Our Kind of American: Christian Nationalism, Race, and Contingent Views of Cultural Membership

ABSTRACT

American “Christian nationalism” is strongly associated with ethno-racial prejudice and xenophobia, particularly among White Americans. Yet research to date ignores the possibility of Christian nationalism operating similarly for ethno-racial minorities depending on the latter’s identification with the in-group vis-à-vis another out-group. Integrating insights from Christian nationalism research with social identity and group threat theories, we theorize racial identity differentially shapes the association between Christian nationalism and attitudes about ethno-racial exclusion or assimilation, contingent on which group is situated as the out-group. Data from the 2014 General Social Survey show Christian nationalism predicts White adults hold more exclusionary and assimilationist views. However, among Black adults (not Hispanic adults), Christian nationalism is associated with a rejection of domestic racial/ethnic minorities assimilating. And among Hispanic adults (not Black adults), Christian nationalism is associated with a rejection of exclusionary or assimilationist views when immigrants are under consideration. In contrast, when immigrants are the target, Christian nationalism operates identically for Black and White Americans. The same is true for Hispanic and White Americans when domestic minorities are the target. For ethno-racial minorities, Christian nationalism is thus linked with both inclusive/pluralistic views and exclusionary/assimilationist views, contingent on whether it connects them to the in-group or out-group.

KEYWORDS:

• Christian nationalism
• race
• assimilation
• exclusion
• identity

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Notes

1. The ISSP module on national identity was also administered in 1996 and 2004, but several of our questions of interest where not included in these additional survey waves. The ISSP module for national identity was included on form X of Ballot A, B and C while the ISSP module for Citizenship II was administered on form Y across the same ballots. See: https://gss.norc.org/Documents/quex/2014_Quex_Map.pdf

2. The choice to treat these three outcome measures as binary is driven by our theoretical and empirical concerns. Theoretically, we are simply more interested in the broad agreement/disagreement with these views rather than focusing on the level of agreement. This is also necessarily empirically, because several of the extreme options have problematically small cell sizes. For example, for the outcome statement “Those who Don’t Share Customs/Traditions can’t be fully American” only 49 respondents report strongly agreeing. For the other two multi-category outcome, we find only 30 respondents strongly agreeing and 21 respondents strongly disagree respectively. However, this does not make full use of the information provided in these data. We run a series of supplemental models on the full outcome variable treating it both as ordinal (See Table A1) and multinomial (See Tables A2–A4, Figure A1) outcomes.

3. We treat those who provided the neutral response – i.e. “Neither Agree nor Disagree”- as part of the comparison group. We believe that this is the most conservative approach to addressing our research concerns about assimilation/exclusionary attitudes. We recognize, however, that some respondents may be providing this neutral response to avoid the more “exclusionary” response due to social desirability bias. To test this, we ran a series of multinomial logistic regression models on a three-outcome coding of the dependent variable focusing on those who agreed (strongly or otherwise), were neutral, or disagreed (strongly or otherwise) to these three statements. For outcome 2, we find little evidence of any of this sort of social desirability. For outcome 3 and 4, there may be some indication of this for Hispanic respondents, but not for White non-Hispanic or Black non-Hispanic respondents. Since changing our coding of the binary outcomes would increase our effect size across the board here, we chose to retain the more conservative coding because we cannot be sure whether the response here signal real neutrality or social disability among these respondents.

4. Here, we follow Perry and Schleifer (2023) by treating this measure as a continuous outcome. In additional models (See Table A5), we treat this as a set of categorical indicators compared to those who believe it is “Not at All Important” and a comparable pattern of results.

5. Equivalized income is produced by taking the reported family income divided by the square root of the number of individuals in the home. Income is adjusted to 2014 dollars and we take the natural logarithm to account for skew in this measure (see Perry and Schleifer 2023).

6. The GSS party ID measure includes the categories: “Strong Democrat,” “Not Strong Democrat,” “Independent, near Democrat,” “Independent,” “Independent, near Republican,” “Not Strong Republican,” “Strong Republican,” and “other Party” For our indicators, we combine the first three categories for Democrats and categories 5 through 7 for Republicans. For the indicator comparison groups, we include the category for political independents and those reporting “other Party.”

7. Along with issues surrounding the interpretation of interactive coefficients for categorical models, Gelman, Hill, and Vehtari (2020), have raised concerns about sample size, statistical power, and interactional strategies. Given the relatively small GSS sample used here, we wanted to test what the necessary sample size was to detect interactional effects for our models presented here. We ran a series of model simulations taking our interactive models as the baseline and compared model likelihood ratio to the same model without the interaction while varying the sample size to determine at what sample we achieved. 80 statistical power to capture interaction effects here at the alpha = 0.05 level. We focused on outcome 1 here, as this outcome has the smallest sample size across our outcomes. These analyses show that under these conditions, we can detect with .8 power and 95% confidence interval the interactional effect of the size uncovered in our real analyses with a sample size of around 800. While we recognize that this test may be fairly novel, we are unaware of a more established test that meets our criteria here (i.e a power test that focuses on interaction, accounts for non-continuous variables, and allows for multiple interaction effect). For our purposes, this test bolsters our confidence in the detection interactional effects in our statistical analyses under these conditions.

8. We recognized that some argue that weighting regression models is inappropriate (see: Winship and Radbill 1994). In Table A6, we present our results using three different weighting strategies to show our results are robust to our weighting decisions. All models presented below are model weighted regressions.

Additional information

Notes on contributors

Samuel L. Perry

Samuel L. Perry is Professor of Sociology at the University of Oklahoma. His work focuses on the interplay of religion and power. His recent articles have appeared in Social Forces, Sociological Theory, and the Annual Review of Sociology. He is also the author of five books, including the award-winning Taking America Back for God (with Andrew Whitehead) and The Flag and the Cross (with Philip Gorski).

Cyrus Schleifer

Cyrus Schleifer is an Associate Professor of Sociology at the University of Oklahoma. His research examines different statistical approaches to modeling religious change and within-occupational inequality. His research has appeared in the American Sociological Review, Journal of Sociological Methods & Research, Social Forces, Sociological Science, Social Science Research, and other outlets.

Andrew L. Whitehead

Andrew L. Whitehead is an Associate Professor of Sociology at the Center for the Study of Religion and American Culture at IUPUI and author of two books on Christian nationalism.

Kenneth E. Frantz

Kenneth E. Frantz is a doctoral student in the Sociology Department at the University of Oklahoma. His research interests include religion, politics, and disability.