Banal proclamation about midterms on Tuesday like "It's about turnout!" aside, as I've told readers, we're largely in uncharted territory when it comes to polling for this year's elections and that's because we really don't have any idea what "likely voter models" really represent in 2018, because we're so far outside the norm.
Turnout in 2010 was decent, 41%, but Republicans destroyed the Democrats and picked up 63 House seats. Turnout in 2014 was dismal, 36%, but as a result Republicans regained the Senate and took the largest House margin they've had in generations by picking up 13 more seats. And midterm turnout since I've been alive has largely been hovering around that 40% mark. Back in the sixties and in 1970, midterm turnout was much
higher, 47 or 48%, and Democrats had huge House margins, unthinkable in today's era of gerrymandering down to city blocks.
So what does all that mean in 2018? As Vanity Fair's Peter Hamby reminds us, if Virginia in 2017 is anything to go by
, the polls break down completely if midterm turnout is as high as it was 50 years ago
In the final week before Election Day last November in Virginia, where the commonwealth was electing a new governor, the polls were tightening. Republican Ed Gillespie, was, like Donald Trump before him, tapping into immigration fears by running campaign ads about the threat of the violent MS-13 gang and sanctuary cities. Polls showed Gillespie was suddenly within 3 points of the Democratic front-runner, Ralph Northam, who had held a sturdier lead for much of the year. Were Democrats about to blow another big race in a battleground state? Were last minute Republican efforts to exploit racial fears actually working? Could Northam lose even with an unpopular president in the White House? That scenario would defy the logic of every previous off-year election in modern times, but never mind. The punditry machine, huffing Twitter fumes, gassed itself up. The day before the election, three out of four panelists on MSNBC’s Morning Joe solemnly predicted a Northam loss.
The fourth, Harold Ford Jr., predicted a toss-up, but not without pre-writing an obituary. “Democrats are going to look back and wonder, if he does not win, did we lose on the crime issue, did we lose on the public-safety issue?,” Ford offered. Meanwhile, Mika Brzezinski wondered whether the governor’s race in a state populated by 8.5 million people might hinge on “the Donna Brazile stuff,” a niche beltway Twitter scandal so forgettable that I had to Google it to recall what it was about, despite the fact that I cover politics for a living and grew up in central Virginia. Turns out Brazile wrote a book trashing Hillary Clinton’s campaign, which surely weighed heavily on the minds of voters in Roanoke.
Not only did Northam win the Virginia race, he won by 9 points, a polling error more substantial than anything we saw in 2016. Gillespie, it turns out, actually won more votes than any previous statewide Republican candidate; he had enthusiasm at his back. It’s just that Democrats had more—and they blew the doors off on Election Day. The whole spectacle was yet another blow to polling and to punditry, two industries sullied by Trump’s victory in 2016.
The Virginia result went mostly unexamined after the results came in, as everyone in politics quickly moved on to the latest Trump thing. But the question of why Northam outperformed the polls, and why polls continue to wield such mystical power over the political press, is worth keeping in mind as America prepares to head to the polls next Tuesday. Every piece of evidence we have about voting behavior during the Trump presidency—special elections in various corners of the country, public and internal polls, early voting data in key states—indicates that we are heading for a midterm election with explosively high turnout. University of Florida professor Michael McDonald, who studies voting patterns, estimated recently that almost 50 percent of eligible voters could cast ballots this year, a turnout level not seen in a midterm election in 50 years. Trump, in his way, is loudly trying to juice Republican turnout in red-leaning Senate races by demagoguing the threat of illegal border crossings, which happen to be at their lowest point in decades.
Enthusiasm in this election, though, is mostly fueled by Democrats. Aside from college-educated white women, much of the Democratic coalition in 2018 is comprised of voters—young people, African-Americans, and Hispanics—who don’t typically show up in midterm elections. And the main thing to remember about high-turnout elections, especially ones that bring non-traditional voters into the mix, is that strange things can happen. House seats once thought to be safe are suddenly in jeopardy, like Republican Steve King’ssolidly red seat in Iowa now appears to be.
Still, in the press, it seems written in stone that Democrats will take back the House but fail to take the Senate, thanks to an unfavorable map that has too many Democratic incumbents running in Trump-friendly states like Missouri, North Dakota, West Virginia, Indiana, and Montana. The prospect of a House-Senate split is the most likely outcome according to the polls and veteran handicappers, and that probability has already started congealing into conventional wisdom. New York Times columnist Ross Douthat, writing last weekend, said this scenario is “the sensible thing to root for,” the best way to constrain Trump’s impulses but also an unchecked liberalism.
Polls remain our best tool for reading the electorate and discerning important trends, which is why journalists, handicappers, and campaign managers depend on them so much. Entire media companies are devoted to explaining them. But polls are not predictive. They are wobbly around the margins. Pollsters, the honest ones at least, know this and repeat the warning over and over again. Yet even the shock of 2016 hasn’t stopped people in the media from making predictions about next Tuesday. Journalists, at least on their Twitter accounts, have started to write off certain Senate races. Tennessee is one, North Dakota another. Joe Donnelly was left for dead last week, until a new NBC/Marist poll came out this week, showing him ahead by 2 points. In Nevada, a recent Emerson pollshowed incumbent Republican Dean Heller ahead of challenger Jacky Rosen by 7 points, prompting a chorus of worried groans from Democrats. People who know better urged caution.
“Consistently, the public polling here is garbage,” Nevada political journalist Jon Ralston told me. He pointed out that public polling in Nevada underestimated Democratic performance in every one of the last six competitive statewide elections. In 2010, Harry Reidwas losing to Republican Sharron Angle by 3 points heading into election day. He won by 5.7. “Polls here under-represent Democratic turnout in general. They under-represent Hispanics,” Ralston said. “I don’t know why no one has learned.” Funny stuff happens when people who don’t mainline CNN for a living actually vote.
Then there’s Texas, where Democrat Beto O’Rourke appears to have reclaimed some late momentum against Republican Ted Cruz,who expanded his lead in the race after the Brett Kavanaughhearings energized G.O.P. voters. Right-leaning analysts have fallen all over themselves to mock the endless stream of Texas polling and the glowing coverage O’Rourke has received from the national press. Republican pollster Patrick Ruffini tweeted that “Beto is and has always been fanfic.” The Weekly Standard published an otherwise sensible piece about the race on Wednesday under the headline “Beto-Mania Is a Joke (Probably).” Yeah, O’Rourke might lose. That’s the most likely outcome and the best bet. But here’s a wild concept: he also might win.
So yes, the blue wave scenario is actually really simple: Democrats turn out in numbers that utterly overwhelm Republican voter suppression efforts. If this is another year where turnout is 38-42%, the polls are probably pretty accurate, because that's what the likely voter models are based on. That's still pretty decent news for the Dems as we've seen, at least for the House.
But if turnout is closer to 50%, the likely voter models come apart instantly
. Odds are those additional voters are going to be pulling the levers for the Democrats. At that point, all bets are off as to where things go, but the possibility of not just a blue wave but a surprise blue tsunami does exist. I remain extremely skeptical, precisely because of those wildly successful GOP voter suppression efforts in states where Dems are trying to defend those Senate seats.
But let's remember that even in "high" turnout years, the vast majority of eligible voters still don't show up
in midterms, and that only helps the GOP.
We need to make 50%+ turnout midterms and 75%+ turnout presidential election years the norm