If you’re a regular reader of FiveThirtyEight you’ll know that our Senate forecast has said pretty much the same thing every day. When we officially launched our model in early September, it gave Republicans a 64 percent chance of winning a majority in the Senate. Today, the number is similar: 63 percent.
Republicans’ odds have never been higher than 66 percent — a figure they reached late last week — or lower than 53 percent. The informal model updates we published going as far back as March also had Republicans as 55 or 60 percent favorites.
To an extent, this stability reflects the noise-reducing features of the FiveThirtyEight model. Our program examines the polls for signs of statistical bias, and weighs them more heavily when they have larger sample sizes, better methodologies and better track records — which can reduce the impact of outliers. The FiveThirtyEight model is also fairly conservative in estimating the uncertainty associated with each race and the disposition of the Senate overall. At times in the past, the polls in most swing states have been biased in the same direction (either toward Democrats or Republicans).
But this degree of stability is unusual. In pretty much every election we’ve covered, the polls have more clearly broken toward one or another party by this point.
Then-Sen. Barack Obama was an increasingly clear favorite over the course of the 2008 presidential election and almost certain to win by Election Day that year. Obama’s position also became safer over the final 60 days of the 2012 campaign, although with a major hiccup after his first debate against Mitt Romney in Denver.
In September and October 2010, Republicans became progressively more certain to win the U.S. House — while Democrats became more sure of keeping the Senate. In 2012, the Senate broke toward Democrats in September and stayed there in the closing weeks of the campaign.
This year has been different. While there’s been movement in individual states, it hasn’t been in a consistent direction. Whenever it seemed Democrats were poised to turn the race into a true tossup, a new problem has emerged for them: the strong break against them in the polls for their incumbent Mark Udall in Colorado earlier this month, for example. But just the same, whenever Republicans seemed poised to have a clearer advantage, they’ve encountered a roadblock. The Kansas Senate race became highly competitive in early September, and more recently Georgia, which previously leaned toward the GOP, has become a 50/50 proposition.
The large batch of polls released on Sunday morning — including Marist College polls in six states and YouGov polls of every Senate race — followed a similar pattern. Each party had its share of good and bad news.
Republicans can point toward each pollster’s survey of South Dakota, which had their candidate Mike Rounds well ahead. Polls there in early October suggested the GOP could have another Kansas-type problem on its hands with a potentially competitive three-way race between Rounds, the independent Larry Pressler and the Democrat Rick Weiland. Rounds, however, is a 96 percent favorite on the basis of the new polling.
The polls also brought good news for Republicans in Kansas. YouGov had their incumbent Pat Roberts up by 4 points against the independent Greg Orman, while the Marist poll had Orman up one — but down from a 10-point advantage in their late September survey. That election is now a tossup.
Democrats can take some encouragement from the polls of Colorado. YouGov had Udall one point ahead (although down from a 3-point lead in its previous poll) while Marist had the Republican Cory Gardner up by one point instead. Those aren’t great results for Udall, but they’re better than most of the recent nonpartisan surveys of the state, which had shown leads of as large as seven points for Gardner. Republicans remain favored in Colorado, but Udall’s chances of keeping his seat are improved slightly.
Another endangered Democratic incumbent, Mark Begich of Alaska, has also seen marginal improvement in his position. YouGov had Begich trailing by four points against the Republican Dan Sullivan — better than the 6-point deficit they had for him previously. The bigger news for Begich was a Hellenthal & Associates poll released on Friday, which showed Begich with an unlikely seeming 10-point lead over Sullivan. That poll is almost certainly an outlier but it speaks to the uncertainty in the contest; Alaska is notoriously a challenging state to poll.
That the overall forecast has been stable doesn’t mean nobody has the advantage: Republicans continue to have the stronger hand to play. If they win each Democrat-held state where the FiveThirtyEight forecast has them as at least two-to-one favorites — Iowa (where Republican chances are 68 percent), Alaska (72 percent), Colorado (77 percent), Louisiana (77 percent), Arkansas (86 percent), South Dakota (96 percent), West Virginia (99 percent) and Montana (99.7 percent) — they’ll take the Senate, even if they lose the tossups in Georgia and Kansas (and fail to convert North Carolina, where their chances have improved).
But Democrats have a good drawing hand: their chances are equivalent to completing a flush with two cards to come in Texas hold ‘em. Perhaps Begich’s ground game will save him in Alaska. Perhaps Mary Landrieu will survive in a runoff in Louisiana. The Democrat Alison Lundergan Grimes also remains close to Republican incumbent Mitch McConnell in some polls of Kentucky.
We wouldn’t bet on any precise one of those outcomes occurring. But the Democrats have a lot of lottery tickets, and there’s a good chance that at least one draw will come through for them. In only 30 percent of our model’s simulations on Sunday morning did Republicans sweep each of Alaska, Arkansas, Colorado, Iowa, Kentucky, Louisiana, Montana, South Dakota and West Virginia. That means Republicans often will have to win one or more of Georgia, Kansas, New Hampshire and North Carolina if they want to take over the Senate.
Part of the reason for the uncertainty is that many of the important races are associated with circumstances that produce larger polling errors. Louisiana will almost certainly have a runoff and Georgia is more likely than not to have one, which means there’s quite a lot of campaigning left in each state. The polling has been sparse in Arkansas and both sparse and inconsistent in Alaska. Kansas, and to a lesser extent Kentucky, feature a divergence between the polling and the Republican-leaning “fundamentals” of each state.
If we were seeing the same set of results in a presidential election — where there’s both more abundant polling and more accurate polling — Republicans would be much clearer favorites: perhaps somewhere in the range of 75 to 80 percent to take the Senate, about where Obama was in his reelection bid at this point in 2012. Republicans have a great opportunity — most of the key Senate elections are being held in very red states — and are more likely than not to convert on it. But it’s been the same story every day. Republicans have the edge, but they haven’t been able to put Democrats away.