Is American Religion Partisan Sorting in the Trump Era?

By Paul A. Djupe

Social scientists are increasingly concerned about social sorting – how individuals choose social environments based on political agreement. All sorts of ills accompany such a trend. People become less open to new ideas, strengthen their (political) identities, homogenize their news sources, become more prejudiced, and more. There are many reasons why James Madison, writing in Federalist 10, was concerned about homogeneous states. And the same is true for religious groups (Madison explicitly referenced homogeneous religious groups in Federalist 39). Among other things, the lack of independence from political parties limits their reach and moral authority. While some religious groups have been associated with a political party for a long time now, can we find evidence that religious people are sorting based on their politics?

I’m drawing on a variety of data sources that I’ve gathered with some excellent colleagues (see the data note for details) from just before the 2016 election and almost every year thereafter. You’ll note that not every year has an estimate for every group, which is a limitation of some of our samples. But we did ask nearly the same question: “If you had to guess, how supportive of Donald Trump are the following figures and groups?” The most inclusive version (in 2020) listed: Head Clergyperson, My congregation, My friends, My community, and My state. The scale ranged from 0 to 100 (labeled “Fully supportive of Trump”). I need to issue a caveat that there is variation in the composition of the samples and how some of the religion questions were asked, so an unknowable amount of this movement may be due to these factors. To maximize comparability, I’m going to focus on whites.

White religious groups perceive shifts in their social worlds across Trump’s presidency. It is no surprise that white evangelicals perceive the highest support for Trump across their social contexts and the layers are tightly integrated. That is, there are effectively no differences between their states, congregations, clergy, communities, and friend groups. Moreover, there is an uptick in support across these four years from the mid 50s to the low 60s. For a group that is often described as “embattled”, we should understand that to mean from people outside of their social worlds. Of course, this is just what might be contributing to their sense of threat felt from others. This pattern also makes sense given the structure of the religious group – it is highly decentralized with clergy entrepreneurs looking to attract people from a particular community. The rise of highly localized, non-denominational Christianity certainly exacerbates these trends.

Other white religious groups show some division amongst their social layers. For instance, Jews in 2016 saw a sizable difference between Trump support in their communities and states compared with support in their congregations and from their rabbis. Notably that gap has closed by 2020. Mainline Protestants show consistent gaps throughout, with clergy and congregations the least supportive and their political divisions the most, but all have moved more supportive since 2016. Catholics begin the same place as mainline Protestants in 2016, but the gap between contexts has fully closed by 2020, with the most movement found among priests. Notably, the nones are the only group who have friends out of step with their communities and congregations (yes, some have congregations).

Why? One possibility is that Trump supporters have become less shy over time. Trump is now arguing that the silent majority will speak on November 3, trying to explain why the “fake polls” are showing him with a huge deficit to Biden. There may have been a hint of that in 2016, but most accounts now find little evidence to support the existence of shy voters. The appearance of growth in Trump support could simply be people feeling free to express their support for him across his presidency.

As political scientists are excited to show, politics in congregations is likely to cause some people to leave and extreme religious politics (e.g., the Christian Right) is associated with people de-identifying from religion altogether. Donald Trump has wrapped himself in evangelicalism and evangelicals have returned the embrace. Perhaps the link of religion with the most controversial president in at least modern history has pushed liberals and moderates to stop identifying with the faith, meaning more Trump supporters left behind. Are liberals and moderates surrounded by more conservatives in their congregations?

The figure below plots personal Trump support (via a feeling thermometer) and perceived congregational support in 2016 and 2020. It is apparent that the two don’t quite march in lockstep – ardent Trump opponents see an average support score in the upper 30s in their congregations, whereas Trump supporters see an average score of 75. Those scores do not vary much by 2020. The average difference in personal versus congregational support is 23 in both years, though it varies a bit by religious tradition. All three groups averaged 23 in 2016, but diverged slightly by 2020 – evangelicals averaged 21, while mainliners were at 24, and Catholics were at 26. That is, Catholics were slightly more diverse in 2020, while evangelicals were slightly less diverse. At most, evangelical congregations became slightly more homogeneous, but it was not a trend across white Christian American religious groups.

This question about the sorting of American religion based on politics is not a new one and has been addressed before, but in a different way. Michele Margolis argues that when people tend to take some gap years from religion in their 20s, their partisanship solidifies before their religious identity. Therefore when the question arises about finding a congregation again, partisan stereotypes guide them – Democrats stay away from religion, while Republicans head back. That process skips over politics in congregations and should result in 1) ever more conservative congregations and 2) more homogeneous congregations. While Ryan Burge doesn’t find much evidence of taking gap years from religion (aka the life cycle effect) anymore, it’s important to note that other processes may be at work in polarizing religion than the one rooted in congregations I laid out above.

The difference between the two approaches is important. One (mine) suggests that congregations host some level of diversity, which incentivizes clergy to work to create conditions that maintain that level of diversity rather than lose members. Therefore, clergy are often careful about whether they bring up politics as well as how they talk about political issues. The other approach (Michele’s) suggests that congregations are (increasingly) homogeneous, and that some of the most consequential religious decision making happens outside of congregations under the influence of partisanship. Moreover, clergy are almost immaterial to this approach since the decision to rejoin religion is asserted to be stable across the life course. Are congregations important or derivative of individual decisions?

In my view, the US has been roiling under the Trump Administration given its conduct and policies, so it is astounding to find such stability in the general patterns of congregational political diversity that I show in this post. Just like Lewis Carroll’s Red Queen who needs to run faster and faster in order to stay in the same place, I suspect that clergy have been working harder and harder to maintain their place and weather this Trump storm.

Paul A. Djupe, Denison University, is an affiliated scholar with PRRI, the book series editor of Religious Engagement in Democratic Politics (Temple), and co-creator of religioninpublic.blog (posts). Further information about his work can be found at his website and on Twitter.


Data Note. I gathered the 2016 data with The Brian Calfano (1000 white Christians). The 2017 data were part of a 3 wave panel of American adults I gathered with Amanda Friesen and Anand Sokhey – we added a fresh cross section to that panel in May 2017 with Andrew Lewis (total of 1137 cases). The May 2018 data were collected from American adults through Qualtrics Panels with Andrew Lewis, Jake Neiheisel, and Anand Sokhey (1429 cases, gender, age, and region Census quotas). Finally, the March 2020 data were collected with Ryan Burge and Andrew Lewis (3,100 cases) through Qualtrics Panels using gender, age, region, and race quotas set with Census figures.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s