One of the aspects of the vaccine discussion that I find intriguing centers on the unvaccinated population. We know that Republicans are more likely to tell pollsters that they don’t intend to be vaccinated against the coronavirus and we know, too, that counties where former president Donald Trump won by a larger margin in 2020 are counties in which vaccination rates tend to be lower.

But the question of community spread of the virus depends on the density of the unvaccinated population. And most of the country’s unvaccinated population lives in counties won by Joe Biden, not Trump.

As I wrote on Thursday, there are a lot of reasons for this, the most important being that large cities have lots of people, including many who voted for Trump. When we’re talking about large populations of unvaccinated people, though, this is important, particularly because cities also tend to exhibit a great deal of partisan and racial sorting. In other words, in those cities with lots of unvaccinated people, it’s fair to assume that the unvaccinated are not evenly distributed.

My story about the unvaccinated prompted an interesting question: what counties that voted for Trump were the most heavily vaccinated? In other words, we know that Trump-voting counties tend to be less densely vaccinated, but which are the most vaccinated? And why?

Let’s start with the counties themselves. Here they are, according to the Centers for Disease Control and Prevention’s Oct. 6 data. (Those data, incidentally, exclude counties in Texas and some in California.)

So what makes these counties special?

In some cases, it’s clearly demographics. Like Sumter County, Fla., home of the mega-retirement complex the Villages. In others, it may be proximity to more-blue areas, like Delaware County, Ohio, just outside of Columbus. But the overarching pattern I noticed is that there isn’t an obvious pattern.

To that point, below I created a comparison of those 20 counties with Trump-voting counties overall. It looks at six metrics — vote margin in 2020, density of White, Black and Hispanic populations, the percent of the population aged 65 and over and the percent living in poverty — and compares the 20 counties with the average for all Republican counties. I also included the split within the 20 counties relative to the overall average; that is, the number of the 20 counties that were over or under the average.

Vote margin
White %
Black %
Hisp %
Over 65
Poverty
All counties
Trump +42
81%
6%
7%
19%
15%
Most vaccinated
Trump +24
73%
6%
5%
21%
11%
Over average
3
11
6
3
10
4
Under
17
9
14
17
10
16

A few things jump out, some of which we’ll come back to.

For now, let’s focus on the fact that these counties were much less favorable to Trump than others. If we look at the data from our first graph (that depiction of the unvaccinated population by 2020 vote) in another way, we see why that’s the case. The darkest-blue counties are the most vaccinated; the darkest-red, the least.

Notice, too, that the largest circles on that graph — that is, those that cast the most votes for Trump — are ones that represent counties that voted for Biden overall. It’s a reminder of a data point I mentioned on Thursday: 45 percent of Trump voters live in Biden-voting counties.

But, again, the point is the correlation between 2020 vote and vaccinations, as depicted in the table and in that beeswarm graph. If we use a more-traditional scatterplot, we see the correlation: the dots generally slope from upper left to lower right, meaning that vaccination rates drop as vote margins shift more to Trump.

Now, though, we come to those other demographic factors depicted in our table. Like the percentage of older people in the county. There’s not much difference between the average in the 20 most-vaccinated counties and overall. In fact, half of those counties had more older people than the average for Trump-voting counties and half had fewer.

So let’s look more broadly at vaccination rates by 2020 vote. Below, I’ve grouped all the Biden-voting counties into 10 buckets, with the 10th of those counties that had the lowest vaccination rates in one group all the way up to the 10th of Biden counties with the highest vaccination rates. Then I did the same with Trump-voting counties and figured out the average vaccination rates, vote margins and demographic factors for each bucket.

That allowed me to make cleaner scatter plots comparing red and blue counties. Like so:

Some notes on some of the graphs.

Vote margin. Notice that the drop-off in Trump-voting counties was sharp. In other words, the Trump counties that are the least vaccinated were far more supportive of Trump than the Trump counties that are the most vaccinated. Counties that preferred Biden by wider margins are more vaccinated, but less dramatically. (You’ll notice, too, that the Democratic values jump around more; that’s in part because there are fewer of them.)

White population. One reason there’s no effect of White population in our 20 most-vaccinated Trump counties is that they’re all pretty White. In Democratic counties, there’s a bigger relationship between the density of the White population and vaccination rates.

Black population. That’s in part because there’s a strong correlation between the density of Black residents in a Biden-voting county and how densely vaccinated it is. This has two contributors: White Democrats are more likely to say they’re vaccinated than Black people (though Black people are about as supportive of vaccines as White people overall, thanks to White Republican opposition) and because of apparent systemic issues.

College education. For example, vaccination rates in both red and blue counties track with the density of college-educated residents. In both Trump- and Biden-voting counties, the correlation between the density of the Black population and vaccination rates is about the same as the correlation between college graduates and the vaccination rate.

Poverty rates. Education also correlates with income, which itself correlates with race. In Trump-voting counties, poverty was the strongest indicator of vaccination rates: more poverty means less vaccination. In Democratic counties, the density of the Black population was, though not by much.

It’s worth breaking out the question of race and education separately. If we take all counties (not just red or blue ones individually) and compare the density of the Black population with the density of college graduates, we can measure how vaccine density shifts.

This time, I used quintiles (buckets of fifths) of Black population densities and the density of college graduates. What emerges is that there’s not much variation in the vaccination rates of the Black population regardless of the density of college graduates, though there is more vaccination in the counties with the most college graduates.

What sets apart the Trump counties that are the most vaccinated? A few things, including that they tend to be better educated than most Trump-voting counties and, relatedly, have fewer residents in poverty. The median income in those counties is about 20 percent higher than the average across Trump-voting counties.

But this, again, correlates to 2020 vote. What sets apart the Trump counties with the highest vaccination rates? They’re also the least Trump-y.