Not on the Same(-Sex) Page: How Religion Shapes Opposition to Same-Sex Marriage Among Sexual Minorities and Heterosexuals

Approximately 7–10% of sexual minorities report opposing same-sex marriage. This estimate is based on pooled General Social Survey (2008–2016), pooled American National Election Survey (2008, 2012), and Cooperative Congressional Election Study (2016) data. Due to data sources not consistently measuring sexual minority identities and same-sex marriage concurrently, we are unable to examine how these trends may have changed over time.

When referring to our sample, we use “LGBQ + ,” “LGB,” and “Sexual Minorities” interchangeably. However, when making more general claims or statements, we use “LGBTQIA + .” When referencing other authors’ work(s), we mirror their language.

In the Catholic Church, there remains much complexity and disconnect surrounding Catholic leaders and congregants’ support for same-sex marriage. Catholic teachings situate LGBTQIA + individuals as “objects of derision, prejudice, hatred, and discrimination” (Jelen, 2011:208). Historically, Catholic teachings have maintained that same-sex relationships (including marriage) are sinful and Catholic officials (e.g., the Pope and Diocese) have publicly condemned same-sex marriage and/or (re)affirmed marriage is reserved for the union of one man and one woman (Jelen, 2011; Love, 1998). Indeed, as an official position, the Catholic Church continues to view marriage as a monogamous, heterosexual union designed for procreation (Chappell, 2021).

However, this does not necessarily mean that anti-LGBTQIA + beliefs, views, and practices do not remain in collectivistic religious traditions. As Coley (2018:426) states, “religious traditions that teach that homosexuality is a “sin” might nevertheless choose not to discriminate against sexual minority students, that is, if these…religious traditions [sic] have well-developed bodies of social justice teachings, comfort with the idea of universal human rights, and/or concern for the way that societies fail their most vulnerable members.”

We report how the 2016 CCES data compares to that of other nationally representative surveys/estimates in Appendix Table 3.

Questions about sexual orientation and gender identity have only recently been included in large-scale social science surveys (Jones, 2021). Even in surveys that do include questions about sexual and gender identity (e.g., the General Social Survey [GSS] and the American National Election Survey [ANES]), they are limited in the number of sexual minorities and the ability to conduct analyses/capture proper variation (see Appendix Fig. 7). This is especially the case for religious variation among sexual minorities (Ratcliff and Haltom, 2021; Woodell & Schwadel, 2020). Due to small samples of LGBT individuals and fewer sexual minorities identifying with conservative religious traditions in the USA (Woodell & Schwadel, 2020), the GSS and ANES do not adequately capture the nuanced statistical relationship between sexuality and religion (also see Appendix Table 4).

We also conducted models where bisexuals were coded independently to determine whether the opinions of bisexuals were distinct from those of lesbian women and gay men. The results were substantively similar and are available upon request.

Being “born-again” and “evangelical” may have unique effects. As Smidt (2021) notes, conflating the two identities can lead to issues when examining socio-political outcomes. However, we rely on the born-again measure and the religious affiliation measure to identify evangelical Protestants, a limitation of the CCES. Despite differing approaches, they often result in similar findings/takeaways (Smith et al., 2018).

We also conducted our models using religious attendance, religious importance, and prayer frequency as individual variables. These do not substantively change results.

In ancillary analyses/robustness checks, we conducted heterosexual-only and sexual minority-only models. Since our imputations included sexuality, the samples vary among the imputed datasets for the sexual minority and heterosexual only models. The same method is applied to our pooled GSS and ANES (2008–2014) sensitivity analyses.

Tables that include the regression results are found in Appendix Tables 5 and 7. Although the focal interaction is not significant, traditional statistical significance tests are inefficient when estimating interaction effects in non-linear models, especially logistic regression (Buis, 2010; Norton et al., 2004). As Norton et al. (2004) note: “The statistical significance cannot be determined from the z-statistic reported in the regression output. The odds-ratio interpretation of logit coefficients cannot be used for interaction terms” (p. 154).

The final sample size (post-imputation) is not 64,000 because we use the LGBT-specific weights which are only available for those that responded to the post-election survey. We re-estimated our models (and imputations) using the pre-election survey weight which resulted in a final sample of 64,000; the results are substantively similar and available upon request.

Ratcliff and Haltom (2021) find that sexual minorities are more likely to be an “Other” religious group because the 2013 Survey of LGBT Adults included “Nothing in Particular” in the operationalization of “Other” religious identities. Instead, here we follow recent work on religion non-affiliates and include respondents who report “Nothing in Particular” with Agnostic and Atheist respondents (Schwadel, 2014, 2020).

The resulting sample of the 2008–2014 GSS/ANES data is 18,519 of which only 4.4% identify as lesbian, gay, or bisexual. Given these few observations and limited variability at the religion-sexuality nexus (see Table 2), we collapse variables like religion (0 = Not Evangelical Protestant, 1 = Evangelical Protestant), race (White/Racial Minority), and region (South/Not South).

We conducted supplemental analyses examining the intersection of sexuality, religious affiliation, and race. However, we do not discuss them here as they have very small underlying counts at some intersections (e.g., sexual minority Black mainline Protestants that oppose same-sex marriage). These are available upon request from the corresponding author.