We're in the final hours of the 2018 mid-term elections, and there's more than enough information out there to keep you busy until the polls actually close Tuesday night. However, I wanted to take this opportunity to answer some questions I've been receiving from students and other political observers about this election cycle.
Is Everything Really Different Now?
A number of people have asked me if the unwritten rules governing American political campaigns have been thrown out the window. After all, Donald Trump violated so many norms and still managed to win the GOP nomination and the general election.
But I don't think Trump's success means the old rules don't still apply. For one thing, the 2016 election ended up remarkably close to what a "normal" election would have looked like. Modest economic growth following two terms of a Democratic presidency tends to yield a slender victory for a Republican.
For another thing, various forecasts seem to be lining up. My simple "fundamentals" forecast model—based on economic growth, presidential popularity, and the number of House of Representatives seats held by the majority party—predicts Democrats will pick up around 38 seats this year. The most recent polling-based forecasts, aggregated by FiveThirtyEight, suggest Democrats will pick up 35 to 40 seats. The Cook Political Report is suggesting 30 to 40 seats. This doesn't mean that outcome is guaranteed, but if economic forecasts (with no polling information) and polling forecasts (with no fundamentals) are lining up, that suggests the political realm is likely operating as it has in the past.
Are the Polls Wrong?
Polling gets a bad rap. If there are 100 elections, chances are polls will predict nearly all of them within a point or so, but will miss on a few, and everyone will cite those few as evidence that Polls Are Wrong. Indeed, quite a few people insist that polls can't be trusted based solely on the 2016 presidential election, even though nationally they were only off by a bit over a single point.
Don't be fooled by all this noise. The best way to know what's going to happen in an election—short of waiting to see what happens—is to ask people how they're going to vote. Chances are the polls we have are pretty close.
But they may well be off by a point or two. And here's the thing about that—they'll likely be off by a point or two in the same direction across much of the country. The errors, that is, are correlated, usually because pollsters are underestimating or overestimating how some key subgroup is going to participate. The problem is, we can't know that ahead of time.
So even a pretty modest error, correlated across the country, can have enormous consequences. There are enough Senate races within just a few points of each other that such an error could spell the difference in majority control. Same with the House.
What Does the Early Vote Tell Us?
In a recent podcast, Nate Silver thoughtfully explained why he doesn't include early vote statistics in his forecast models. In a nutshell: That data is incomplete, and, as a result, inconclusive.
Yes, early votes can tell us, in some states, how many Democrats, Republicans, and unaffiliated voters are showing up to vote. But not only doesn't that tell us how they voted, it's also a biased sample of the electorate, and we don't know how it's biased, or by how much, until the election is over. All this makes early vote statistics a pretty poor predictor of actual voting results.
Now, there may be other reasons to study the early vote. It can be an interesting indicator of voter enthusiasm, or of various turnout efforts by the respective parties. And overall turnout levels are themselves a phenomenon worth examining. But following this won't give you a great idea of who's going to win or lose on Tuesday.
What's the Lesson of This Election?
A mid-term can be "about" a great many things, as it will turn on hundreds of state and local elections. However, this election, more than most mid-terms, is extremely national in its focus. Voters overwhelmingly say that they will base their congressional vote on the candidates' attitudes toward Trump rather than on any single issue.
What's more, part of a campaign is an attempt to define just what the election is about. To a very large extent, Trump has attempted to define this election as one about race and nationality. Just as he made the filling of a Supreme Court vacancy a referendum on whether victims of sexual assault should be taken seriously, so has he made this election a referendum on white supremacy. For evidence, see his late-game advertisement—which suggests that new immigrants will murder police officers—and his decision to dispatch thousands of soldiers to the Mexican border to repel the migrant caravan.
This all means many political observers and Republican elites will be attempting to divine a message from the election returns. If Republicans fare better than expected, that would likely be interpreted as a signal that explicitly nativist and racist messaging is a winning formula for party candidates, and 2020 would likely bring more rhetoric along those lines. And if Republicans perform worse than expected, the result could be construed as evidence that nativist and racially charged messaging is not a winning ticket. Basically, the question for the GOP is whether the increased enthusiasm by racially resentful whites offsets the losses of racial minority groups and more moderate women.
There's no precise science in this interpretation, of course. Many Republican strategists believe Trump's nativist rhetoric paid off for him in 2016—he is in the White House after all—ignoring the fact that he underperformed a generic Republican candidate by several points and is likely less popular than he should be given the strong economy because of such language. But the campaign to define the lesson of the mid-terms will begin almost the moment the polls close on Tuesday, and it will have a substantial effect on 2020.