The Political Bias in Online Sermon Sharing

By Paul A. Djupe, Denison University

Lost in the excitement over the release of PRRI’s new 2020 Religious Census report was a new study by Pew Forum on the sermon content in American Christian churches in the runup to the 2020 elections (thanks for sharing, Val!). They find some shocking results, at least compared to what someone who has studied such things for over 20 years (ahem, me) would expect. That is, they find that evangelical Protestants were the most likely to share a political sermon while Catholics were the least likely; Black and mainline Protestant churches were equally likely. Overall, 67 percent of churches shared at least one sermon that mentioned the election. Not all of these sermons were necessarily divisive – some simply encouraged participation, especially among Black Protestants, but many others discussed issues, candidates, and/or parties.

Why is this unexpected? Because every single academic study of clergy (or congregant reports of what clergy said) has come to the opposite conclusion. They all say that Black Protestants and Catholics have the highest levels of political engagement, while evangelical and mainline Protestants have the least. Those are not just my results and these studies have used a variety of measures to reach that conclusion. Moreover, it has been quite consistent over time, ranging from data from 1995 to more recent data from the last few years (forthcoming in Brian Calfano’s new book).

So, what gives? The difference must lie in who is likely to share sermons online and the competing pressures across religious groups. Naturally, we would expect that sermon sharing would be higher among those who feel greater competition. Especially during the past year, being online was the only way to reach people, but, for this reason, the survey question I’ll draw on asked if the house of worship offered online services before the pandemic. Sharing requires literal and figurative bandwidth that some houses of worship simply don’t have.

How would those two forces relate to politics? I think it will depend on the religious group. That is, Catholic churches are geographically ordered in a way to undermine competition, though not all Catholics observe parish boundaries. Without that competition, politics may not enter the calculation. Evangelical Protestants, however, are unorganized, plant churches where the people are, and tend to face stiff competition. Especially during the 2020 election season, engaging with politics through online sermons strikes me as a good measure of felt competition – it is what the people apparently wanted and certainly would draw more attention than not addressing the biggest story in the US (right along with COVID-19). So, competition may drive up sermon sharing among Protestants, especially evangelicals and non-denominationals, and may drive up the amount of politics in those shared sermons.

We were more concerned with COVID-19 when we posed these questions, but our October 2019 survey of American adults [note] included a question asking, “My congregation put worship services online before the coronavirus pandemic.” Because of my predilections, the survey also asked whether they had heard their clergy address anything from a list of 14 items salient during the election season. Three explicitly concerned the election (Trump, Biden, and the “importance of voting/participating in politics”) and the other 11 covered a range of issues including Black Lives matter, guns, health care, immigration, poverty, and others. Because not everyone reported attending worship services, I am limiting the analysis to those who report attending at least a few times a year. I am measuring religious groups using denominational coding, though I am separating non-denominationals from other Christians.

What Christian groups are more likely to share online? We can see the worth of separating non-denominationals given that many more of them had an online presence before the pandemic. Catholics were the least likely – 25 percentage points less than non-denominationals. Denominational Protestants were online at the same rate – just under half. The pattern appears to implicate market competition. Without a denominational label to help attract people, non-denominationals take their message directly to the people. But the most important point for evaluating the Pew report is that the congregations sharing online represent minorities of most of these religious traditions, save non-denominationals. Of course, now we need to see if online congregations are distinctive in their political engagement.

Do churches differ in their engagement with electoral issues when they have an online presence? Based on the results below, the answer is yes – within every group, the churches online had at least a slightly higher engagement with the elections. But it is only statistically significant among non-denominationals. Offline non-denominationals had the lowest levels of electoral engagement, while Black Protestants online had the highest levels. Catholics were on the low end.

This general pattern comports with the argument of Felipe Mantilla in his new book How Political Parties Mobilize Religion – Catholics tend to avoid partisan politics to protect their brand, while non-hierarchical religious groups like Sunni Muslims or evangelical Protestants are unable to fend off parties politicizing the faith. The pattern, especially the non-denominational one, is also loosely consonant with Richard Nielsen’s Deadly Clerics, in which Muslim clerics outside of established institutions use politics to attract attention and build their following.

What about non-electoral concerns? Obviously, all of the issues we asked about could be engaged during election season, but they are not explicitly about candidates or elections. The figure below shows the average number of topics survey respondents report hearing from their clergy. There are technically no statistically significant differences within these religious groups based on whether they were online or not. However, the average number of topics is only higher for online evangelicals and non-denominationals. And if we put all of the topics together, the only groups that differed significantly by their online presence are non-denominationals and evangelicals – the groups that Pew finds to have engaged elections at a higher rate. While I separate non-denominationals, the most common coding scheme simply allocates non-denominationals who attend regularly to the evangelical category. I show here clearly that they deserve to be singled out. It’s important to note that Pew’s coding decision contributes to the very high rates of electoral engagement of evangelicals in their online sample.

I am not saying that Pew is overclaiming – they acknowledge that theirs is a sample of churches with an online presence and they cannot tell whether it is representative or not. There are a number of reasons to think it’s not – churches with an online presence are younger congregations, have more resources, are much larger, are more likely to be non-denominational, are less likely to be Catholic or Black Protestant, and tend to be more politically engaged. According to previous research, Catholics and Black Protestants are more politically engaged than Protestants. But in this sample, evangelical Protestants and non-denominationals with an online presence are more politically engaged, especially concerning the 2020 elections. That coupled bias – of being online and more political when they are – contributes to a distorted view of American religion, though still important in the view of American Christianity available online.

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


Note: The October survey was administered just before the 2020 election with nearly 1800 survey respondents supplied by Qualtrics Panels. This is not a probability sample, but we did apply quotas for region, gender, and age to closely match Census numbers for adult Americans (18 or over). We then generated raking weights to help balance the sample for those characteristics as well as education and race. The project was reviewed and approved by the Denison University IRB.

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