Understanding the Complex Influence of Religiosity on the Race Gap in Support for Proposition 8

We provide evidence from several surveys showing that religiosity can not only explain (mediate) but also influence (moderate) the relationship between a voter’s race and/or ethnicity and her attitudes about Proposition 8 (California’s Proposal to amend the state’s constitution and ban same-sex marriage in 2008) and related issues. These findings advance our understanding of the complex and enduring role of the Black church as a socializing institution in African-American communities, and they speak to the need among proponents of same-sex unions to inform churchgoers about the socio-political implications of anti-gay and exclusionary policies.


Introduction
In Chicago's famous Grant Park, and in front of thousands of tearfully-jubilant supporters, then-senator Barack Obama became the United States President-elect.Of course, scholars and journalists will debate the racial implications of Obama's accomplishment (compare, e.g., Mansfield, 2008to Morris & McGann, 2008), but it is difficult to overstate the historical significance of having the first President of the United States (POTUS) of color.Many Americans, regardless of race or ethnicity, did not believe that such an accomplishment would happen in either their own or their grandchildren's lifetimes, and by shattering what was hitherto considered one of the highest political glass ceilings, Obama forces the nation to engage in vital (but often uncomfortable) dialogues about race relations (Asim, 2009;King & Smith, 2011;Walters, 2007).
While election night was considered a victory for many African Americans, another historically marginalized group suffered a devastating loss: Gay, lesbian, bisexual, and transgender (GLBT) communities in California a state known for its progressive stance on same-sex-unions (Stryker & Van Bushkirk, 1996) were having their right to marry reversed.It hardly seemed fair, especially since, compared to Whites, a disproportionate number of Blacks voted to amend the state's constitution and overturn the legitimacy of same-sex marriage (Besen, 2010;Egan & Sherrill, 2009;Murray, 2009), and some believe that African-American turnout was critical to the passage of this amendment.The most conspiratorial of these claims imply that pro-Obama enthusiasm among Black voters was exploited to mobilize them against same-sex unions (Abrajano, 2010;CNN Wire Staff, 2012;Miller, 2009;Sherkat, Vries, & Creek, 2010).Such claims begged the question: "how could African-Americans, a group far too familiar with the abuses of social exclusion, vote to deny other minorities the right to equal citizenship?"Considering what Blacks endured, one would expect them to be more sensitive to the plight of gays and lesbians.Cartoonist Darin Bell lampoons these narratives of Black hypocrisy in the November 17 through 19 2008 issues of his comic strip, "Candorville", in which he uses a conversation between two of his characters to exaggerate the dilemma of "racial pride and anti-gay prejudice" (http://candorville.com/archives/).
The issues to which Bell refers presuppose that homophobia played a key role in African Americans' votes for the California Marriage Protection Act (Proposition 8).We offer a different explanation, for we believe that it is difficult to discuss opinions about same-sex marriage without considering the importance of voters' religious values.We begin by referencing the literature on race and sexuality to develop the argument that religious upbringing mediates (explains) as well as moderates (influences) the race gap in Proposition 8 support.We examine these "mediator" and "moderator" effects using a combination of national-and state-level public opinion surveys.Consistent with the mediator argument, controlling for church attendance cancels out the impact of race on the likelihood that a Californian votes for the proposal.Moreover, the effect of race is conditional upon religiosity: racial differences in support for Proposition 8 tends to exist at lower levels of religiosity; when we consider voters who attend church frequently, both racial groups express similar degrees of disapproval for same-sex marriage.We conclude by discussing how these findings advance our understanding of the complex role of the Black church on African Americans' attitudes about sexuality.

Race, Religiosity, and Proposition 8
Research on race and attitudes about sexuality yields mixed results.Some find that, compared to their White colleagues, African Americans are noticeably less accepting of homosexuality as a lifestyle (Glick & Golden, 2010).Others demonstrate the opposite: Whites, not Blacks express viewpoints that are more homophobic (e.g., Lewis, 2003).There are even scholars who fail to observe racial differences in attitudes toward GLBT persons (Herek & Capitanio, 1995).In short, race either polarizes viewpoints about sexuality or has no impact whatsoever.These conflicting finding prompted Shaw and McDaniel (2007) to note that the literature lacks an adequate interpretation of the relationship between race and GLBT attitudes.
Research on attitudes about Proposition 8 is similarly inconsistent.For instance, Egan and Sherrill (2010, Table 1) use precinct-level and survey data to demonstrate that exit polls may overstate the level of Blacks' support for Proposition 8 and that alternative data sources show racial gaps that are more narrow.These results, while divergent, are not incompatible.Conventional explanations of the race gap tell different stories under different circumstances; the challenge, therefore, is determining which circumstances affect the stories these explanations tell.As Baron andKenny (1986, p. 1173) note, mixed conclusions often suggest the existence of an intervening factor, an additional concept that can reconcile the uneven findings in a literature.Recent research suggests that religiosity (commonly defined as the frequency of church attendance) is such a factor.As an intervening concept, religiosity can function as a mediator (in the sense that the alleged race gap in Proposition 8 support disappears after accounting for it) or a moderator (that is, the impact of race on support for same-sex marriage differs according to levels of religiosity).To our knowledge, scholarship has focused on either the mediating (e.g., Egan & Sherrill, 2009;Ellison, Acevedo, & Ramos-Wada, 2011) or moderating (Taylor, Mattias & Chatters, 1999) properties of religiosity.Such "either/or" characterizations paint an incomplete picture of the complex and poorly understood relationships between race, religion, and Proposition 8. Accordingly, we distinguish ourselves from past research by considering the dual functions of religiosity simultaneously, for examining its mediating and moderating effects will enable us to explain why the hotly-debated relationship between race and Proposition 8 exists, and, more importantly, how church habits factor into this relationship.

Religiosity as a Mediating Concept
There is a well-established relationship between a person's race and her level of religious devotion (Calhoun-Brown, 1999;Cohen, 1999;Harris, 1999).In addition, a widely documented link exists between an African American voter's churchly dedication and her support for same-sex marriage (for recent evidence, read Griffith, & Cruz, 2010).More relevant to this paper, several scholars (Egan & Sherrill, 2010;Sherkat, Vries & Creek, 2010) and journalists (Besen, 2010;Grad, 2008;Roker, 2008;Silver, 2008;Wilson, 2009) discuss the impact of race on attitudes about Proposition 8 specifically.
Taken together, these studies suggest that religiosity mediates the relationship between race and support for Proposition 8.In a report to the National Gay and Lesbian Task Force Policy Institute, Egan and Sherrill (2009, p. 9) make the case that: "…much of African Americans' support for Proposition 8 can be explained by the fact that [B]lacks tend to be more religious than Californians as a whole" (italics added).Similar arguments appear throughout the literature on race and sexuality (e.g., Herek & Capitanio, 1995), and Schulte and Battle (2004, Table 3) are perhaps the first scholars to subject a variant of this "mediation hypothesis" to an empirical test.In light of this evidence, we too expect that Black votes for Proposition 8 are primarily an artifact of Blacks' historically strong adherence to religious principles.Specifically, we expect divergent attitudes about same-sex weddings to disappear after accounting for racial differences in levels of religiosity.
Hypothesis 1 (Mediator Effect): When controlling for the level of church attendance, a voter's race will have little, if any, impact on his or her support for Proposition 8.

Religiosity as a Moderating Concept
Above, we hypothesize that the race gap in support for Proposition 8 will disappear after accounting for religiosity.What is it about religiosity that closes gap?We believe that exploring religiosity as a "mediator" requires that we also consider its role as a "moderator".Voters tend to base their support for Proposition 8 on their religious teachings, and this pattern holds true regardless of race.However, the works of Harris-Lacewell (2004, chap.2), Jacobson (1990, p. 258) andMcDaniel (2008) remind us that the importance of religiosity as a "socializing agent" is greater for African Americans than it is for Whites.Furthermore, Shaw and McDaniel (2007) acknowledge the prevalence of research examining the implications of deeply internalized homophobic beliefs in Black communities-beliefs that contribute not only to secret encounters between Black men who "have sex with other men" but reject a gay identity (Boykin, 2006;King, 2004) but also to racial disparities in HIV infection rates (Glick & Golden, 2010).
If Blacks are more heavily influenced by church attendance than Whites are, then religious devotion among African Americans may pull their viewpoints about sexuality away from those of Whites.Put differently, the impact of religiosity on support for Proposition 8 might vary by race.Both groups are susceptible to the influence of religiosity, but the degree of susceptibility, we believe, is greater among African Americans.This argument suggests the following expectation: Hypothesis 2 (Moderator Effect): If religiosity is at its lowest level, then there will be no race gap in support for Proposition 8.However, as the level of religiosity increases, its effect on support for Proposition 8 will be greater for African American voters than it will be for White ones.

Survey Data
We use several data sources to examine mediator and moderator effects.The National Election Pool ("NEP") conducted 51 separate state-level exit polls for the 2008 Presidential Election, and there was there was also a national survey.The California exit poll of 2,360 voters included 765 respondents who were interviewed by telephone.The telephone survey adjusted for those who voted early or by absentee ballot.The NEP asked respondents "How did you vote today on Proposition 8 to amend the state constitution to eliminate the right of same-sex couples to marry"?Because the sub-sample sizes were large enough, we were able to compare four different racial groups: Whites (N=1,400), Blacks (212), Hispanics/Latinos (407), and Asians (125).
We also wondered whether the racial differences would also occur at the national level on questions concerning gay marriage.Accordingly, we looked at four identical survey questions in the April 2009 through July 2011 waves of the ABC/Washington Post poll (henceforth called "ABC/WP") and 6 identical questions asked in the April 2009 through March 2011 waves of the CBS/New York Times poll (hereafter referred to as "CBS/NYT").(Note 1) Supplementing our state-level data with national polling results allows us to situate out findings more easily within the extant literature on racial differences in political opinions, particularly those opinions pertaining to issues of racial and ethnic identity, religiosity, and sexuality.(Note 2) 2.4 Plan of Analysis 1.We begin by crosstabulating responses to Proposition 8 by race (see Table 1).We also crosstabulated for each of the four racial groupings and nine other predictor variables by Proposition 8 (see the third bullet point for details).Unfortunately, the NEP used 4 different survey forms and the question asking whether the respondent was born-again was not used in one of these and religious service attendance was not used in two of the four forms.In Table 1, we also display the relationship between race and church attendance, but, because of space limitations, the control variables are not displayed.
2. We used log-linear models to determine whether the relationship between race and Proposition 8 was statistically significant after controlling for the predictor variables (the mediation effect).Furthermore, we used log-linear modeling to determine whether the relationship between the predictor variables and Proposition 8 was significantly affected by race (the moderating effect).We use similar analyses to examine the national surveys (see Table 3).

Results and Discussion
The first column in Table 1 shows the relationship between a respondent's self-reported race/ethnicity and his or her support for Proposition 8.One clearly sees that the frequency of Blacks voting "yes" on the California constitutional amendment is higher than that of respondents from any other racial or ethnic group and that there is a 21-point gulf (70%  49%) between White and Black respondents.Likewise, the second column of the table reveals a connection between race/ethnicity and church attendance.The percentage of respondents who attend church at least once a week is higher among African Americans than it is for Whites, Hispanics, or Asians (61%, 28%, 30% and 45%; respectively).In fact, there is a 33-point gulf [61%  28%] between Blacks' and Whites' levels of religiosity.Furthermore, Table 2, shows that respondents who are present at weekly religious services are more likely than those who attend church less frequently to support Proposition 8 (a 48 point chasm [84%  36%], see Smith and Seltzer [2001] for a discussion of gaps, gulfs and chasms).The results are less consistent in the national surveys.We found a significant racial difference regarding whether it should be legal for gay and lesbian couples to marry in one of the four waves of the ABC/WP survey (results not shown).In July 2011, 55% of Whites and 42% of Blacks believed that it "should be legal for gay and lesbian couples to get married" (p=.01).For three of the six CBS/NYT surveys, there were no significant differences regarding whether gay couples should be able to legally marry, form civil unions, or have no legal recognition (see Table 3).(Note 4) However, in June 2009, African Americans were more likely than Whites to favor no legal recognition (47% versus 33%).Similarly, in April 2010, Blacks were more likely than Whites to favor no legal recognition (47% v 31%).The results were borderline significant in March 2011 (43% v 34%; p=.07).Results from the multivariate analyzes generally comport with Hypothesis 1. Specifically, in a log-linear model, race was not statistically significant in affecting attitudes toward Proposition 8 after controlling for frequency of attendance at religious services (Likelihood Ratio ["LR"]=1.0;df=2).The July 2011 ABC/WP survey did not ask about church attendance.As a proxy, we used whether the respondents said they were a "born again" or "evangelical" Christian, and race was no longer significant after controlling for this variable (LR=3.6,df=1).Similarly, the June 2009 CBS/NYT survey did not ask about church attendance.We controlled for whether the respondent said they were evangelical or born again Christian, and race was no longer significant afterwards (LR=3.5, df=2).As discussed above, the April 2010 CBS/NYT survey also found that African Americans were more likely than Whites to believe that there should be no legal recognition.This remained significant after controlling for either being evangelical (LR=15.8,df=2) or pious (LR=20.4,df=2).In summary, of the eleven surveys we examined (ten national and one in California), race did not significantly relate to attitudes about gay marriage in seven of them.In three other surveys, race was not significant after controlling for religiosity.However, in one survey, race remained significant after controlling for religiosity.Table 4 presents the results of a logistic regression of the determinants of Proposition 8 support.Here, the dependent variable is the yes/no Proposition 8 question appearing in the California exit poll from the 2008 National Election Pool (NEP) and the covariates are the predictors mentioned in the previous section.Table 4 reveals that church attendance was the strongest predictor variable (odds ratio ["OR"] =10.4), followed by whether the respondent voted for Obama (OR= 6.50), was not a Democrat (OR=3.13),did not attend college (OR=2.06),and was not a liberal (OR=2.77).Race was not significant in the logistic regression equation.(Note 5) Note: results are the percentages of African-American and White respondents (presented separately) who express opinions to the following question: "what comes closest to your view?Gay couples should be allowed to legally marry, or gay couples should be allowed to form civil unions but not legally marry, or there should be no legal recognition of a gay couple's relationship?" Having discussed its role as a mediator, we now explore religiosity as a moderating variable.Table 5 allows us to evaluate Hypothesis 2 (i.e., that the influence of religiosity on opposition to gay marriage is greater for Black respondents than it is for White ones).We find evidence of this only in the CBS/NYT April 2010 survey (the one survey in which race remained significant after controlling for religiosity).(Note 6) Among Whites, 22% of those who attend religious services weekly support gay marriage compared to 53% of those who attend religious services less often (a 31-point gulf).Among Blacks, the corresponding figures are 6% and 60%; a 54-point chasm.Essentially, the effect of religious service attendance had a greater impact among African Americans than among Whites, a finding that is consistent with our expectations (LR=12.0,df=2, P=.002).Among Whites, those who attended services weekly were highly likely to believe that there should be no legal recognition (49% v 20%; gulf=29).The corresponding difference among Blacks was noticeably greater (83% v 23%; chasm=50).
As we note in Table 6, there were three situations in which the relationship between a predictor variable and Proposition 8 was affected by race.The first was with ideology: Among all four racial groupings, liberals were less likely to support Proposition 8.For Whites (40% v 88%), Hispanics (54% v 79%), and Asians (47% v 71%)-moderates were less likely to support Proposition 8 than were conservatives.However, there was no difference between moderates and conservatives (76% v 77%) among Black respondents.The conditional effect for age is also worth noting.Among Whites, Asians, and Hispanics, younger respondents were more likely to vote "no" on Proposition 8 than were older respondents.However, among Blacks, it was respondents between the ages of 45 and 59 who were most likely to oppose the California amendment.For Blacks, respondents over the age of 60 and those under the age of 30 were particularly likely to support Proposition 8.There are similarly fascinating findings with regard to education.For Whites, greater levels of education led to greater opposition to Proposition 8.However, this increase in opposition toward same-sex unions is statistically insignificant, and there were no appreciable "education effects" among the other three racial and ethnic groups.Note: Estimates are percentages.
Perhaps the biggest result in Table 6 is the non-finding for religiosity.Contrary to Hypothesis 2, there was no significant interaction effect between race, religious attendance, and support for Proposition 8.The lack of significant racial differences among highly religious respondents suggests that religiosity, not race, was the major motivator of opposition to same-sex marriage.Rather than driving Blacks and Whites apart (which is what we initially assumed), appeals to morality and piety are universally effective in mobilizing voters to amend the state's constitution.When such appeals are made with sufficient strength, then we can expect support for Proposition 8 to increase, regardless of race.Politicians who understand U.S. race relations are particularly likely to succeed when they direct their anti-gay messages at African-American churches, but pious Blacks are no more susceptible to these messages than other God-fearing voters are.Taken together, this non-finding suggests that Black voters have received much undeserved attention for their role in the passage of Proposition 8.

Concluding Remarks
Using a combination of state-and national-level survey data, we examine the conditional impact of religiosity on the race gap in attitudes about same-sex attitudes more generally and support for Proposition 8 in particular.We find mixed results.On the one hand, there is clear evidence of religiosity's mediating effect (Hypothesis 1): any Black/White differences in the likelihood of voting "yes" on the California amendment disappear once we account for the role of church attendance.For the national data, race was insignificant in seven surveys questions on gay marriage and did not remain significant in two surveys after controlling for religiosity.However, in one national survey, race still affected attitudes toward same-sex marriage even after controlling for religiosity.For the most part, Blacks and Whites have similar attitudes toward gay marriage, or Back opposition is attributable to Blacks being more religious.Our expectation of "religiosity as a moderator" (Hypothesis 2) was not supported in examining Proposition 8.However, the moderator effect was present in one of the national surveys.When religiosity levels are low, Blacks and Whites have similar levels of Prop 8 support; at higher levels of religiosity, Whites are substantially more likely to express anti-gay marriage sentiments than Blacks are (see Table 5).As much as we would like to speculate why there are inconsistent findings for the April, 2010 survey (Congress allowed D.C. to legalize same-sex marriage in March 2010, perhaps the lead up to the 2010 mid-term elections, and so on), we don't know why the results for this survey were different).This result did not occur in the three CBS/NYT surveys before April, 2010 or in the two after this date.
Beyond the empirical contributions these findings make, the recent controversy over same-sex marriage in California provides us with a unique opportunity to understand the complex role of race and religion on support attitudes about sexuality.We opened this paper with a discussion of the alleged contradiction among African Americans who, to borrow Miller's (2009) phrasing, supported Obama but not same-sex marriage.As noted, Blacks' support for Obama and their opposition to same-sex marriage became the topic of spirited debates.The results of this paper inform these debates by demonstrating that it is just as problematic to blame Black Californians for the passage of this state constitutional amendment as it is to hold Blacks responsible for the outcome of the 2008 presidential election.In both cases, electoral success depended on the both White and non-White support, and our findings suggest that the race gap in votes for Proposition 8 tells us more about the effectiveness of issue framing than it does about group conflicts between Blacks and GLBT persons: Strongly religious citizens, regardless of race, tend to base their political attitudes on the messages they receive in church.Furthermore, the literature shows that Proposition 8 was framed primarily as a "matrimonial" rather than a "civil rights" issue, and it is clear that many churchgoers are ambivalent (if not antagonistic) toward same-sex unions.
When considered in this light, it is not surprising that any citizen who harbors fundamentalist beliefs would vote in favor of an amendment that she or he believes preserves the sanctity of marriage.
The polarizing impact of religion necessitates that proponents of same-sex marriage educate and mobilize voters.For example, if the rhetoric of the religious implications of same-sex had been minimized, or if Proposition 8 been effectively packaged by politicians and framed by the media as a "civil rights" issue, then the outcome in 2008 may have been different (Ghavami & Johnson, 2011).Current political events lend credence to this speculation, for recent polling results show that support for same-sex unions is increasing-particularly among African Americans (Ergun, 2012), and one can credit this opinion change not only to the Obama administration's "evolving" stance on the issue (Hirshman, 2012) but also to coalition building between prominent civil rights and gay rights leaders and organizations (Cohen, 2012;Taylor, 2012).

Notes
Note 1.The surveys were provided by the Roper Center for Public Opinion Research.The ten national surveys occurred over 3+ days of data collection and the sample sizes were approximately 1,000.
Note 2. We acknowledge that using multiple surveys introduces issues of measurement and commensurability; however, the fact that these data sources converge upon a similar set of findings gives us some confidence that, despite the flaws, our analyses paint a reasonable picture of the role of religion in attitudes about same-sex unions more generally and Proposition 8 specifically.
Note 3. Some of these dichotomies might seem non-intuitive (i.e., non-liberal instead of liberal).This recoding was done so that all odds ratios would be greater than 1 and hence more interpretable.
Note 4. One survey (December, 2010) showed racial differences that were "borderline": the result (p = .06)fell just shy of reaching statistical significance at the conventional level of .05.
Note 5. When we modeled the data using several different stepwise regression strategies the same results occurred.Neither races by itself, nor race in interaction with church attendance, were significant after introducing the control variables.
Note 6.Unfortunately, CBS/NYT did not often ask respondents how often they attended religious services.However, our other indicator of religiosity (being an Evangelical Christian) only exhibited a significant three-way interaction in the month discussed above (April, 2010).

Table 1 .
Support for Proposition 8, religiosity by race California exit poll data from the 2008 National Election Pool (NEP).Note 1. Exact wording for Prop 8 question: "How did you vote today on Proposition 8 to amend the state constitution to eliminate the right of same-sex couples to marry?" Results in first column represent percentage of respondents who voted "yes."Note 2. Exact wording of religiosity question: "How often do you attend religious services?"Results in the second column are the percentage of respondents who claim to attend church services "at least once a week."Note 3. p < .000for both questions

Table 2 .
Religiosity and support for Proposition 8

Table 3 .
Black/White differences in attitudes about sexuality more generally: results from CBS News/New York

Table 4 .
Logistic regression models of support for Proposition 8 California exit poll data from the 2008 National Election Pool (NEP).

Table 5 .
Interaction of race, support for gay marriage, and church attendance

Table 6 .
Significant 3-way interactions with race and Proposition 8