The Washington PostDemocracy Dies in Darkness

The demographic surges that spurred 2018′s heavy turnout

Analysis by
National columnist
November 13, 2018 at 10:09 a.m. EST
Polling workers help people to check in at Miami Beach Fire Station No. 3 on Tuesday, Nov. 6, 2018, in Miami-Dade County, Fla. (David Santiago/Miami Herald via AP)

On any Election Day, the first analyses of turnout are anecdotal. Long lines at particular places. Commentary from poll workers about how the turnout looks relative to prior years. Stories of unexpectedly long lines or the unexpected absence of same. As with any similar reporting, caution is warranted in extrapolating: Just because the line at your polling place was long doesn’t mean that turnout nationally was unusually high.

Sometimes, though, that line isn’t giving the wrong impression. Last week, turnout was unusually high, according to estimates from the University of Florida’s Michael McDonald. It lagged behind presidential-year voting, but that has happened consistently for nearly 200 years.

Compared to other midterms, 2018 turnout is estimated at the highest in a century, edging out a surge in turnout in the middle of the last century that peaked in 1966.

Put another way: Midterm turnout is estimated to have been higher than at any point since women were given the right to vote.

But why?

Exit polling would suggest a surge in older voters. In 2016, voters aged 65 and older were estimated to make up 16 percent of the electorate. Last week, exit polls had that figure at 26 percent.

Those numbers probably aren’t directly comparable. As the New York Times’s Nate Cohn notes, the methodology used by Edison Media Research, the go-to exit polling firm, changed between the two elections. This year had a higher-than-normal midterm turnout election, but 2016 turnout was still higher. So it makes sense that the density of older voters would be higher in 2018 than 2016, given that older people tend to vote more regularly than younger people. But a 10-point jump? That should raise some eyebrows.

Catalist, a left-leaning data firm, did its own analysis of the electorate in 2018, using its voter file and reported data to compile estimates for the composition of the electorate that can be compared to 2016 and the 2014 midterms. It reports a jump in 65-and-older voters from 2016, but of seven points, not 10. Correlated: Density of younger voters was down by five percentage points.

But comparing to 2016 doesn’t necessarily tell us why midterm turnout was up. Comparing to 2014 answers that question more directly. Since 2014, the percentage of the electorate that is under the age of 40 was up more than three percentage points. Voters under 30 made up 2 percent more of the electorate than in 2014, as did voters age 30 to 39. Those younger voters, exit polls tell us, voted overwhelmingly for Democratic candidates.

Now bear in mind that composition of the electorate is evaluated in percentage points, not number of voters. So if the number of voters surged in 2018 relative to 2014, there were increases in the number of voters in each group. It’s just that the younger and older voters had a larger relative increase than did voters age 40 to 64. There were still far more voters in that midrange age group! But fewer relative to younger voters than in 2014.

Shortly before the election, The Post and our partners at ABC News released a poll looking at how different educational and gender groups planned to vote. Among white women with a college degree, the Democrats had a 16-point advantage. Among white men without a degree, the Republicans had a nearly 40-point lead.

So how did those groups turn out? Catalist’s data offers some insight.

White men without a degree made up the same percentage of the electorate as in 2016 but, according to Catalist’s data, 3 percent less of the electorate than in 2014. White women with a degree made up slightly more of the electorate relative to the last two elections.

These are subtle changes. When we overlay another factor -- urban vs. rural -- the figures stand out a bit more.

Whites with college degrees (who backed Democrats over the Republicans regardless of gender) who live in suburban areas made up four percentage points more of the electorate than in 2014. Those without a degree in suburban areas made up three points less of the electorate.

There were still more of the latter voters in 2018, but while in 2014 suburban whites without college degrees made up nine percentage points more of the electorate than those with degrees, last week Catalist estimates that the difference was only two points.

That comports with the results of the election. Our analysis of where House seats flipped shows that it was largely in suburban areas. (The dark areas on the chart below show seats that flipped or where Democrats lead in districts held by Republicans. The light areas are seats won by each party that didn’t flip.)

The Catalist data suggest that turnout was up broadly, but that bigger increases relative to the composition of the 2014 electorate were seen among younger and older voters and among whites with college degrees in suburban areas.

There’s another trend noted in that data. The density of the white vote in the electorate has slid downward over time. The 2008 electorate that elected Barack Obama as president had the same percentage of nonwhite voters as did the electorate in 2014. Catalist estimates that the density of the white vote in 2018 was three points lower than in 2014, and nonwhite voters broadly supported Democrats over Republicans.

The turnout surge was across the board, but uneven. Where the surge was larger, it tended to favor the Democrats. As the results indicate.