I came across my article about that data last week and thought it might be interesting to re-create. So, using county-level demographic and voting data, I made maps comparing the 2016 and 2020 elections.
The metric on which these are measured is a bit complicated, so it’s worth a quick explanation. In 2016, the average number of votes cast in a county was 57.5 percent of the number of residents over 18. So the maps show counties which exceeded that percentage and those that fell short of it — in both 2016 and 2020. Since the average was 63.7 percent in 2020, given higher national turnout, I decided to compare apples and apples on the turnout metric. (Those percentages, by the way, are based on five-year population averages for 2016 and 2019, the most recent available.)
So here’s a comparison of three things: the actual turnout percentage, the two-party margin and the density of Whites in each county. The color scales on both the 2016 and 2020 maps are the same, so, for example, if a county is dark red in 2016 and light red in 2020, it voted more heavily Democratic last year, though it still went for Trump.
A few things to notice. First, and most obviously, there are far more counties that passed the 2016 average in 2020. Again, that’s because turnout was up. But you’ll also notice that that meant a lot more counties that went for the Democrat show up in the 2020 map than in the 2016 one. In other words: More blue counties had turnout above the 2016 average.
That correlates to the bottom two maps. A lot more less-densely White counties show up in 2020, since turnout was higher in those places. And less-densely White counties tend to vote more heavily Democratic.
In 2016, counties that exceeded the 57.5 percent average turnout relative to the adult population gave Trump 1.2 million more votes than Hillary Clinton earned. She won counties with below-average turnout by 4 million votes. In 2020, Biden won 2.7 million more votes in places that topped 2016 turnout — and 1 million more votes in counties that also topped the 2020 average.
If we look at the other counties, those with turnout below the 2016 average, the patterns are reversed. Here we’ve added Black and Hispanic population density, since there are more of them on the 2016 maps.
Some of the below-average counties with a high density of Hispanics are places where there are a lot of Hispanic residents who are not citizens and therefore can’t vote. That caveat aside, there are obviously fewer Hispanic and Black counties in 2020 that were below the 2016 turnout average than there were four years prior. Also compare California in 2016 with California in 2020. A much different picture, even if the electoral-vote picture remained the same.
What stands out on these maps are the densely White places where turnout was below the 2016 average. It’s a chain of counties that follow the spine of the Appalachian Mountains — territory where Trump did very well. Its turnout was relatively low in 2016, too, but it’s still worth noting, given the election outcome.
If we break turnout into 5-percentage-point chunks and find the average density of the White population, the result looks like this.
Almost uniformly, the purple dots (2020) are to the left of the orange dots (2016) as you move up the scale — indicating that the density of the White population in counties was lower at each interval of turnout. In other words, more densely non-White counties were turning out at higher rates in 2020, pulling the average White density down. In 2016, counties that had above-average turnout were on average 83 percent White. In 2020, counties that were above the 2020 average for turnout were on average 70 percent White.
This doesn’t necessarily mean that the difference was made by non-White voters. It could be increased turnout of White people in less densely White areas, like large cities. But it does show one way in which Trump was unable to re-create his success in 2016.