By Ryan P. Burge, Eastern Illinois University

Donald Trump has said some amazing things in his illustrious career. One of my favorites Trump clips comes from a press conference after he won the Nevada Republican primary in February 2016. He notes, “I won the evangelicals.” Then later he famously states, “I love the poorly educated.” I think about those two statement probably more than I should. But, I’m always interested in understanding whether Trump’s support is more rooted in educational attainment or is more strongly tied to religion. Fortunately, I have the data to test that.

I love the Cooperative Congressional Election Study. It has become my “go to” source for understanding contemporary politics and religion. The main reason: huge samples. Both the 2016 and 2018 versions have more than 64,000 respondents. That means that even if I am looking at just white evangelicals, I start with nearly 10,000 people. Now, I can subdivide that even further. That means I can look at white evangelicals who have a bachelor’s degree and attend church once a week and still have a decent sized subsample. That’s amazing. So, that’s what I did. I used the 2018 CCES and because Trump was not on the ballot, I used Republican identification as a proxy. I think it’s a good proxy of Trump support, because 90% of Republicans approve of the job that he’s doing. Does Trump do better among poorly educated Americans? Or low attenders? Or is it a combination of both?

The heat map above is the percentage of each square that identifies as a Republican based on each level of education and church attendance for white evangelicals. Note that the size of these squares begins around 50 respondents (never attending with a graduate degree), but goes all the way up to 855 (attending weekly with a high school diploma). The average square has 257 respondents, which gives me decent statistical power.

First thing that jumps out is that the bottom right is very red, and the top left is grayish-blue. The most die hard Republicans are those without a high school diploma who attend church more than once a week. The least Republican are white evangelicals who never attend and hold a graduate degree. The other thing that is noticeable is that as you move from left to right, the columns move from gray to red. It’s clear that there is a positive relationship between church attendance and Republican identification for white evangelicals.

However, what about the rows of the heatmap that denote different levels of education? Clearly education matters less than church attendance. For instance, take a look at the far right column – people who attend more than once a week. If you scan up from the bottom, you notice that each square becomes slightly less Republican but the differences are relatively small. Considering that someone with a graduate degree is just 8% less Republican than one without a high school diploma, that’s not a substantial difference. This pattern exists for most of the columns as well – the education differences are small compared to the church attendance differences.

The second heatmap is just for white Catholics, and obviously there is far less red here and much more gray-blue. The most Republican squares are devout Catholics with moderate levels of education. However, as one moves to the middle of the heat map the Republican support begins to significantly wane. Catholics in the middle of this heatmap are 8-10% less Republican than their evangelical counterparts.

In addition, education seems to play a stronger role here. Remember that there were small differences in the attendance columns for evangelicals – the divergence is much larger here. For instance, the differences between the most educated Catholics and the least educated Catholics who attend church the same amount is 15-20%. That’s a significant difference. It appears that education has a way of tamping down Republican support among Catholics that doesn’t exist for evangelicals.

Just for comparison, I made the same heatmap for non-white Catholics and Protestants (of all traditions) just to get a sense of how significant race has become in American political life. As you can tell, we have a sea of gray and blue here. In fact, not a single square even comes close to being majority Republican. The closest this group gets is people who attend more than once a week and have a graduate degree, but even there just one in three identify as a Republican. It’s hard to overstate how important race has become when it comes to looking at religious groups.

I felt like I needed to dig just a big deeper on the relationship between education, church attendance, and political partisanship. I broke the attendance variable into three categories (low, medium, and high) and then estimated a logit regression predicting the share of each group that identifies as Republicans. The first graph is just white Catholics. As you can tell – there’s really no difference in the slope of the lines. The more education Catholics receive, the less likely they are to identify as a Republican. Additionally, more devout Catholics are more likely to be Republicans, while those who attend infrequently are less likely to affiliate with the GOP. That’s what I would expect to see. But what about white evangelicals?

Here, the slopes look different. For low attending evangelicals there is no relationship between education and identifying as a Republican, but for Catholics it was clearly downward. In other words, for evangelicals who attend a few times a year or less their political affiliation looks no different if they have a graduate degree or lack a high school diploma. The slope for those of high attendance is downward, but very slightly. Consider this: a white evangelical who attends once a week and has no high school diploma is only about 5% more likely to identify as a Republican than the same white evangelical with a graduate degree. Here, it seems that attendance is drowning out education. Those in the middle category look like Catholics, however. The relationship here is negative for those attend a few times a year. The more education they obtain, the less likely they are to identify as a Republican. However, large numbers still identify as Republicans in this group – 55% of those with a graduate degree still affiliate with the GOP.

Taken together, the conclusion is clear for white evangelicals – church attendance is a much stronger predictor of political affiliation than education. On the other hand white Catholics seem to demonstrate the negative relationship between education and Republican affiliation that is prevalent among the general public. What might be going on here is a selection effect. People who attend conservative Protestant churches know what they are getting themselves in to. They know that the messages from the pulpit and in the pews will be overwhelmingly conservative. If they didn’t like it, they could leave. Consider that approximately 8 in 10 white evangelicals who attend church more than once a week identify as Republicans. If one continues to attend those churches, the assumption is that this is what they want to be exposed to – and education shouldn’t mitigate that. Obviously, what we can’t see here is the number of people who were raised evangelical, gained a higher level of education, and traded their conservative Protestant affiliation for membership in a mainline church. Or no church at all. The upshot for me is simple: evangelicalism has put down a clear marker and the public has responded. The sorting has probably reached its apex. The educated moderate evangelicals have seen the writing on the wall and have headed for the exits.

Ryan P. Burge teaches at Eastern Illinois University in Charleston, Illinois. He can be contacted via Twitter or his personal website. The syntax for the post can be found in here.