexual community in these select counties, cities, and towns by acting strongly on symbolic issues such as same-sex marriage. Such action would play politically to perceived greater negative public response across the entire state on the issue of same-sex marriage. This would explain the direction of the sign for gay and lesbian interest groups in the model, as the political effect sought by the state legislature is mitigated by the activities of organized interest groups that support lesbian, gay, and bisexual rights at the state level.
This explanation seems consistent with the finding that the greater the number of localities within a state that offered domestic partnerships, the more likely the state was to enact an anti-same-sex marriage law. It appears that domestic partnerships might in theory be a stepping stone toward popular acceptance of same-sex relationships, but they cannot be identified in practice as a stepping stone toward statewide legislative endorsement of same-sex marriage. In fact, the current model suggests the contrary. It appears that a large number of localities offering domestic partnerships within a state actually encourage state legislatures to take action to preclude such partnerships from being translated into any form paralleling heterosexual marriage. Clearly, there is an element of symbolic politics in such actions, since legislatures and governors play to different constituencies than local elected officials. However, the resultant state law still has the effect of precluding same-sex marriage in the respective state.
Our findings challenge some popularly accepted beliefs about same-sex marriage in the states. While some states moved with alacrity and ease on this issue, many states moved with reticence and division. Moreover, a fifth of all states did not move at all. Thus, the states were neither monolithic in response nor always hostile in their consideration of this issue--a surprising fact given the ease with which same-sex marriage was rejected by state courts and simultaneously deemed not worthy of consideration by most state legislatures when it first appeared in the early 1970s.
Contrary to earlier research on sexual-orientation policies in localities, religious groups appeared to play a less prominent role in shaping the political debate over same-sex marriage in the states. In part, this finding might be explained by the incorporation of other demographic and political factors, but it also may be that religious configurations play a less defining role in relation to same-sex marriage, an issue about which individuals are more likely to have predefined opinions. Clearly, church groups have historically played a national role in defining the parameters of marriage, as well as shaped popular opinion on homosexuality. Similarly, it would be consistent to believe that local religious configurations would have a strong influence over the debate in cities and counties. However, it may be that popular opinion is less influenced by local religious configurations as manifested in any one individual state. Either way, this finding raises interesting ground for future research.
Surprisingly, it appears that a larger presence of lesbian and gay couples increases the likelihood associated with a state introducing anti-same-sex marriage laws. It appears that whereas local governments within states have responded to the presence of these residents with increased domestic-partnership coverage and the extension of antidiscrimination laws to sexual orientation, state legislatures have been less willing to translate the presence of such residents into positive action on same-sex marriage.
Finally, we raised a different role for the number of localities with domestic-partnership registries within a state. We showed that one factor often identified as politically important in establishing the legitimacy of same-sex relationships and their equivalence to their heterosexual counterparts actually has a negative effect in relation to statewide laws on the issue of same-sex marriage.
Overall, we presented a comprehensive picture that demonstrated differing impetus for the divergent paths that states pursued during this period in dealing with same-sex marriage. Although a number of demographic and political factors were shared, the current findings highlight differences in the factors that researchers should be considering as they turn from the examination of antidiscrimination laws introduced by localities to the area of same-sex marriage and state laws. The battle over same-sex marriage is clearly of a different political nature than earlier sexual-orientation political battles. At most, we have just begun to map the territory on which this battle occurs within the states.
Table 1. Laws Enacted Restricting Same-Sex Marriage by State by Year, 1990-2001 State 1990 1991 1992 1993 1994 1995 Alabama Alaska Arizona Arkansas California Colorado Connecticut Delaware Florida Georgia Hawaii X Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah X Vermont Virginia Washington West Virginia Wisconsin Wyoming State 1996 1997 1998 1999 2000 2001 Alabama X Alaska X X Arizona X Arkansas X California X Colorado Connecticut Delaware X Florida X Georgia X Hawaii Idaho Illinois X Indiana Iowa X Kansas X Kentucky X Louisiana Maine X Maryland Massachusetts Michigan X Minnesota X Mississippi X Missouri X Montana X Nebraska Nevada New Hampshire New Jersey New Mexico New York North Carolina X North Dakota X Ohio Oklahoma Oregon Pennsylvania X Rhode Island South Carolina X South Dakota X Tennessee X Texas X Utah Vermont Virginia Washington X West Virginia X Wisconsin Wyoming Table 2. General Estimation Equation Using a Logit Regression Model of Enactment of Anti-Same-Sex Marriage Laws in the 50 States, 1990-2001 Independent variables Coefficient (standard error) Institutional interactions Prior state court decisions -0.4258 (0.4325) Prior laws on sanction -2.695 (0.7138) ** Prior attorney general's opinion -0.3827 (0.3642) Statewide referendum or popular initiative 0.7455 (1.942) Current state court decision 0.2045 (1.568) Partisan politics Percent Democrats in House -0.0206 (0.0141) Democratic governor -0.4619 (0.4134) Election year--governor -0.9031 (0.5408) Election year--legislature 1.711 (0.5152) ** Population characteristics Percent of racial diversity 0.0095 (0.0196) Percent of urban density in 1990 -0.0103 (0.0168) Percent of college educated in 1990 -0.4576 (0.1880) * State divorce rate -0.1660 (0.1549) Gross State Product per capita 0.0500 (0.0292) Ideology of the citizenry -0.0212 (0.0195) Religious Factors Percent of population in hostile religions -0.0123 (0.01890) Percent of population in supportive religions -0.0776 (0.23921) Lesbian, gay, and bisexual presence Percent same-sex couple households 5.899 (1.989) * Gay interest group membership per capita -314.96 (124.88) * Prior state action on sexual orientation Percent of population with sexual orientation -0.0021 (0.0095) anti-discrimination coverage Number of localities in state with domestic 0.3532 (0.1110) ** partnership coverage Constant 7.579 (4.064) N = 600 Groups = 50 States Correlation Structure: Exchangeable Wald [chi square](21) = 82.23 P = 0.00 * Significant at the 0.05 level ** Significant at the 0.001 level.
(1.) Hawaii--Baehr v. Lewin 74 Haw. 530; Alaska--Brause v. Bureau of Vital Statistics, No. 3 AN-95-6562 CI, 1998 WL 88743 (Alaska Super. Ct. Feb. 27, 1998); and Vermont--Baker v. Vermont 170 Vt. 194.
(2.) See, for example, in 1996--Hawaii S.B. 3112; 1997--Illinois H.B. 389, Nebraska L.B. 407; 1998--Hawaii H.B. 1424, Minnesota H.B. 3773, Nebraska L.B. 407, Rhode Island 7994; 1999--H.B. 5517; 2000--H.B. 5517, H.B. 7589 and S.B. 2380; 2001--Rhode Island H.B. 5608.
(3.) For example, Beth Robinson, one of the lawyers representing the three couples seeking same-sex marriage in the case of Baker v. State (170 Vt. 194), noted in her oral argument in 1998 before the Vermont Supreme Court that the state of Vermont had over 1,000 legal rights and obligations that were directly tied to marital status (Barclay & Marshall, 2002).
(4.) See, for example, the New York case on this issue: Storrs v. Holcomb (1996, Sup) 168 Misc.2d. 898; Storrs v. Holcomb 245 A.D.2d. 943 (1997).
(5.) Since the only laws being recorded by the dependent variable are laws hostile to same-sex marriage, positive actions by a state are coded in the complementary. Within this coding, positive state action in this context occurs in the realization of laws endorsing an equivalent to same-sex marriage, as in Vermont in 2000, or in the failure of a state to act at all on this issue during the period, as in New York and eight other states.
(6.) State statutes were searched by year in Lexis using the following search query: "marriage/s same sex or same sex or (man/s woman) or (male/s female)."
(7.) The search consisted of searching state court opinions in Lexis using the keywords "same-sex marriage" or "same sex marriage."
(8.) The New York case was Anonymous v. Anonymous 67 Misc.2d 982 (1971).
(9.) The Minnesota case was Baker v. Nelson 191 N.W.2d. 185.
(10.) For example, the following cases were brought: Jones v. Hallahan 501 S.W.2d 588 (1973) (Kentucky) and Singer v. Hara 11 Wn. App. 247 (1974) (Washington).
(11.) The search consisted of searching state attorney general opinions in Lexis using the keywords "same-sex marriage" or "same sex marriage."
(12.) The search consisted of searching state constitutions and codes in Lexis using the keywords "marriage/s same-sex or same gender or (man/s woman) or (male/s female)." This list is more expansive than the list provided by Lambda Legal Defense Fund, Inc., on their Web site (http://www.lambdalegal.org) and differs slightly from Eskridge (1999).
(13.) The search consisted of searching state court opinions in Lexis using the keywords "same-sex marriage" or "same sex marriage."
(14.) We used linear interpolation for two reasons. First, we did not believe that the racial numbers reported in the Current Population Survey would be as accurate a representation of this category as were the more comprehensive data from the full Census count. Second, we believed that linear interpolation was a supportable hypothesis given such aggregate population characteristics. Further, this method of accounting for missing data was least likely to inappropriately influence our resultant statistical equations given our use of pooled times series data.
(15.) The SPPQ Data Resource is available at http://www.unl.edu/SPPQ/.
(16.) Figures reported on the Web site of the National Gay and Lesbian Task Force (http://www.ngltf.org). Figures for 1990 and 2000 were confirmed by information from the 1990 and 2000 Census but see the note (Technical Note on Same sex Unmarried Partner Data From the 1990 and 2000 Censuses) by the Census Bureau concerning the comparability of 1990 and 2000 data on same-sex couple households.
(17.) Although linear interpolation may not completely and accurately capture the change in same-sex couple households during this period or the likelihood of such couples self-reporting on the census, we believe it is a logically and statistically consistent way of accounting for the increase in reported same-sex households.
(18.) The 1990 values were obtained directly from Donald Haider-Markel as used in Haider-Markel and Meier (1996). The reported 2000 values were supplied directly by Lambda Legal Defense Fund, Inc., and record membership for June 2000. Values for all intervening years were interpolated. Interestingly, it should be noted that Lambda Legal Defense, Inc., has not been at the heart of the initiation of legal challenges for same-sex marriage during this period. In fact, Paula Ettelbrick, who was legal director of Lambda for the early part of this period, was one of the more outspoken opponents of same-sex marriage (Ettelbrick, 1997). Cases in Hawaii, the District of Columbia, and Vermont were all filed using private attorneys (Eskridge, 2002, pp.16-17). The Vermont case received support and co-counsel from Gay and Lesbian Advocates and Defenders (GLAD)--a New England-based lesbian and gay rights organization, but it was initiated by private attorneys in Vermont. (Barclay & Marshall, 2002).
(19.) Figures were based on information posted on their Web site (http://www.lambdalegal.org). The city and county populations were dealt with in such a way to avoid double counting.
(20.) We used an equal correlation structure because of the nature of our dependent variable. In essence, each state legislature begins each calendar year as a "clean slate" uninfluenced by bills introduced in the prior year. Although a legislature could be influenced by its own prior laws, we accounted for this possibility in the independent variables. Consequently, we believed that within each unit it was logical to assume exchangeability. We used a population averaged model because our interest was in the common factors that existed across states in relation to the impetus for same-sex marriage rather than consideration of the level of individual unit effects. We believe that such a model did not misrepresent the data structure given the nature of the data and our specified model.
(21.) All models were run in Stata 7, Special Edition.
Barclay, S., & Marshall, A. M. (2002). Supporting a cause, developing a movement, and consolidating a practice: Cause lawyers & sexual orientation litigation in Vermont. Paper presented at the annual meeting of Law & Society, Vancouver 2001.
Berry, F. S., & Berry, W. D. (1999). Innovation and diffusion models in policy research. In P. A. Sabatier (Ed.), Theories of the policy process (pp. 169-200). Boulder, CO: Westview.
Berry, W. D., Ringquist, E. J., Fording, R. C., & Hanson, R. L. (1998). Measuring citizen and government ideology in the American States, 1960-1993. American Journal of Political Science, 42, 327-348.
Bradley, M. B., Green, N. M. Jr., Jones, D. E., MacLynn, & McNeil, L. (1992). Churches and church membership in the United States 1990, Atlanta: Glenmary Research Center.
Dorris, J. B. (1999). Antidiscrimination laws in local government: A public policy analysis of municipal lesbian and gay public employment protection. In E. D. B. Riggle & B. L. Tadlock (Eds.), Gays and lesbians in the democratic process (pp. 39-61). New York: Columbia University Press.
Eskridge, Jr., W. N. (1996). The Case for same-sex marriage. New York: The Free Press.
Eskridge, Jr., W. N. (1999). GayLaw: Challenging the apartheid of the closet. Cambridge, MA: Harvard University Press.
Eskridge, Jr., W. N. (2002). Equality practice: Civil unions and the future of gay rights. New York: Routledge.
Ettelbrick, P. (1997). Since when is marriage a path to liberation. In R. M. Baird & S. E. Rosenbaum (Eds.), Same-sex marriage: The legal and moral debate, edited by Robert M. Baird and Stuart E. Rosenbaum. New York: Prometheus Books.
Green, J. C. (2000). Antigay: Varieties of opposition to gay rights." In C. A. Rimmermann, K. D. Wald, & C. Wilcox (Eds.), The politics of gay rights (pp. 121-138). Chicago: University of Chicago Press.
Haider-Markel, D. P., & Meier, K. J. (1996). The politics of lesbian and gay Rights: Expanding the scope of the conflict. Journal of Politics, 58, 332-349.
Herman, Didi. (1997). The antigay agenda: orthodox vision and the Christian Right. Chicago: University of Chicago Press.
In the Matter of the Adoption of Caitlin (1994), 163 Misc.2d. 999, p. 1008.
Klawitter, M., & Hammer, B. (1999). Spatial and temporal diffusion of local antidiscrimination policies for sexual orientation. In E. D. B. Riggle & B. L. Tadlock (Eds.), Gays and lesbians in the democratic process (pp. 22-38). New York: Columbia University Press.
Lewis, G. B., & Rogers, M. A. (1999). Does the public support equal employment fights for gays and lesbians? In E. D. B. Riggle & B. L. Tadlock (Eds.), Gays and lesbians in the democratic process (pp. 118-145). New York: Columbia University Press.
Schroedel, J. R. (1999). Elite attitudes toward homosexuals. In E. D. B. Riggle & B. L. Tadlock (Eds.), Gays and lesbians in the democratic process (pp. 89-117). New York: Columbia University Press.
Vermont Civil Union Review Commission Report (2002). Vermont Civil Union Review Commission.
Wald, K. D., Button, J. W., & Rienzo, B. A. (1996). The politics of gay rights in American communities: Explaining antidiscrimination ordinances and policies. American Journal of Political Science, 40, 1152-1178.
Zorn, C. J. W. (2001). Generalized Estimation Equation Models for correlated data: A review of applications. American Journal of Political Science, 45, 470-490.
Scott Barclay is an Associate Professor in the Department of Political Science at the University of Albany: SUNY.
Shauna Fisher is a doctoral student in the Department of Political Science at the University of Washington.…
Questia, a part of Gale, Cengage Learning. www.questia.com
Publication information: Article title: The States and the Differing Impetus for Divergent Paths on Same-Sex Marriage, 1990-2001. Contributors: Barclay, Scott - Author, Fisher, Shauna - Author. Journal title: Policy Studies Journal. Volume: 31. Issue: 3 Publication date: August 2003. Page number: 331+. © 1999 Policy Studies Organization. COPYRIGHT 2003 Gale Group.
This material is protected by copyright and, with the exception of fair use, may not be further copied, distributed or transmitted in any form or by any means.