2018 midterms

FiveThirtyEight’s Forecast Shows Very Tight Battle for the Senate

Republicans are favored to hang onto the Senate, but it’s a near thing. Photo: John Lund/Corbis via Getty Images

There hasn’t been remotely as much discussion of the Senate as the House in speculation about the 2018 midterms. Part of that disparity is simply because the Senate landscape is so wildly skewed toward Republicans (with ten Democrats running in states carried by Donald Trump last year, as opposed to just ten Republican incumbents — most in safe seats —altogether) that it often did not look like control of the chamber was in question. But another reason is that there are fewer objective tools with which to analyze the Senate picture. Because Senate races are only happening, by definition, in a third of the country, there’s no “national generic” vote, and in fact, national polling in general (say, of the president’s approval ratings) is of limited utility.

So highly quantitative assessments of Senate races have been few and far between. That’s why the release of FiveThirtyEight’s Senate Forecast is of considerable interest. As Nate Silver explains it, his model relies on various sources of data:

[We] take lots of polls, perform various types of adjustments to them, and then blend them with other kinds of empirically useful indicators (what we sometimes call “the fundamentals”) to forecast each race. Then they account for the uncertainty in the forecast and simulate the election thousands of times. Our models are probabilistic in nature; we do a lot of thinking about these probabilities, and the goal is to develop probabilistic estimates that hold up well under real-world conditions. 

So his forecasts have probabilities of victory in both individual races, and for the overall outcome nationally. The latter is what’s going to get the most attention: FiveThirtyEight currently forecasts that Republicans have a two-in-three (or 66.1 percent) chance of hanging onto the Senate, meaning Democrats have a one-in-three (or 33.9 percent) chance of flipping the chamber. That’s all from Silver’s “classic” forecasting methodology, which includes polls and “fundamentals” such as past voting results, historical trends and fundraising; there are more and less complicated versions that add or subtract certain kinds of data and produce marginally different probabilities.

When you get to specific races, FiveThirtyEight is more precise than other rating services that make largely subjective “calls” on who’s ahead and by how much. Each Senate race, like the national contest, is assessed in terms of the probability of victory. Races where no one has more than 60 percent odds of winning are toss-ups; those with 60 to 75 percent probable winners lean one way or the other; and those with 75 to 95 percent probable winners are likely to go to the favored party.

Right now this system projects two toss-up races, in Republican-held Nevada and in Democratic-held Florida. Aside from Florida, incumbent Democrats, remarkably, lead in all of the states carried by Trump in 2016. They are likely winners, in fact, in Indiana (Joe Donnelly), Montana (Jon Tester), and West Virginia (Joe Manchin), and solid favorites in Michigan (Debbie Stabenow), Ohio (Sherrod Brown), Pennsylvania (Bob Casey), and Wisconsin (Tammy Baldwin). That leaves just two lean-D states, Missouri (Claire McCaskill) and North Dakota (Heidi Heitkamp). Democrats also have a narrow advantage in Republican-held Arizona.

If the Senate elections all break the way the probabilities suggest, and the parties split the two toss-ups, Republicans would hang onto the Senate by the narrowest possible margin, holding 50 seats plus Mike Pence’s tiebreaking vote. But the range of probabilities makes other possibilities entirely feasible. Altogether FiveThirtyEight suggests seven competitive races, three in which Democrats are narrowly favored (Arizona, Missouri, and North Dakota), two in which Republicans have a narrow advantage (Tennessee and Texas), and then two toss-ups in Florida and Nevada. A very slight shift in probabilities in a state or two could change the forecast considerably.

Considering the heavily skewed landscape, the notion that Democrats are this close to winning back the Senate is astounding. If it happens, the consequences would be even greater than a flip in control of the House. Republicans would lose the control over judicial confirmation, which Mitch McConnell has used to give Trump exceptional power over the future shape of the courts. Republicans would also no longer be able to pass legislation on strict party-line votes using the budget reconciliation process. And senators have at least as much investigatory power as members of the House, with which to hold the administration accountable.

There are enough close races, however, to make the outcome quite uncertain, and as we learned in 2016, probabilities have two sides, particularly if they are not particularly lopsided. In 2012, Democrats seemed to win all the close Senate races, and Republicans pulled off the same trick in 2014. It could easily all come down to late returns from somewhere in the West on the evening of November 6 — or later than that.

FiveThirtyEight Forecast Shows Tight Battle for the Senate