The Washington PostDemocracy Dies in Darkness

The fascinating swings in 2016 voting by state and type of county

People vote in 2016 at a polling place in Los Angeles. (Nick Ut/AP)

If you want to understand politics, one of the people you should pay attention to is Cook Political Report’s Dave Wasserman. He is one of those guys who has at his immediate disposal various bits of data specific to the nearest thousandth about how some suburban county in Wisconsin voted in 1956 and how that informs the likely winner of the 2024 Democratic presidential nomination.

This is hyperbolic, but only slightly.

On Wednesday, Wasserman weighed in on the ongoing debate over the extent to which Texas and Arizona could be in play for Democrats in 2020. As we’ve noted, the results in Texas in 2016 alone gave Trump his electoral college win; flip that state and Hillary Clinton is president. As we’ve also noted, Texas was closer three years ago than Iowa, a state Barack Obama won in 2012. And, as we’ve additionally noted, no state Trump won in 2016 was more lukewarm about his performance as president last year than the Lone Star State.

Generally, Texas’s accessibility for Democrats is described as being a function of the extent to which it’s becoming more diverse. But Wasserman points out another important factor: It’s also becoming more urban.

The Department of Agriculture has an index by which it ranks how urban or rural a county is. It’s a nine-point scale, running from most urban (1) to most rural (9). Sixty of Texas’s 254 counties are ranked as a 1 or a 2. Three of Arizona’s 15 are. Those data are from 2013. When the Department of Agriculture did the same index in 2003, only 47 of Texas’s counties were considered that urban.

If we look at how states voted in 2016 and compare them to the density of votes cast in those most-dense urban counties, we see a weak correlation: The higher the percentage of votes cast in counties rated 1 or 2, the better Clinton tended to do.

This in itself isn’t new; the urban-rural divide in electoral politics is by now a thing of common knowledge. If we look at the counties fitting each of the nine density types, in fact, we consistently see that more-urban counties vote more heavily Democratic than more-rural ones. In fact, depending on the state, it’s often only those most urban counties that preferred Clinton to Trump in 2016.

We’ve highlighted Texas and Arizona on that chart. Texas in particular is interesting. The most urban counties went slightly for Trump in 2016. The counties that were slightly smaller voted for Clinton by a little more than 2 points.

But, of course, those more-urban counties are also where more people live. While it’s only the more-urban places that consistently backed Clinton, that’s also where far more people live. This is the old “the electoral map is all red” problem: Empty space may be very partisan, but it’s not very populated.

There were a lot more votes in those heavily urban Texas counties in 2016 than in the rest of the state — more than twice as many, in fact. Trump won that vote narrowly (by about 50,000 votes) and won the state by about 800,000.

That’s a big gap to make up, which is one reason that observers think Texas remains an unlikely pickup for the Democrats.

But it’s worth noting, too, that there were only 15 states in 2016 in which at least 80 percent of the votes cast came from the more-urban counties in the state. Clinton won 11 of them. The four she lost included Pennsylvania and Florida, which she lost by 0.7 and 1.2 percentage points, respectively.

The other two were Texas and Arizona.