by Nate Silver, 538
Democrats are slight favorites to regain control of the Senate, according to the FiveThirtyEight Senate forecast, which launched today. But the map is wide open, with at least a dozen competitive races — none of which are certain pickups for Democrats — including some states where Democrats are playing defense.
In fact, while it’s possible that Democrats will wind up controlling 54 seats or perhaps even more, the most likely outcome is a much more closely divided chamber, including the possibility of a 50/50 split in which control of the Senate would be determined by whether the vice president is Kamala Harris or Mike Pence. (Joe Biden and Harris currently have a 76 percent chance of winning the presidential race, according to our forecast.1)
Our Congressional model (our forecast for House races will be released soon) is largely the same as the version we built in 2018, which was quite accurate in predicting the number of Senate and House seats that each party would win. We’ve made a handful of changes since 2018, most of which were designed to create more consistency with our presidential forecast, including assuming that uncertainty is slightly higher this year because of an increase in mail voting under COVID-19. But these adjustments don’t greatly change the outlook. For a complete list of changes, see our methodology guide.
As in 2018, there are three versions of the model, which build on one another and become increasingly complex:
The Lite version of the model relies as much as possible on polling. In races that don’t have much or any polling, it calculates the candidates’ standing from other races that have been polled.
The Classic version relies on polling but also incorporates “fundamentals” such as fundraising, incumbency and a state’s partisan lean relative to the rest of the country.
By default, we’re showing you the “Deluxe” version of the model this year. It’s supposed to be the most accurate one and — given everything going on — we’re inclined to cut to the chase. But you can toggle between the versions using the magnifying glass icon at the bottom of the page.
You might want to get in the habit of doing this, too, because there are some fairly large differences between the model versions this year. This reflects the fact that the polling in individual Senate races is generally quite good for Democrats, while other indicators and expert ratings are more equivocal. For instance, the poll-centric Lite version of the model currently gives Democrats a 68 percent chance of winning the Senate, as compared to a 64 percent chance in the Classic version and a 58 percent chance in the Deluxe version.2
Model Talk: Democrats are slightly favored to win the Senate | FiveThirtyEight
These differences stem from the fact that despite their strong polling, conditions for Democrats are inherently a little challenging in the Senate. They need to gain a net of three or four seats to win control, depending on if they also win the presidency. More likely, though, they will need to flip four or five Republican-held seats, because chances are good that one seat will flip from blue to red: Democrat Doug Jones is an underdog to keep his seat in Alabama, which he won against a very bad opponent in a special election in 2017.
Furthermore, while there are a great many Republicans up for re-election this year, only two of them (Colorado’s Sen. Cory Gardner and Maine’s Sen. Susan Collins) are in states that Hillary Clinton won in 2016. Thus, Democrats will have to pick up seats in states that have traditionally been Republican-leaning, such as Georgia and North Carolina. They also have some more ambitious (but redder) targets in races in Kansas, South Carolina and Alaska.
National conditions are pretty good for Democrats — they lead on the generic congressional ballot by 6.4 percentage points, and President Trump is fairly unpopular and an underdog for reelection — but they are not spectacular. (Democrats won the popular vote for the House by nearly 9 points in 2018, for example, a bigger margin than their current generic-ballot lead.) Moreover, Democrats are mostly aiming to defeat Republican incumbents, and even though the incumbency advantage has diminished in recent years, it’s rare that you can take much for granted in races featuring incumbents.
Still, the sheer number of plausible Democratic pickup opportunities is surprising — and favorable for the party — given that the conventional wisdom in 2019 was that Democrats might have trouble finding enough targets to take the Senate. The table below is a list of Republican-held seats where Democrats have at least a 5 percent chance of winning in at least one version of our model. There are 16 (!) of these. In some cases, the model is being exceptionally conservative because of a lack of polling in the state, but the first dozen on the list are realistic pickup opportunities for Democrats.
Where Democrats are most likely to pick up Senate seats
Republican-held seats where Democrats have at least a 5 percent chance of winning in at least one version of FiveThirtyEight’s model, as of 5 p.m on Sept. 17
Democrats’ single best pickup opportunity is probably in Arizona, where Republican Sen. Martha McSally — who lost to Democrat Kyrsten Sinema in the race for Arizona’s other Senate seat two years ago — badly trails Democrat Mark Kelly in polls. Nor is McSally, who was appointed to the seat following the death of Sen. John McCain, likely to benefit much from incumbency, as appointed incumbents typically perform much worse than elected ones.3
Meanwhile, the two Republicans in Clinton-won states, Gardner and Collins, are also behind in polls. Still, it may be premature to write their political obituaries, especially for Collins. She won her last race by a large margin, she has a fairly moderate voting record, Maine only barely voted for Clinton in 2016, and experts continue to rate the race as a toss-up, all factors that help keep her afloat in our Classic and Deluxe models.
The fourth state where Democrats are currently favored for a pickup is in North Carolina, where Democrat Cal Cunningham, a former state senator, leads Republican incumbent Sen. Thom Tillis in polls and is also a slight favorite according to the “fundamentals” our model evaluates. And troubling for Tillis is that he has some of the hallmarks of a weak incumbent: He only barely won his seat in 2014, which was a much better political environment than the one Republicans face now, and he badly lags Cunningham in fundraising.
Indeed, many Republican incumbents, such as Sens. Joni Ernst of Iowa, David Perdue of Georgia and Dan Sullivan of Alaska, face some version of this problem, as many of them won by single-digit margins in 2014, a year when Republicans won the popular vote for the U.S. House by about 6 points. But this year, the national environment favors Democrats by 6 or 7 points, so that’s around a 12-point swing, putting Republicans who won by narrow margins last time in the danger zone.
Other Demoratic opportunities are more idiosyncratic. In Montana, they’re hoping to benefit from the personal popularity of Steve Bullock, who is currently the governor there (although the Republican incumbent is favored in our forecast at the moment). And in South Carolina, Democrat Jaime Harrison has run a surprisingly competitive race against incumbent Sen. Linsday Graham, who once had a reputation for moderation but has now become a major defender of Trump, a shift that may mean he’s now not satisfying voters in either camp. In addition to Perdue’s seat, there is also a second seat open in Georgia, currently held by the appointed incumbent Sen. Kelly Loeffler. (This special election is unusual, though, in that multiple candidates from both parties will face off on Election Day, and if no one gets a majority, two candidates will move to a runoff in January.)
In an average simulation, our forecast has Democrats picking up about six Republican-held seats in the Lite model, around five in the Classic model and about four and a half in the Deluxe model.4 And that would be enough for them to win control of the Senate — even if they fail to capture the presidency.
However, Democrats also have some seats of their own to defend. Most notably, there is Jones’s seat in Alabama. Running for reelection as a Democrat in one of the reddest states in the country puts him at a disadvantage, and this time around he doesn’t have the benefit of running against Roy Moore, who multiple women accused of pursuing sexual relationships with them while they were in their teens. (Instead, his opponent is the former Auburn football coach Tommy Tuberville.) But like Collins in Maine, whose chances shift significantly depending on which version of the model you look at, Jones might not be completely done for. The polls do make it look like Jones is a goner (hence his low chances in the Lite forecast), but the fundamentals — including the fact that he’s raised a lot of money and has a track record of moderation — give Jones an uphill chance at victory.
After Alabama, there is a big drop-off in the likelihood of possible Republican pickups, but their next best chance is probably in Michigan, where John James, who ran a surprisingly vigorous race against Demoratic Sen. Debbie Stabenow in 2018, is within striking distance of Michigan’s other Democratic senator, Gary Peters. Minnesota, New Mexico and New Hampshire also represent long-shot opportunities for Republicans, although those seats will likely change hands only if the night is going very badly for Democrats.
So while we know you’re tired of hearing this … the outcome is, at this point, uncertain.
There are plausible upside scenarios for Democrats where tight polling margins against incumbents like Graham prove to be the canary in the coal mine for widespread Republican problems, and Biden and Harris get to craft an agenda with as many as 54 or 55 Democratic senators.
At the same time, though, the Senate is probably a heavier lift for Democrats than the presidency. Let’s imagine, for example, that Biden flipped Wisconsin, Michigan, Pennsylvania and Arizona and retained all the states Clinton won, which would imply a solid but not overwhelming margin in the presidential race. In that scenario, even if every Senate race went the same way as the presidential election, Democrats would still come up one seat short, picking up Maine, Colorado and Arizona but losing Jones’s seat in Alabama.
And while it’s certainly possible that most of the toss-up races will break in the same direction as the presidential race, Senate races can be quirky. Democrats got a mixed set of results in 2018, for example, picking up seats in Arizona and Nevada — but also seeing four of their incumbents lose.
One thing Democrats don’t have to worry about: If they pick up the Senate, it’s very likely that they’ll have retained the House, too; Republicans are starting out with a big deficit there and are playing at least as much defense as offense. There are scenarios where Republicans could win the House, but they involve the national climate unexpectedly turning into a Republcian landslide, in which case Democrats won’t be competitive in the Senate anyway. But we’ll discuss that more when we release our House forecast. In the meantime, please drop us a line if you have any questions about the Senate.
Nate Silver is the founder and editor in chief of FiveThirtyEight. @natesilver538