The Most Republican Jobs in America? Start with the Church.

An incredibly new databases opens up all kinds of possibilities for analysis.

I am running a holiday special - 20% off an annual subscription to this newsletter. That brings the cost down to just $48 per year. Take advantage by clicking this button:

---

Boy, do I have a treat for you all today! Do you ever wonder how the partisanship of people who work for religious organizations varies compared to folks who work in other industries? Of course you do—and I think I can answer that question with a high degree of specificity, thanks to an amazing data collection effort from a team of researchers.

They created what they call VRScores, a database of over 500,000 people that links their workplaces with their voting records. I know you have lots of questions about how that’s even possible, and I don’t want to bog you down with all the nuts and bolts, but here’s the simple version:

Some states require you to declare yourself as a Republican or a Democrat so they know which ballot to give you during a primary election. These are called closed primary states. You can find a map of those here. A company called L2 has aggregated that registration data with a bunch of other variables, which the VRScores research team then matched with employment data from Revelio Labs. Of course, there are lots of states that don’t have closed primaries, so the research team used sophisticated prediction models to extrapolate to a much larger sample.

I know some of you really want to dig into this more, so here’s the methodology page from Politics at Work. And here’s a link to the preprint of the paper where they lay out all the nitty-gritty details on SSRN. I was skeptical of the accuracy of this data too, dear reader—but after working this dataset over in a variety of ways, I can say this: it absolutely passes the sniff test. It’s really good.

Let me show you what I mean. This first graph is just me grabbing a handful of the largest employment sectors in the data and comparing the partisan lean of those sectors to folks who work in “Religious Organizations.” (Don’t worry—I’ll get more granular later.)

The partisan lean is a really simple calculation—it’s just the Republican share minus the Democratic share in each industry. Is anyone shocked that people who work in higher education are 32 percentage points more Democratic than Republican? Of course you aren’t, because that’s certainly the perception. The next most left-leaning sector is people working in law offices, followed by software engineers. Again, that definitely has a lot of “face validity” to it.

There was one large industry sector with zero lean at all—and that was real estate agents. Which actually makes quite a bit of sense from my interactions with Realtors. There are certainly more right-leaning folks in that segment of the economy compared to those who work in higher education. But people who worked for religious organizations are clearly the only group to the right of center. Among those employed in this sector, Republicans outnumbered Democrats by seven percentage points.

That tracks, right? So let me show you an expanded list of the 50 largest industries by total number of employees. The metric I used here was the share who were Republicans.

I think this list looks really similar to what I would expect if I were asked to rank the most liberal professions and the most conservative ones. The most left-leaning include people working in internet services, clothing retail, media and streaming, higher education, and fast food restaurants. In each of those cases, no more than 30% of employees identified as Republicans.

But as you move up the list, a lot of the usual suspects start showing up near the top. As you can probably guess, folks working for religious organizations are the most likely to be Republicans—just below 50%. Right behind them are people who work in commercial construction and real estate appraisal. The next two on the list are insurance brokers and real estate agents.

Again, I think these scores line up well with our general understanding of where various professions fall on the political spectrum. But the conclusion—both in this graph and the prior one—is pretty simple: people who work for religious organizations are, on average, more Republican than any other large employment sector in the United States.

Here’s what I thought would be a fun little data exercise: I pulled out ten industries that most closely resemble folks who work for religious organizations in terms of politics. I did this for both the data from the first time this was collected in 2012 and the most recent version from 2024.

Those real estate appraisers show up a whole bunch, don’t they? In both 2012 and 2024, they were among the industries most politically similar to people working in religious organizations. There are a bunch of sectors that appear in both lists, really. For instance, folks who manufacture aerospace instruments show up twice. So do car dealers and insurance brokers. Engineers and people working in the mortgage industry, too.

It’s an odd assortment when compared to religious organizations, right? I don’t know how much a used car dealer has in common with a pastor, but they do tend to have a similar partisan composition. I think there’s a good reason for this, though: almost every group on this list either tends to have a high income or works in an industry that’s heavily regulated by the government. So it logically follows that these folks would favor a party that focuses on lower taxes and less regulation.

I don’t think the vast majority of people who work in religious organizations are focused on economic issues—my guess is that they’re more concerned with topics like abortion, same-sex marriage, and other questions of morality. So these groups aren’t at all alike on many issues, but they still manage to support the same candidates.

Now here’s where my data-analytics heart really starts to sing. The dataset I’ve analyzed so far is pretty easy to parse because it has a column called NAICS code (which is a unified system for categorizing people into industries like mortgage brokers or car dealers). It doesn’t take much work to classify folks when you have a variable like that. But in this case, that’s a pretty blunt instrument—it lumps together priests, pastors, imams, rabbis, church secretaries, support staff, and everyone else into one single category.

I wondered if I could get into the weeds a bit more by poking around a dataset that actually lists the name of each person’s employer—not just the generic industry they work in. So I set out to do that with the help of ChatGPT. I basically had it look over the list and generate a set of terms that might have religious connotations. It pulled together a corpus of words like “church,” “synagogue,” and “Catholic,” then identified the rows in the dataset where the employer name included any of those key terms.

I iterated over this a couple of times to make sure it wasn’t pulling in folks who worked for Catholic hospitals or religious universities, and then I set out to validate my filtered data to see if I had managed to pull together a solid subset of people truly working in religious settings.

Here was my first test:

I think I passed with flying colors because the partisan lean of those folks identified as religious employers is exactly what I would have expected. While the rest of the data—over 500,000 people—had a clear Democratic lean (about 11 percentage points), the composition of my sample of religious employers leaned about five percentage points toward the Republican side of the ledger.

That’s exactly what I wanted to see, and it validates the data-cleaning exercise. This five-point estimate is almost identical to what you saw in the first bit of analysis I did here, where folks working in “religious organizations” based on NAICS codes leaned around seven percentage points to the right.

Feeling pretty good about myself, I wondered if I could subdivide the data into larger religious families. So I had ChatGPT generate a list of words that would clearly identify someone working for a Muslim, Jewish, or Christian house of worship. Then I ran the same calculation as before—the partisan lean of each subgroup.

And wow, this result is better than I could have expected. Among folks who worked for Muslim organizations, the partisan lean was 65 points toward the Democrats. For those who worked for Jewish groups, it was D+50. For anyone who’s read my work before on how different religious groups voted in the 2024 election, you’ll know that these results align really well with what we already know about how those groups tend to vote.

That became even more apparent when looking at the results for people working for Christian employers, who were about 15 percentage points toward the Republicans. This makes it clear to me that the reason folks working in generic “religious organizations” weren’t more Republican overall is because the non-Christians in that subgroup were pulling the average to the left. If you look only at people working for employers with words like “Catholic,” “Baptist,” or “Pentecostal” in the name, you find that these folks may be among the most Republican-heavy sectors of the workforce.

But wait—there’s more! I took this down one more level of abstraction, to individual Protestant denominations, by using key words like “Methodist” or “Lutheran.”

And you better believe this data looks exactly like I would’ve guessed before I even generated the graph. Folks who work for Baptist and non-denominational churches are the most right-leaning, followed by Pentecostals and Lutherans. In each case, the Republican share outnumbers the Democratic share.

So which denominations tend to be more Democratic-heavy? It should come as no surprise that it’s the Episcopalians, the Adventists, and the Methodists. This tracks perfectly with what we know about how both the clergy and laity feel about politics in those more mainline traditions.

Man, I love doing my job—especially on days like this. I’m working with an incredibly cutting-edge dataset that would’ve been impossible to even conceive of just a few years ago, and it’s adding to our understanding of the world in ways I probably haven’t even thought of yet.

But rest assured—I’ll be coming back to VRScores more than once in future posts.

Code for this post can be found here.

---

Ryan P. Burge is a professor of practice at the Danforth Center on Religion and Politics at Washington University.