Abstract
This article examines how intersecting identities influence beliefs about the legality of same-sex marriage focusing on how religious, political, and sexual identities intersect with ethnicity and gender. Support for same-sex marriage has shifted considerably in the last decade. However, changes in attitudes toward same-sex marriage have not been uniform. Religious and political identities have provided a foundation for support or opposition to marriage rights, and sectarian Protestants and fundamentalist Christians and people who identify as political conservatives or with the Republican Party have opposed marriage equality. Ethnic identities intersect with religious and political identities as well as with sexual and gender identities and family factors to structure support for marriage. Using data from the 2006–2014 General Social Surveys, I examine how commitments to conservative Christianity and the Republican Party have influenced attitudes toward same-sex marriage, and how ethnicity intersects with religious, political, and other identifications to structure support for same-sex marriage.
In the wake of the U.S. Supreme Court’s 5-4 decision in Obergefell v. Hodges granting the right of marriage to same-sex couples, some analysts suggested that Obergefell signals the end of Christian right mobilization against marital and family rights for same-sex couples (Fetner 2015). However, other scholars have noted that the ruling has energized social movement organizations and activists working against equal rights for LGBT (lesbian, gay, bisexual, and transgender) persons on a variety of fronts (Meyer 2015). Indeed, within five months of the Supreme Court decision, a nondiscrimination ordinance was voted down by a decisive 61-39 majority in Houston, the fourth largest city in the United States where only 26 percent of residents are non-Hispanic whites (quickfacts.census.gov/qfd/states/48/4835000.html). Yet, as was the case in 2008 for California’s Proposition 8 (Abrajano 2010; Egan and Sherrill 2009), the ethnic context of Houston may not indicate disproportionate opposition to civil rights among African Americans, Latinos, or Asians.
In the last decade, there has been a considerable shift in public opinion on marriage rights for same-sex couples, and the majority of Americans now favor same-sex marriage—moving from 29 percent in 2004 to more than 56 percent in the 2014 General Social Survey (GSS). In the face of this rapid change in public opinion, it is unclear how the intersection of varied identities has informed increasing support for marriage rights. Studies examining how social and personal identities structure opinions about marriage equality have focused on religious and political identities that are considered foundational for structuring beliefs about family and community (cf. Sherkat, De Vries, and Creek 2010; Sherkat et al. 2011; Wadsworth 2011). Yet, some pundits have claimed that the influence of religious and political identifications on attitudes toward same-sex marriage is waning (http://www.politico.com/magazine/story/2014/07/evangelicals-gay-marriage-108608.html#.VS51x_nF_hk; http://time.com/3669024/evangelicals-gay-marriage/; http://www.huffingtonpost.com/2014/09/15/us-evangelicals-gay-marriage_n_5824174.html).
Intersectional approaches to religion and politics amplify that the influence of political and religious identities, gender, and sexuality may be quite different for groups with different levels of privilege (Wadsworth 2011). African American opposition to same-sex marriage has been shown to be largely a function of commitments to biblical inerrancy and identification with sectarian Protestant denominations (Sherkat et al. 2010; Wadsworth 2011); however, there is limited research on how opinions are structured among Latinos or Asians. I examine how the intersection between ethnicity and other identities helps structure support for marriage equality using data from the 2006–2014 GSS across four ethnic groups: (1) Anglo-whites; (2) African Americans; (3) Latinos; and (4) Asians. The analyses enable comparisons of the level of support by ethnicity, and also how ethnicity influences the impact of political, religious, gender, and sexual identities.
Foundational Identities and Marriage Equality
Intersectional approaches to research on power and domination have taken varied forms, with many scholars amplifying the importance of the intersection of identities such as ethnicity, gender, and sexuality (cf. Allison 2011; Dhamoon 2011; Hancock 2007; Hankivsky and Cormier 2011; Harris 2009; McCall 2014; Steinbugler, Press, and Dias 2006; Wadsworth 2011). On the issue of marriage equality, prior research using an intersectional approach suggests that religious and political identifications are of principal importance for structuring opinion and activism (Wadsworth 2011). Intersecting identifications are part of a larger cultural contest about the conferral of rights in the realm of family and sexuality, and the overlapping nature of cultural fields complicates how actors’ intersecting identities influence policy preferences (Hankivsky and Cormier 2011).
Family, Sexuality, and Cultural Conflict
The evolution of the contemporary debate over marriage rights for same-sex couples is grounded in the overlapping cultural structures of family/sexuality, religion, and politics. In each of these realms, cultural capital is used to maintain and contest hierarchies of domination. The overlapping nature of cultural fields enables contesting actors to transpose cultural schemas across cultural fields, and to reinterpret cultural capital in accordance with schemas derived from different cultural positions (Bourdieu 1991, 1993; Sewell 1992; Sherkat 2014; Sherkat and Ellison 1997). The hierarchy of cultural capital in the realm of family and sexuality has been fiercely contested over the last five decades. Although the patriarchal nuclear family enjoyed a long reign of domination, feminism challenged patriarchy and disconnected sexuality from reproduction (Powell et al 2010; Treas 2002). Gradually, homosexuality also became a contested issue, with LGBT people asserting the right to be included as legitimate actors in families and upholding their sexuality to be a vital aspect of their humanity. The defenders of patriarchal heterosexual privilege have not taken these challenges to their cultural authority lightly, and backlash against both feminism and LGBT rights has been considerable (Fetner 2008; Myers and Diaz 2009).
The patriarchal family enjoys considerable support in some quarters of the religious marketplace, and many religious groups in the United States use their resources to promote a patriarchal family structure. Condemnation of all forms of nonmarital sexuality is commonplace in American Christianity, but strongest among white sectarian Protestants, who also tend to embrace a literalist reading of the Bible (Sherkat 2014). A sizable body of research has consistently found that people who identify with sectarian Protestant denominations view homosexuality as morally wrong, and they are substantially less supportive of civil rights for persons who have same-sex relations compared with Americans who identify with other religious groups or have no religious identification (Bolzendahl and Brooks 2005; Burdette, Ellison, and Hill 2005; Campbell and Monson 2008; Gay et al. 2015; Hill et al. 2004; Loftus 2001; Merino 2013; Olson, Cadge, and Harrison 2006; Perry 2015; Powell et al. 2010; Powell, York-Quadlin, and Oren Levy 2015; Schnabel 2016; Sherkat et al. 2010; Sherkat et al. 2011; Whitehead 2014a, 2014b; Whitehead and Perry 2015). Notably, sectarian religious and fundamentalist religious commitments are associated with low levels of economic and educational privilege (Keister and Sherkat 2014; Sherkat 2014). And, religious support for the patriarchal family is conditioned by ethnic privilege and religious interpretations forged in ethnic experiences, and this shapes the meaning and importance of marriage and sexuality (Cohen 1999; Wadsworth 2011)
Although the religious cultural field is dominated by Christianity in the United States, the field of Christianity is contested by a wide variety of organized religious groups (Bourdieu 1991)—many of which have embraced feminism and supported alternative family relations (Edgell 2005; Edgell and Docka 2007). Indeed, several major religious denominations officially endorse same-sex marriage, including the Unitarian Universalists (since 1997), the United Church of Christ (since 2005), and more recently the Presbyterian Church (USA), the Episcopal Church, and the Evangelical Lutheran Church in America. And, congregations and activists within virtually all religious denominations have affirmed the value of same-sex relations and families, often in the face of considerable conflict (Cadge, Day, and Wildeman 2007; Cadge, Olson, and Wildeman 2008; Comstock 2002; Djupe et al. 2006; Wood and Bloch 1995). Importantly, over the last two decades, large proportions of Americans have spurned religious identification and religious beliefs (Hout and Fischer 2014). In the 2014 GSS, more than one in five Americans do not identify with a religious group, believe that the Bible is a book of fables, or do not believe in a God. These secular Americans also tend to reject the value of patriarchal family relations, and are open to accepting varied forms of family and sexuality (Sherkat 2014).
Political institutions are powerful producers of cultural knowledge and personal preferences (Armstrong and Bernstein 2008; Wildavsky 1987). In the contemporary United States, the Republican Party has embraced the patriarchal family, and has included opposition to same-sex marriage as an official plank in their Party Platform since 1992. The embrace of socially conservative issues by the Republican Party has made them more attractive for sectarian Protestants, and religious identities help structure political identification (Brooks 2002; Brooks and Manza 2004; Layman 2001). However, political conservatives and Repub-lican Party identifiers also pose a secular rationale for opposing same-sex marriage—amplifying the need to combat community disorganization, ensure economic well-being, and eliminate government subsidies which they link to nonheterosexual and nonpatriarchal family forms (https://www.gop.com/platform/renewing-american-values). However, as the American public has increasingly embraced marriage rights for same-sex couples, some commentators have claimed that Republicans are also moving to the center (Fetner 2015), hence the effect of Republican Party identification should be weakening over time.
Sexual orientations and identities are strongly linked to support for same-sex marriage, and people who identify as gay, lesbian, or bisexual are more enthusiastic about marriage equality than heterosexuals, though research examining the influence of sexual identities on support for marriage equality is limited (Lannutti 2005). Sexual orientation and sexual identity are not isomorphic, and we might expect that those who openly embrace an LGB identity will be more supportive of marriage equality than individuals who consummate same-sex relations but do not necessarily identify as LGB. Sexual identities are forged in ethnic experiences, and intersect with religious and political identifications (Cohen 1999; Harris 2010a, 2010b; Wadsworth 2011). Many LGB persons maintain religious commitments and seek to negotiate religious and sexual identifications, and ethnicity is often an important factor in maintaining these connections (Acosta 2013; Harris and Battle 2013; Schnoor 2006; Wilcox 2003; Yip and Page 2013). Sometimes this involves embracing new religious identifications and rejecting types of religious expression that conflict with sexual identifications (Sherkat 2002, 2016; Yip 2002, 2005). Religious ties may suppress outward expressions of sexual identity and may lead LGBT people to be less enthusiastic about same-sex marriage. On the contrary, religious identities may promote the value of marriage and increase support for same-sex unions. Similarly, some LGB people retain Republican and conservative political identifications, and groups like the Log Cabin Republicans have mobilized in favor of marriage equality (Rogers and Lott 1997). Yet, some LGB conservatives and Republicans remain opposed to marriage equality, as do some liberal LGB persons who argue that marriage destroys sexual freedom and undermines collective identities (Lannutti 2005).
Ethnicity and Intersecting Identifications
Ethnic identities are forged in families and communities and nurtured in the socio-political circumstances of varied ethnic groups. Advantaged ethnic groups enjoy dominance over the preferred interpretations of family/sexuality, status, religion, and gender, while relatively disadvantaged ethnic groups often must contest and problematize definitions of family/sexuality, gender roles, religious identities, and status (Burton et al. 2010). Religious institutions and identifications have been crucial for the development and maintenance of ethnic identities (Min 2010). African American ethnic identity is often tied to Protestant religious identifications(Lincoln and Mamiya 1990), and religious identification and institutions are a source and sustainer of ethnic identity for Latino and Asian Americans (Bankston and Zhou 1995, 1996; Cadge and Ecklund 2007; Ebaugh 2003; Ebaugh and Chafetz 2000; Menjívar 2003; Min 2010; Min and Jang 2015b; Pena and Frehill 1998). Family ties are most often bound through marriage, and ethnic intermarriage is less common than homogamy (Kalmijn 1998; Qian and Lichter 2007) and is further reinforced through religious homogamy (Kalmijn 1998; Sherkat 2004). The relationship between ethnicity and politics appears to be more of a function of economic interests leading disadvantaged ethnic groups to favor more liberal identities; yet, the meaning of political liberalism/conservatism and motivations for attaching to the Republican Party may be quite different for whites than for African Americans, Latinos, or Asians (Lee and Pachon 2007; Leighley and Vedlitz 1999; Lien 1994).
Ethnicity plays a crucial role in the ongoing political discussion of civil rights for LGBT persons. What is important from an intersectional approach is that ethnic identities are forged in historical connection with dominant political, religious, and family institutions, and these identities also help construct institutions like marriage (Wadsworth 2011). In 2008, this came to the fore in California, where the Presidential bid by eventual winner Barack Obama coincided with a ballot initiative (Proposition 8) to ban same-sex marriage. Some have speculated that the passage of Proposition 8 may have been motivated in part by high minority turnout and lower support for civil rights for same-sex couples among African Americans, Latinos, and Asians (http://www.washingtonpost.com/blogs/the-fix/wp/2013/03/26/how-proposition-8-passed-in-california-and-why-it-wouldnt-today/). While political scientists have generally concluded that African American and Latino support was unlikely to have been key for the passage of Proposition 8 in California (Abrajano 2010; Egan and Sherrill 2009), African Americans are substantially less likely to favor marriage rights for same-sex couples, and religious identifications and beliefs play a principal role in African Americans’ rejection of civil rights for same-sex couples (Sherkat et al. 2010). However, research on the civil rights preferences of other ethnic minorities is scant.
Using a 2006 Pew study of U.S. Latinos, Ellison, Acevedo, and Ramos-Wada (2011) reported 65 percent of U.S. Latinos opposed marriage rights in 2006, while the 2006 GSS shows 51 percent opposition and 35 percent supporting marriage in the full sample (14 percent were unsure, which was not an option in the Pew data). Diversity is also a problem when talking about broad ethnic groups like Latinos, and Ellison et al. indicated that Puerto Ricans were significantly more supportive of marriage rights than were other Latinos. Studies of U.S. Latinos have found that Latino Catholics are more supportive of same-sex marriage when compared with Latino Protestants (Chaturvedi 2014; Ellison et al. 2011). Notably, the Republican Party has actively sought to recruit both Latino and African American voters with socially conservative positions like opposition to same-sex marriage and abortion, and candidates and party officials have actively used conservative Christian religious themes for this purpose.
Asians are an even more diverse category, and few studies have examined the social and political preferences of this large and growing set of ethnicities. A substantial fraction of Asian immigrants hail from relatively conservative nations, particularly nations with high pluralities of Muslims, and there has been some suggestion of a general “Confucian” opposition to homosexuality (Adamczyk and Cheng 2015; Adamczyk and Pitt 2009). Yet, Asian immigration has been highly selective, and large numbers have come to the United States pursuing professional employment secured because of high educational attainment (Alanezi and Sherkat 2008; Amin and Sherkat 2014; Min and Jang 2015a; Sherkat 2014). Indeed, unlike African Americans and Latinos, the 2006–2014 GSS shows that large proportions of Asians are either non-Christian (29 percent) or have no religious identification (25 percent). Only a small fraction of Asians identify with sectarian Protestant denominations (less than 9 percent). It is also important that few ethnic minorities identify with the Republican Party—which secures the commitments of under 18 percent of Asians, 12 percent of Latinos, and only 5.5 percent of African Americans. There is evidence that ethnic minorities are attracted to the Republican Party for reasons other than social conservatism (Alvarez and Bedolla 2003; Wong 2000).
The centrality of religious identities for ethnic identity is most clear among African Americans and Latinos—50 percent of African Americans identify as members of sectarian Protestant groups, while 63 percent of Latinos are Catholic. Because these religious groups represent the majority of their ethnic group, participation and identification are “semi-involuntary” which leads to a weaker relationship between religious beliefs and identification and participation (Ellison and Sherkat 1995). It seems likely that in marginalized communities where religious identities are more uniform, there may be a tendency toward greater support for marriage equality. Empathy for LGBT persons among religious African Americans was also spurred by the HIV epidemic, which affected the entire community (Cohen 1999; Harris 2010a, 2010b). Hence, we should expect African American sectarian Protestants to be more supportive of same-sex unions than their white, Latino, or Asian counterparts, while we should also expect greater support for marriage equality among Latino Catholics when compared with Catholics of other ethnicities.
The intersection between ethnicity and social status may inform beliefs about marriage rights differently across ethnic groups. It is difficult to predict whether education and income will be more informative of support for marriage equality among ethnic minorities, or if income and education will have less of an effect. It is possible that social status will have a more limited impact in ethnic groups with higher status (Asians and whites), while it may have a greater impact on disadvantaged minorities (African Americans and Latinos). On the contrary, ethnic identities to disadvantaged groups may blunt the positive impact of education or income on support for marriage equality (Wadsworth 2011).
Finally, ethnicity may intersect with family, gender, and sexual identities to inform support for marriage equality, and research consistently shows that married respondents are less supportive of same-sex marriage compared with those who have never married or who have experienced divorce (Merino 2013; Olson et al. 2006; Perry 2015; Sherkat et al. 2011; Whitehead 2014a, 2014b). Yet the importance of marital status likely varies across ethnic groups with different rates of marriage and divorce. Previous examinations consistently show substantially greater support for same-sex marriage among women in the general population (Merino 2013; Olson et al. 2006; Perry 2015; Sherkat et al. 2011; Whitehead 2014a, 2014b; Whitehead and Perry 2013) and for Latinas (Chaturvedi 2014; Ellison et al. 2011). However, gender differences in support for same-sex marriage have been found less substantial among African Americans (Sherkat et al. 2010). Quantitative research examining support for same-sex marriage by sexual identity is lacking, and this study will provide the first examination of how sexual orientation and identification intersect with ethnicity and gender to influence support for marriage equality.
Data and Measures
Data from the 2006–2014 GSSs are analyzed to examine trends and predictors of support for same-sex marriage. The GSS has continuously asked about same-sex marriage since 2004; however, the critical item on beliefs about the Bible is on an opposing interview schedule in 2004. The GSS is the only high-quality nationally representative survey which asked questions about same-sex marriage across a substantial time period. Because questions about religious beliefs, attitudes toward same-sex marriage, and indicators of sexual orientation and identification can appear on different interview schedules in different years of the GSS, the effective sample size varies depending on which indicators are included in the models. To make certain that estimates are robust, I present multivariate analyses in three separate models: (1) baseline models that exclude sexual orientation and sexual identification (n = 6,088); (2) models controlling for sexual orientation (n = 4,508); and (3) models controlling for sexual identification using the 2008–2014 surveys (n = 4,022). Across all models, missing values on all variables are treated as missing, with the exception of income which is set to the mean.
Support for Same-sex Marriage
I examine a single ordinal indicator of support for same-sex marriage. Respondents were asked to what extent that they agree or disagree with the statement, “Homosexual couples should have the right to marry one another,” with responses coded (5) strongly agree, (4) agree, (3) neutral, (2) disagree, and (1) strongly disagree.
Religious Factors
I investigate the impact of religious identification, religious attendance, and biblical fundamentalism and secularism on support for same-sex marriage. Three focal religious identities are examined: (1) sectarian Protestant (e.g., Baptist, Pentecostal, Nazarene, Church of Christ), (2) Catholic, and (3) No religious identification. These three identifications are distinguished from a mostly mainline and other Protestant comparison category. Because of the small size of several identification groups across ethnicities, a more fine-tuned set of religious identifications is not feasible; however, these three identifications capture most of the distinctiveness on beliefs about same-sex marriage.
Frequency of religious attendance is measured by a single-item indicator in which respondents are asked, “How often do you attend religious services?” with responses ranging from (0) never to (8) several times a week. Biblical fundamentalism and secularism are operationalized with dummy indicators for the responses: (1) “The Bible is the actual word of God and is to be taken literally, word for word” and (2) “The Bible is an ancient book of fables, legends, history, and moral precepts recorded by men.” Compared to “The Bible is the inspired word of God, but not everything in it should be taken literally” and a small number of “Other” responses. In the ethnic-specific models, only the dummy for biblical fundamentalism is used because of distribution issues for some ethnicities.
Political Identification and Beliefs
I examine two political variables, one tapping party identification and the other examining liberal-conservative self-rating. Individuals who identify as Republican are compared with all other respondents using a dummy indicator. Political conservatism is measured using an item asking respondents whether they “think of themselves as liberal or conservative” with responses ranging from (1) extremely liberal to (7) extremely conservative.
Sexual Orientation and Identity
Sexual orientation is indicated using a question asking respondents about the sex of their sexual partners in the last five years (male, female, or both), and together with self-reported sex, I created an indicator for those who reported having sex with someone of the same sex. In 2008, the GSS added a question on sexual identity—asking, “which of the following best describes you? (1) gay or lesbian; (2) bisexual; or (3) straight/heterosexual?” Because of the relatively small numbers of gays, lesbians, and bisexuals, I do not examine differences between bisexuals and gays or lesbians, though the intersectional models by gender reveal some differences between the latter two groups.
Sociodemographic Control Variables
I control for a host of demographic characteristics. Preliminary analyses revealed some cohort differences in support for same-sex marriage; however, these differences were relatively small and mostly confined to whites. Because the ethnic minority groups have much smaller numbers, estimating cohort parameters is inefficient relative to a linear control for respondent’s age. I examine a binary indicator comparing females to males. Separated and divorced respondents and those who have never married are compared with married and widowed respondents. A relatively small proportion of respondents are widowed, and they are concentrated in older ages and hold similar views on same-sex marriage to married respondents. I include a measure of the number of children living in the household at the time of the survey. Family income is examined using a standardized measure across years of the GSS (CONINC) with missing values set to the mean. The dollar value for income is divided by 10,000 in the multivariate models. Educa-tional attainment is assessed using an ordinal indicator of degree attainment ranging from (0) no high school degree, (1) high school degree, (2) associate’s degree, (3) bachelor’s degree, to (4) graduate degree. I also use dichotomous indicators for respondents who are lifelong residents of rural areas, and for respondents who have continually resided in the South.
Analytic Strategy
Descriptive statistics are presented for support for same-sex marriage and other covariates for the full sample and by ethnic group. Next, I present sets of heterogeneous ordinal logistic regression models in the full sample, over time, across ethnic groups, and by gender. These models estimate an error variance parameter for groups so that the estimates are more comparable; hence, it gives a better indication of ethnic variation, changes over time, and differences by gender than a regular ordinal logistic regression model (Mood 2010; Williams 2009). This method is the current state-of-the-art for comparing models across groups using full interactions, and models for time comparisons, ethnic comparisons, and gender comparisons parallel the models presented in Williams (2009:549: Table 5, Column 4).
Findings
Table 1 presents the descriptive statistics by ethnicity, and shows an upward pattern of support for same-sex marriage in every ethnic group. African Americans are consistently less supportive of marriage rights than are other ethnic groups—with the exception of Latinos in 2010. Latino and Asian support fluctuates more from year to year; however, support increased from 37 to 51 percent among Latinos, and from 42 to 64 percent among Asians. In several years of the study, Asians have the highest support of any ethnic group (2006, 2010, 2014), and in 2012, support is highest among Latinos.
Descriptive Statistics by Ethnicity.
Note. SSM = same-sex marriage.
Table 2 presents exponentiated estimates from three sets of heterogeneous ordinal logistic regression models (baseline, controlling for sexual orientation, and controlling for sexual identification). Ethnic differences are significant and substantial in Model 1 and Model 2, controlling for baseline demographics and religious identification (Model a) and in models controlling for both sexual orientation (Model b) and sexual identification (Model c). However controls for church attendance and Bible beliefs in Model 3 eliminate the significance of the difference between whites and African Americans and Latinos—though Asians remain consistently less supportive of same-sex marriage than whites net of other factors across all sets of models. Adding controls for Republican identification and political conservatism renews the significance of the difference between whites and African Americans and Latinos. Taken together, this suggests that the intersection between ethnicity, religion, and political identities produces varied effects on support for marriage rights. Notably, because of the very small number of Asians who identify as LGB, Model 4c does not estimate a separate variance parameter for Asians, thus reducing the degrees of freedom to 24 and likely deflating the Asian coefficient slightly compared with Models 4a and 4b.
Heterogeneous Ordinal Logistic Regression Model of Support for Same-sex Marriage: 2006–2014.
Baseline.
Controls for sexual orientation.
Controls for sexual identification.
p < .05. **p < .01. ***p < .001.
Both sexual orientation and sexual identification have a substantial impact on the odds of being more favorable toward marriage equality net of all other factors. LGB orientation increases the odds of a more favorable view by 268 percent, while identifying as gay, lesbian, or bisexual increases the odds of a more supportive response by 258 percent net of all other factors. This finding suggests that sexual orientation is no less important than sexual identity for structuring support for same-sex marriage.
Sectarian Protestant identification has a significant negative impact on support for same-sex marriage compared with the largely mainline Protestant excluded category in all of the models. Net of all other factors, sectarian Protestants have 37 percent lower odds of being in a more supportive category when compared with the mostly mainline comparison group controlling for sexual orientation (Model 4b), and the odds are 38 percent lower controlling for sexual orientation (Model 4c). Respondents with no religious identification are significantly and substantially more favorable toward marriage equality compared with the mainline in Model 2, though Model 3 shows that much of the difference between mainliners and nonidentifiers is explained by religious participation and beliefs about the Bible—nonidentifiers have low rates of church attendance, eschew fundamentalist beliefs about the Bible, and tend to view the Bible as a book of fables. Controls for Republican and conservative identifications eliminate all differences between nonidentifiers and the mostly mainline Protestant comparison category.
Catholic identifiers are significantly more supportive of same-sex marriage compared with the mostly mainline Protestant comparison group until controls for political identification are introduced in Model 4. Catholics are significantly more in favor of same-sex marriage in the baseline (Model a) and in the model controlling for sexual identification (Model c), but not in models controlling for sexual orientation (Model b). Religious attendance has a significant negative impact on support for marriage equality even after controlling for religious and political identifications. An increase in religious participation is expected to decrease the odds of being more favorable toward marriage equality by 12 percent in Model 4b and by 11 percent in Model 4c.
Biblical literalism significantly reduces support for same-sex marriage in all of the models when compared with those who think the Bible was inspired by a God or something else, while believing the Bible is a book of fables increases the odds of being in a more supportive category relative to the comparison group. In both Model 4b (controlling for sexual orientation) and Model 4c (controlling for sexual identity), holding an literalist position on the Bible reduces the odds of being in a more egalitarian category by 60 percent compared with believing the Bible was inspired by God, net of all other factors. In Model 4b, holding a secular view of the Bible is estimated to increase the odds of being in a more favorable category by 65 percent, and the corresponding estimate for Model 4c is a 76 percent increase.
Both political conservatism and Republican identification reduce support for same-sex marriage net of other factors. An increase in conservatism is estimated to reduce the odds of being in a more supportive category by 35 percent in Model 4b and by 36 percent in Model 4c. And, after all other controls, Republican identifiers have 41 percent lower odds of being in a more positive category in Model 4b, and the estimate is 38 percent in Model 4c. Notably, these findings show consistent, substantial, and significant independent effects of both political and religious identifications for structuring support for same-sex marriage.
Across all of the models, women are substantially and significantly more supportive of marriage equality than are men. Compared with men, the odds of being more favorable toward same-sex marriage are 94 percent higher for women net of all other factors in Models 4b and 4c. Support is a negative function of age, with each year expected to reduce the odds of being in a more supportive category by 2 percent in all of the models. Respondents who are divorced or who have never married are significantly more supportive of same-sex marriage compared with married and widowed respondents. Having children in the household significantly reduces support for same-sex marriage net of other controls, though the significance of the effect of children is not found in Model 4b controlling for sexual orientation. Both education and income are significant positive predictors of support for same-sex marriage in the final models, though the effect of income is largely explained by its negative association with literalist interpretations of the Bible. Income becomes significant again with controls for Republican and conservative identification. Residents of rural areas and those who reside in the South are also significantly less favorable toward marriage rights, even after controls. Finally, the trends in support for same-sex marriage is evident across all models, with each year increasing the odds of a more supportive position by 15 percent in Models 4a and 4b, and 17 percent in Model 4c.
Table 3 examines whether the predictors of support shift over time, comparing the effects of covariates in two time periods (2008–2010 and 2012–2014). Model 1 is the baseline model, Model 2 controls for sexual orientation, while Model 3 controls for sexual identity. For brevity, I focus on whether effects of key identifications shifted between the two periods. First, while there appears to be some closure of the gap in support between African Americans and whites, the difference is modest in allthree models and does not approach statistical significance. The Latino-white difference is unchanged across the two periods, while the gap between Asians and whites appears to have grown, though the difference in coefficients is not significant. In addition, the trend effect of survey year is not significantly greater in the later period, and the estimate is actually smaller in 2012–2014 in Model 2. Taken together, this suggests that there was no significant shift in support for marriage equality among African Americans or other ethnic groups following President Obama’s announcement of his support on May 9, 2012, after 53 percent of the 2012 GSS interviews were completed.
Heterogeneous Ordinal Logistic Regression Model of Support for Same-sex Marriage: Comparing Effects in 2008 and 2010 with 2012–2014.
Difference from 2008 to 2010 effect significant at the .05 level (one-tailed).
Difference from 2008 to 2010 significant at .01 level (two-tailed).
Difference from 2008 to 2010 significant at .001 level (two-tailed).
Difference from 2008 to 2010 effect significant at .01 level (one-tailed).
p < .1 (two–tailed). *p < .05 (two-tailed). **p < .01 (two-tailed). ***p < .001 (two-tailed).
The effects of educational attainment and church attendance on support for marriage equality were significantly different in 2008–2010 compared with their effects in 2012–2014. In Model 4c, in 2008–2010, each degree level was expected to increase the odds of support for marriage equality by 11 percent, while in 2012–2014, this increased to 27 percent. Also in Model 4c, an increase in church attendance was expected to reduce the odds of being more favorable toward same-sex marriage by 8 percent in 2008–2010, and this reduction increased to 13 percent in 2012–2014. The effect of children also appears to have grown. Although in 2008–2010 children had no significant effect on support for marriage equality, in 2012–2014 estimates in all of the models are significant and negative, reducing the odds of support by 9 percent in Models 1 and 3, and by 7 percent in Model 2—where the effect difference is found to be significantly different from 2008 to 2010.
In Model 1 (with no control for sexual orientation or identity), the effect of conservatism is significantly lower in the 2012–2014 period than in 2008–2010, suggesting that political identifications may be becoming less consequential for influencing beliefs about marriage equality. However, Models 2 and 3 do not show a shift in the effect of conservative identification. Notably, the influence of Republican identification is unchanged. Similarly, the influence of religious identifications and beliefs remain the same. Sectarians and biblical fundamentalists continued to be opposed to marriage equality relative to mainline Protestants. Model 1 shows that in 2012–2014, Catholics are significantly more supportive of marriage equality than the mostly mainline Protestant comparison category, while in 2008–2010, the effect of Catholic identification is not significant. Catholic distinctiveness is not found in Model 2 or Model 3 controlling for sexual orientation and identification, but in both models, the coefficients are more positive in 2012–2014.
The effect of sexual orientation and sexual identification on support for marriage equality declined considerably across the two time periods; however, the difference in the effects is not significant at conventional levels. Notably, this decrease can be expected given the very high levels of support among LGB-oriented or identified people in both periods. Much of the change in opinion is coming from heterosexuals, and this reduces their difference from LGB persons.
Table 4 examines estimates from heterogeneous ordinal logistic regression models by ethnicity. The baseline model compares effects across all four ethnic groups without a control for sexual orientation or identity. The second model compares effects for whites, African Americans, and Latinos adding a control for sexual orientation. The third model compares the three larger ethnic groups and controls for sexual identification. Because of the very small numbers of Asian LGB-oriented or identified persons, Asians cannot be compared in Model 2 or Model 3.
Heterogeneous Ordinal Logistic Regression Model of Support for Same-sex Marriage by Ethnicity: 2006–2012.
Difference from white significant at .05 level (two-tailed).
Difference from white significant at the .05 level (one-tailed).
Difference from white significant at .01 level (two-tailed).
Significantly different from Latino at .01 level (two-tailed).
Difference from African American significant at .05 level (two-tailed).
Difference from African American significant at .05 level (one-tailed).
Difference from white significant at .001 level (two-tailed).
Difference from African American significant at .01 level (two-tailed).
Significantly different from Latino at .05 level (one-tailed).
p < .1. *p < .05. **p < .01. ***p < .001.
The importance of the intersection between ethnicity and gender for structuring support for same-sex marriage is clear, being female increases the odds of a more supportive position among whites by 96 percent, while in the other three ethnic groups, gender differences are significantly less robust, increasing the odds by about 41 percent in the three other groups in Model 1. Gender and ethnic identities are interconnected in ways that inform how people view the institution of marriage, and gender and ethnic privilege among white men may be a strong conservatizing force in how they view same-sex marriage (Wadsworth 2011). The difference in the gender effect remains significant in Model 2 controlling for sexual orientation, though the gender difference between whites and Latinos is not significant after controls for sexual identity in Model 3. Sexual orientation and identification have strong positive effects on support for marriage equality in all of the ethnic groups. Sexual orientation (Model 2) and identity (Model 3) seem to have larger effects among African Americans and Latinos, although the difference from whites is not significant in either model.
Sectarian Protestant identifications have a significant negative effect on the odds of supporting marriage equality compared with the mainline among whites in all of the models; however, African American sectarian Protestants are not significantly different from other African Americans. The confluence of white ethnic privilege and conservative religious commitments appears more substantial for informing conservative opinions about the institution of marriage. Across every model, the effect of sectarian identity is significantly less negative for African Americans than for whites. Indeed, Model 1 shows that the negative impact of sectarian identity is significantly stronger among Latinos than among African Americans, though controls for sexual orientation and identification eliminates the significance of the difference.
Ethnicity also intersects with Catholic identifications, with both African American and Latino Catholics showing stronger support for marriage equality than is found among whites. The difference is particularly robust in Model 3 controlling for sexual identification, where identifying as Catholic increases the odds of a more supportive position by 66 percent among African Americans and Latinos, while white Catholics are no different from the mostly mainline Protestant comparison category.
Ethnic differences in the effects of biblical fundamentalism are considerable. Among whites and Asians, biblical inerrancy has a profound negative impact on the odds of holding a more favorable position—reducing the odds by 65 percent among whites and 77 percent among Asians, compared with 49 percent among African Americans and 42 percent for Latinos. Biblical inerrancy has a significant and substantial negative impact on support for marriage equality among African Americans and Latinos, but the effects are significantly smaller than are found among whites and Asians in Model 1. The negative impact of identifying as a biblical inerrantist is also significantly stronger among whites than among African Americans and Latinos in Models 2 and 3, controlling for sexual orientation and identification. Ethnicity structures the meaning and consequences of religious commitments, and how one interprets the Bible as the literal word of God varies across interpretive communities bounded by ethnicity.
Political conservatism also has varying effects on support for marriage equality by ethnicity, and again the more economically privileged whites and Asians show stronger effects. Higher positions on conservative self-identification are expected to decrease the odds of support for marriage equality by 38 percent among whites and 35 percent among Asians, while the corresponding decrements are 15 percent for African Americans and 21 percent for Latinos. Model 1 shows that the effect of conservatism—which could also be interpreted as liberalism—is significantly more negative among whites and Asians than it is for African Americans and Latinos. Ethnic differences in the effect of political identity on a liberal-conservative spectrum persist after controls for sexual orientation and identification.
Republican Party identification has a profoundly negative impact among whites, reducing the odds of being supportive by 41 percent in Model 1; yet among the other three ethnic groups, there is no significant difference between Republicans and others, and coefficients for Latinos are positive but not significant. Identifying as Republican is significantly less influential on support or opposition to same-sex marriage for African Americans, Latinos, and Asians.
Model 1 shows that education has a significantly less positive effect on support for same-sex marriage among Asians when compared with whites or Latinos, and the estimated effect of educational attainment is negative but not significant for Asians. For whites and Latinos, education has a strong positive effect on support for marriage equality, while the effect is positive but smaller and insignificant for African Americans in Models 1 and 2. The positive effect of education is significantly stronger among Latinos than among whites in Models 2 and 3.
Interestingly, the negative effect of Southern origins on support for marriage equality is only found among whites—African American and Latino Southerners are significantly more supportive of same-sex marriage and do not differ significantly from coethnics who are non-Southerners. The distinctive conservatism of Southern whites is also evident in Models 2 and 3 controlling for sexual orientation and identification. White ethnic privilege appears to be critical for Southerners aversion to marriage equality.
Table 5 explores how gender intersects with other identifications to structure support for marriage equality. For brevity, I mostly focus on identifications and influences that differ across genders. Although the findings are not significantly discernable, it is notable that the effect of sexual identification appears to be less strongly positive among men than among women. While identifying as gay or lesbian increases the odds of being more supportive by 335 percent for women, among men the corresponding estimate is 179 percent. The gender difference in the effects, however, is not significant at conventional levels.
Heterogeneous Ordinal Logistic Regression Model of Support for Same-sex Marriage by Gender.
Difference from female effect significant at the .05 level (one-tailed).
Difference from female effect significant at .01 level (one-tailed).
Difference from female significant at .01 level (two-tailed).
p < .1 (one-tailed). *p < .05 (two-tailed). **p < .01 (two-tailed). ***p < .001 (two-tailed).
Ethnicity and gender also intersect to structure support for marriage equality, and ethnic differences in support are stronger among women than among men, particularly the difference between white women and African American women and Asian women is larger than is found among men. Male privilege appears to generate uniformity across ethnic groups on marriage equality. In Model 3, the odds of greater support are 43 percent lower for African American women and 49 percent lower for Asian women when compared with white women, but among men, the odds are 22 percent lower for African Americans and 29 percent for Asians.
Divorce appears to have a stronger positive influence on support for marriage equality among women than among men, and the difference is significant in Models 1 and 2. Unmarried women are also found to be significantly more supportive of same-sex marriage in comparison with unmarried men in Model 2, though the difference is not significant in the other models. Children have a significantly less negative effect for men than for women in Model 1, while the difference in the effect of children is not significant in the other models. On the whole, this suggests that marriage and family factors matter more for women’s opinion about marriage equality than they do for men.
Church attendance has a significantly less negative impact on support for marriage equality among men than among women in Model 1 and in Model 3. This suggest that perhaps much of men’s religious participation is uncoupled with religious commitment and is instead more a function of social influences on participation (Sherkat 1998; Sherkat and Wilson 1995). In contrast, secular beliefs about the Bible have a significantly more positive effect on support for same-sex marriage among men than among women. In Model 3, secular beliefs increase the odds of greater support for same-sex marriage by 49 percent for women, while the corresponding increase among men is 114 percent.
Discussion and Conclusion
Religious, political, and ethnic identities have consequences for social values, and these identities often intersect to produce distinctive public policy preferences. Disentangling how ethnic identities intersect with other identities to influence opinions, and potentially mobilization, is difficult (Wadsworth 2011), and the intersectional approach used in this article is limited by its quantitative approach. Still, this research shows that ethnicity plays a strong role in structuring how religious and political identities inform attitudes toward same-sex marriage, and also how gender and sexuality intersect with ethnicity and family factors to influence support for marriage equality. The Supreme Court decision in Obergefell v. Hodges altered the playing field for social movements affirming or denying marriage rights, yet across the nation, the backlash against Obergefell has resulted in over 200 pieces of discriminatory legislation against LGBT persons (http://www.hrc.org/blog/hrc-previews-anti-lgbt-state-local-legislation). The shifting ethnic structure of the U.S. population will continue to influence how political and religious factors affect the continuing LGBT civil rights struggle.
Although Anglo-whites are more favorable toward legalizing same-sex marriage than are people from other ethnic groups after controls for baseline demographics, bivariate support is just as high among Latinos and Asians. Given the demographic growth of Latinos and Asians, this bodes well for social movements seeking to secure civil rights for sexual minorities, yet the difference between baseline and multivariate ethnic evidences that ethnicity intersects with other identities and demographic characteristics in structuring support. Particularly, white ethnic privilege seems to enliven religious and political motivations for denying marriage rights.
The African American cultural experience is steeped in the institutions of the black church, and, consequently, support for marriage equality is lower than in other ethnic groups. Yet, the negative effects of sectarian identities and fundamentalist beliefs have a less negative impact on support for marriage equality African Americans when compared with other ethnic groups. Lacking ethnic privilege, African Americans are perhaps less enthusiastic about applying religious commitments to deny marriage rights to same-sex couples. What it means to be a sectarian Protestant or even to believe that the Bible is the literal word of God is different for African Americans. Indeed, in the wake of the passage of several anti-LGBT laws around the country, African American ministers from otherwise conservative denominations have been key figures vocally opposing discrimination, even as other African American ministers support discrimination. In contrast, white ministers in sectarian denominations have remained silent or enthusiastically supported legislated discrimination.
Asian ethnic identities also appear to intersect with their advantaged economic status. The influence of educational attainment on support for same-sex marriage was substantially less positive for Asians. In contrast, despite their considerable educational deficits, education has an even stronger influence on support for same-sex marriage among Latinos than it does among whites. For Asians, it appears that definitions of appropriate forms of marriage are more connected to ethnic culture than to economic advantage, while among Latinos, the attainment of privilege makes conceptions of family more inclusive. Future qualitative work could productively explore the intersections between social class, ethnicity, and sexuality in greater detail (Cohen 1999).
Political identifications also have varied effects on support for same-sex marriage across ethnic groups. The social conservatism evident among those who identify as Republican or as political conservatives is only salient among privileged whites. For Latinos and African Americans, what it means to be conservative—or liberal—is less salient for informing beliefs about marriage rights. And, while the Republican Party has often viewed opposition to same-sex marriage as something that might attract minority voters, this research has shown that nonwhite Republicans are no different from non-Republican coethnics in their views on same-sex marriage. Social issues do not appear to be what draws ethnic minorities into the Republican Party. Future investigations should examine how ethnicity intersects with political identifications to structure other social and political orientations (Hancock 2007; Hankivsky and Cormier 2011).
Gender differences in support for marriage equality are much more pronounced among whites than for African Americans or Asians, though women are more supportive of same-sex marriage in all ethnic groups. Gender identity also intersects with family factors like marriage and child rearing, and women were found to be more influenced by marital and child-rearing experiences than men. Sexual orientation and identity predictably influence support for marriage equality, with LGB-oriented or identified persons being much more supportive. By and large, however, sexuality does not appear to intersect strongly with ethnicity or gender for informing support for marriage equality. Still, much more research is needed on how sexual identities influence other identities and structure social and political commitments.
In the wake of the Obergefell decision, “Religious Freedom” has come to be a newly amplified value in movements from the Christian right, and it appears that the amplification of the belief that same-sex marriage violates the religious freedom of conservative Catholics and sectarian and fundamentalist Christians is highly resonant (Snow et al. 1986). Although conservative Christians and Republicans may have lost the battle with Obergefell, they will almost certainly continue their war against marriage equality. The newly hatched focus on “religious liberty” as an impediment to the implementation of laws demanding equal treatment deserves immediate social scientific attention. Future studies must begin to tap what religious conservatives mean by religious liberty, and which political policies they believe violate their freedoms. Future research will also need to take into account how and why white ethnic and male privilege seem to energize religious commitments opposing equal rights.
Footnotes
Acknowledgements
Comments from anonymous reviewers, Jim Green, and the editors were helpful for improving this work.
Author’s Note
A version of this article was presented at the 2015 annual meetings of the Southern Sociological Society, New Orleans, Los Angeles.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
