by Ryan P. Burge and Paul A. Djupe

At this point in the pandemic, when the virus is spreading out of control in the South and Southwest, American life has changed a great deal on the margins. To adapt to the new conditions, people are cooking again, they bought puppies, and their interest in cycling has skyrocketed. But the virus is not the only thing that stands out during Spring/Summer 2020 – the image of American Christianity has been cast in stark relief against a backdrop of pepper spray. How is the American religious distribution shifting in 2020? Do the shifts have anything to do with politics?

We never have fine-grained tracking poll data on religion – it’s simply too expensive to ask some of the necessary questions to measure religious affiliation. So, we are tremendously fortunate that the Data for Progress have made available tracking polling data in weekly surveys beginning in April, giving us eight weeks of measurements (so far) to draw on with appropriate sample weights to ensure representativeness. Of course we need to issue caveats. First, these are samples and thus subject to sampling error. Given the sample size (N~1000 each week), we do not expect those errors to be large, but they are present (the weights take care of most of this). Second, religious identification measures like the ones we show are often the last to change. That is, when people leave a particular congregation, they often do not stop calling themselves a Protestant. Leaving a religious identity may come later – disaffiliation precedes de-identification. Of course, we need to remain cautious about making too much of any dip in a weekly polling effort. At most, these potential trends need to be monitored further.

Protestant identification has dropped five percentage points over this period from the week of April 16 to June 9. It did not happen all at once, but has been a steady drip across this short period. This does not mean that these people have left religion, nor does it likely mean that they have left Protestant congregations. Instead, they are calling themselves “other,” which has increased by the same amount during this period. Catholic identification rates have remained about the same at 20-24 percent. Given the movement up and down, we suspect there is no trend among Catholics, though it is pointing up during the last four weeks. The smaller groups are remaining stable, but the very small sample sizes prevent us from making a determination about trends. Last, the nones appear to remain stable as well, though both atheists and agnostics are off 2 points from their high point at the beginning of the series. Again, their sample sizes are too small to assert statistical certainty about these changes, but they are worth monitoring to assess whether there are “no atheists in (pandemic) foxholes.”

Let’s get back to those “Others”. Their numbers have grown over the last 8 weeks – who are they? One thing we know about them is that a majority of them also identify as evangelical as the following figure shows. The percentage shifts a bit over the 8 weeks, but is generally stable at just over 50 percent. If “other” is amassing ex-Protestants, then the category appears not to be taking disproportionately from evangelicals.

Even if the growth of Other is not disproportionately from evangelicalism, we still suspect that politics is playing a role in these shifts. Though it is not new, evangelical Protestants have firmly allied themselves with Trump’s Republican Party. The vast majority approve of his job and they even express warm feelings toward him while many evangelical elites are Trump’s apologists. At the same time, popular support for Trump has nudged slowly downward toward in part because of his mishandling of the pandemic. Is there a religious cost to this political support?

As the following figure shows, it appears the answer is yes. Over the eight weeks, Others have become 15 percentage points more Democratic, while Protestants have become about 14 percentage points more Republican (49 -> 63). At the same time, Mormons have become more Republican (by 10-20 points), while the various categories of nones have become somewhat less Democratic (by perhaps 5 points). While we do not have the luxury of tracking particular individuals, the aggregate shifts are at least consistent with the notion negative views of Trump’s Party has taken a bite out of the religious identity of his most fervent supporters – people not on board with Trump’s politics are taking refuge in other religious homes.

The observation is consistent with a number of social science findings over the past several decades. Some see the increase in disaffiliation as the result of choice that follows a partisan script, while others (like Hout and Fischer in 2002 and 2014 papers) see strident politics driving liberals and moderates out of conservative religion (which Djupe, Neiheisel, and Conger verified to an extent). In other perspectives, political disagreement can be one thing among many encouraging marginal affiliates to leave particular congregations. The data at hand do not permit us to peer under the hood to see which account is the correct one. But the results we find are consistent with the general idea behind all of this research, that politics can be divisive and is helping to reshuffle American religion in its image.

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.

Ryan P. Burge teaches at Eastern Illinois University in Charleston, Illinois. He can be contacted via Twitter or his personal website.