Caught in a Landslide — County-Level Voting Shows Increased ‘Sorting’
The election was close, but not at the local level. More voters than ever live in a county where the victorious candidate won by a landslide. It’s more evidence that we are sorting ourselves into communities of like-minded Americans.
Two facts stand out in this extremely close election:
- Votes continued to shift from Democrats to Republicans in the nation’s small cities and rural counties.
- There was an astounding increase in the percentage of American voters who live in a county where this extremely close election was decided in a landslide.
First, let’s look at the local landslides. Since the mid-1970s, most U.S. counties have become increasingly either Republican or Democratic. (Robert Cushing and I wrote a book on the subject, The Big Sort.)
As a way of showing this trend, we counted the percentage of voters who lived in counties where the presidential election was decided by 20 percentage points or more. In 1976, about 26 percent of the population lived in one of these landslide counties. That percentage has grown steadily since. In 1992, 37.7 percent lived in a landslide county. In ’96, 42 percent. In 2004, 48.3 percent. In 2012, 50.6 percent.
Then, this year, 60.4 of voters lived in a county where an extraordinarily close election was decided by 20 percentage points or more.
That’s a big number. Even bigger, however, is the percentage of rural voters who lived in a landslide county. In this election, more than three out of every four rural voters lived in one of these politically lopsided communities – in an election nationally decided by a fraction of a percentage point.
The chart above shows the change in the proportion of landslide counties from 2004 through 2016 in six different categories of counties, ranging from the most urban to the most rural:
Central Cities. The largest county (or in New York and Washington, DC, counties) in the metropolitan areas of a million or more.
Big City Suburbs. The “ring” counties in metros of a million or more.
Medium-Sized Metros. Counties in metro areas of between 250,000 and 1 million people.
Small Cities Metropolitan areas under 250,000 people.
Rural Near Cities. Nonmetropolitan counties adjacent to metropolitan areas.
Rural and Remote. Nonmetropolitan counties that do not touch a metro region.
All the categories increased their share of landslide counties, but the largest increases were in the rural counties, followed closely by the Central City and Small City categories.
“In a closely fought national election, when you drop down to study particular locations, they might also be closely competitive, or they might be divided into those that tip lopsidedly to each party, showing that the candidates have rival geographic bases of support,” says University of Maryland political scientist James Gimpel. “In this election, we have the latter case: closely fought nationally, but very one-sided locally. This is particularly true at the extreme ends of the urban-rural continuum.”
Now, this is where we need to look at a second chart, the one showing the shift in votes in these same categories. The chart shows the voting percentages in our geographic groups from 2004 to 2016.
Again, the most noticeable shift is in the rural counties, which moved heavily (8 percentage points since 2008) to the Republicans over these four elections.
The smaller city category also shifted Republican, just as the big city counties moved to the Democrats.
“Some view this urban-rural split as a racial division, and there is an element of that present,” Gimpel said. “But it also reflects other big cultural differences by occupation, affluence and religiously-rooted values.”
Put the two charts together and you get the picture of a nation increasingly polarized, where nearly two out of three voters live in a place with an overwhelming political majority. And you can see a nation divided between the central cities in the largest urban areas and the country’s small metro areas and rural counties.
Any common ground between the two sides has nearly disappeared.