The consequences of religious strictness for political participation

For helpful assistance, we would like to thank Franklyn C. Niles, Michael L. Coulter, Paul A. Djupe, Bill Jacoby, Dominik Hangartner, Michael Bernhard, Ken Wald, Jamie Carson, Laurence R. Iannaccone, Gregory Hawrelak, several anonymous reviewers, and anonymous judges from the Institute for Urban Research at Rice University. For providing data, we would like to thank Laurence R. Iannaccone, Adele James, Michael Emerson, David Sikkink, and the investigators from the Portraits of American Life Study. Complete replication information is available at our Dataverse page (http://hdl.handle.net/1902.1/21720).

Abstract

Contemporary media portray highly religious Americans as active political conservatives. This article examines how church strictness influences political participation by churchgoers. We argue that church strictness influences aspects of a person's life that are known to influence political participation, so assessing the effect of religion on participation requires considering intermediate factors. To evaluate our theory, we analyze the 2006 Portraits of American Life Study, which focused on the role of religion in society. We develop a recursive model of political participation, using multiple imputation to address missingness in the survey. The results indicate that indirect effects of strictness—through civic involvement, income, and religious participation—mitigate the assumed direct effect of strictness upon political participation. We conclude that, although religious groups show political activism in some specific arenas, strict churches are not strong political mobilizers in general, as many media portrayals may lead one to believe.

Keywords:

• Political participation
• Religious strictness
• Multiple imputation

Notes

For helpful assistance, we would like to thank Franklyn C. Niles, Michael L. Coulter, Paul A. Djupe, Bill Jacoby, Dominik Hangartner, Michael Bernhard, Ken Wald, Jamie Carson, Laurence R. Iannaccone, Gregory Hawrelak, several anonymous reviewers, and anonymous judges from the Institute for Urban Research at Rice University. For providing data, we would like to thank Laurence R. Iannaccone, Adele James, Michael Emerson, David Sikkink, and the investigators from the Portraits of American Life Study. Complete replication information is available at our Dataverse page (http://hdl.handle.net/1902.1/21720).

1 Tel.: [phone].

2 As a caveat to this pattern, in the 2004 election Ohio counties with more evangelicals supported George W. Bush and Issue 1 (to ban gay marriage), but did not show higher levels of turnout (Smith, 2008, p. 32).

3 R² between the strictness factor we use and its six items is 0.81, above a 0.5 threshold (Grice, 2001). Besides these factor scores, we also considered three alternative measures of strictness, which yielded similar inferences. The first measure uses expert evaluations gathered by Iannaccone (1994) to create an exogenous strictness measure. The second is a seven-point religious liberalism self-report. The third is an indicator for whether the respondent is a member of an evangelical church, since some conservative denominations have unique means of mobilizing their parishoners (Beyerlein & Chaves, 2003). We also a fit a model among Protestant respondents with a categorical measure of religious tradition. This model shows that liberal Protestants, ceteris paribus, are the most participatory and significantly more participatory than evangelicals. This supports past work showing that conservative denominations face theological hurdles to mobilization (Jelen, 1991; Wilcox, 1992). These results are in online Appendix A.

4 R² between our civic engagement factor and its three component items is 0.74, and R² between the church attendance factor and its three components is 0.65. Both exceed the 0.5 threshold (Grice, 2001). The three measures for civic engagement are each recorded on a 0–7 scale and ask respondents the following: (1) “About how many times in the past twelve months have you attended any public meeting in which there was discussion of local government issues, or school or other community affairs?” (2) “About how many times in the past 12 months have you attended a small group that meets regularly and provides support or caring for those who participate in it?” (3) “About how many times in the past 12 months have you volunteered for a project that helps people in the local community?”

5 Tests indicate that ignoring the dispersion parameter (which allows the variance to differ from the mean) or the zero participation logit would result in loss of fit. Hence, a simpler count model would not suffice.

6 The direct effect of strictness is insignificant in six of the nine logit models, but is significantly negative in models of voting, signing a petition, and attending a political meeting or rally. Church attendance remained insignificant in eight of nine models, only having a significant negative effect in the model of giving money to a party, group, or candidate. Income retained its significant positive value in six of nine models, but this effect disappeared in the models of attending a meeting or rally, working for a religious organization involved in politics, or participating in a demonstration or protest. Lastly, in all nine logit models, civic engagement continued to be an important positive predictor of political activity.

7 These numbers are based on a logistic regression model fitted over the Portraits of American Life Data. Our cost index, which ranges from −1.068 to 3.311, was the one predictor, and the response was an indicator coded 1 if the respondent voted for Bush, 0 for Kerry. Similar results hold with cross-tabulations using three other measures of strictness to predict vote choice. For example, non-evangelical respondents voted for Bush at a rate of 50%, but evangelical respondents voted for Bush at a significantly higher rate of 69%.