2014’s Electorate Showed Up to the 2018 Election. But It Voted Differently.

This was a high-turnout midterm election. But the demographics of those who voted didn’t change much.
Photo: Andrew Caballero-Reynolds/AFP/Getty Images

If you heard it once, you probably heard it a hundred times: the 2018 midterm elections, and perhaps all midterm elections, were “all about turnout.” With the electorate polarized down to its every molecule, the winning equation was simply to identify demographic groups that were in or trending towards one’s own side, then nag and scare and excite and anger and knock and drag them to the polls.

If “base mobilization” was in fact all that mattered, then it would be logical to expect that the shape of the 2018 electorate would be dramatically different from that of the 2014 midterms, in which Republicans had a very solid performance, gaining 13 House seats, 9 Senate seats, and 2 governorships.

But a comparison of exit polls, the best preliminary indicator we have of the shape of the 2014 and 2018 electorates, doesn’t show as much change as you might expect. Yes, the 2018 electorate was much bigger than 2014’s: an estimated 114 million people voted this year, as opposed to 83 million four years ago. But the shape of this bigger electorate is familiar.

To get one issue out of the way, there were actually two national exit-poll operations this year: on the one hand, the traditional Edison Research surveys done for ABC, CBS, CNN, and NBC. On the other, a new survey was conducted by NORC for AP and Fox News. For purposes of apples-to-apples comparisons of 2014 and 2018, let’s focus on the Edison exits.

The white makeup of the electorate was 75 percent in 2014 and 72 percent this year, though the modest difference is mostly attributable to demographic change rather than some sort of voter mobilization effort. African-Americans formed 11 percent of the electorate in 2014 and 12 percent in 2018, another modest change. Latinos are a rapidly rising share of the population, so it’s not too surprising that they grew from 8 percent to 11 percent (they were only 9 percent, by the way, in the NORC exit polls).

Looking at the age distribution, did young people turn out in big and atypical numbers in 2018, as some analysts suggested they might? Doesn’t look that way. Under-30 voters were 13 percent of the electorate in 2014 and 13 percent of the electorate in 2018. The percentage of voters who were over 65 actually went up a tick, from 22 percent in 2014 to 26 percent in 2018 (the percentage of white voters over 65, that famously conservative demographic, was stable at 22 percent).

How about that most notorious category of voter, the non-college educated white voter (a.k.a. the white working-class voter)? Its membership constituted 36 percent of the electorate in 2014 and 41 percent in 2018. That’s not very consistent with a demography-driven Democratic Wave, is it? So is gender turnout a factor? Did women show up in droves to punish Trump and the GOP? Well, yeah, women were up a tick (from 51 percent to 52 percent) as a percentage of the electorate. But white women, supposedly super-mobilized, actually dropped from constituting 38 percent of all voters in 2014 to 37 percent this year.

One difference that might look significant is that the percentage of voters identifying as “white evangelical or born-again Christian” dropped from 26 percent of the electorate to 22 percent. But much of that is simply owing to the general relative decline in the size of the white population, compounded by the erosion of membership that is now hitting conservative Protestant denominations just like their more liberal counterparts. It doesn’t mean Republicans didn’t do as good a job herding the Evangelical flocks to the polls as they have in the past.

So if the electorate isn’t all that different in its component parts than it was four years ago, what did change? It’s hard to say definitively, since it’s always possible that one party or the other did better at turning out their particular share of various demographic groups than the other. But it looks like public opinion changed, with our without partisan efforts to sway it.

The example that jumps off the page in reading the exits is voters over 65. Republicans won them 57-41 in 2014, but only 50-48 in 2018. That’s about the same margin as in 2006, the last Democratic “wave” election, before the tea party movement-driven realignment of the electorate made “old” all but synonymous with “Republican.” White college graduates shifted from 57-41 Republican in 2014 to 53-45 Democratic this year. By contrast, white voters without a college degree changed marginally, from 64-34 Republican to 61-37. White women didn’t trend as massively Democratic in 2018 as some of the anecdotal evidence suggested, but did go from 56-42 Republican to 49-49 this year. The 2014 exits didn’t provide a breakdown by race, gender, and education-level, but given the relatively low change in the vote of non-college educated white voters generally, you can figure this year’s 59-39 Democratic margin among college-educated white women was a pretty big shift.

Yes, the exit polls are quite fallible, as the evidence of the undercounting of white non-college educated voters in the 2016 exit polls shows. But if the change of partisan outcome between 2014 and 2018 was strictly a matter of one mobilization machine outperforming another, it would show up pretty dramatically in the numbers.

It’s not fashionable to say it, but perhaps persuasion by candidates and campaigns had a bit more to do with the Democratic surge in 2018 than we might otherwise suspect. And there’s a good chance that objective reality did, as well: the experience of having Donald Trump as president for two years, with a supine Republican Party doing his bidding. It’s worth pondering as 2020 approaches.