The votes-per-acre paradox

One of Donald Trump’s favorite places in the country is Sweetwater County, Wyoming. Not because there’s anything in the county that’s particularly compelling beyond its natural beauty. No, Trump likes Sweetwater County (even if he couldn’t name it) because it is 1) very large and 2) voted for him by more than a 50-point margin each time he ran for president.

In other words, Sweetwater County is one of the largest splotches of red on the maps of election results that Trump likes to wave around in the Oval Office. Only about 41,000 people live there, but it punches above its weight in making the U.S. look much more red than it actually is.

There is an established (and, for some time, strengthening) link between population density and voting. More urban areas vote more heavily Democratic; more rural, more heavily Republican.

The gap between the most urban and the most rural counties narrowed slightly in 2024 but, relative to 2012, it’s still yawning.

One might assume, as I did, that this means that comparing vote counts to the land area in which voting takes place would mean that the average number of square miles (or, as the charts below indicate, square meters) is lower for Democratic voters in denser parts of the country than it is for Republicans.

Using precinct-level data from 2016, though, we see that, averaged across states, it isn’t. On the chart below, dots below the line indicate an average square-meters-per-voter that’s higher for Hillary Clinton voters than it is for Trump voters in that state. And, as you can see, that’s the case for nearly every state.

If we use median square meters rather than average, the picture is slightly different. States with lower median square-meters-per-voter tended to see more square meters per Trump voters than Clinton. States with higher square-meters-per tended to see more square meters per Clinton voters.

You’ll notice that most of the lower square-meters-per-voter values are in states that voted for Clinton. And that, it turns out, is most of the story.

If we look at the average square meters per Clinton voter in individual counties, we see how urban areas stand out. They have lower square-meters-per-voter because there are more voters in a smaller space. In rural Nevada, there’s a much higher value — more space and fewer Clinton voters.

If we do the same analysis for Trump voters, we see a key difference: much more vote density in the areas around the cities that stand out above.

In other words: more Trump voters in the suburbs, increasing the denominator in our calculation.

Since Trump won more rural areas and Clinton won more urban ones, the ratio of area to votes spikes for the Democrat in places where she got blown out. In Sweetwater County, for example, there were 8.4 square kilometers per Clinton voter compared to 2.2 for each Trump voter.

If we look only at the averages in precincts each candidate won, the chart looks more like what I would have expected at the outset. In nearly every state, Clinton voters cast ballots from more densely-packed areas than did Trump voters.

As I was mulling this over in the first place, I thought it would be interesting to see whether the understood correlation between population density and voting could be measured using these data. I reached an unsatisfying answer: Sort of.

Photo: Trump at a campaign event in 2020. (White House archive/Flickr)