Republicans are substantial favorites to take over the U.S. House of Representatives following this November’s midterm elections, but the U.S. Senate is much more competitive, according to FiveThirtyEight’s 2022 midterm election forecast, which launched today. Democrats are also favored to hang on to the governorships in a trio of swing states in the Rust Belt — Pennsylvania, Wisconsin and Michigan — although they are significant underdogs to win high-profile gubernatorial races in Georgia and Texas against Republican incumbents.
Moreover, in striking down Roe and other popular laws like restrictions against the concealed carry of firearms, the Supreme Court has in some ways undermined one of the traditional reasons that the president’s party tends to lose seats at the midterms. Typically, voters like some degree of balance: They do not want one party to have unfettered control of all levers of government. But the Supreme Court, with its 6-3 conservative majority, is a reminder of how much power Republicans have even if they don’t control the White House.
So, this is not a typical, low-stakes midterm election. On the contrary, there are strong forces tugging at each side of the rope, some of which are potentially of existential importance.
But Democrats’ majorities in both chambers of Congress are narrow, the historical precedent toward the president’s party losing seats is strong, and polls so far — such as the generic congressional ballot, which asks voters which party they would support in an election — suggest that voters slightly prefer Republican control of Congress.
Or at least that’s the story in the House, where there are dozens of competitive races and candidates are relatively anonymous. There, big-picture factors tend to prevail. An unusually weak Republican candidate in one district might be counteracted by a strong one in another, for example.
So let’s briefly run through the model’s forecast for House, Senate and gubernatorial races. Then I’ll describe some changes to the model since 2020 — which are modest this year but reflect how congressional races are changing in an increasingly polarized political environment.
Republicans have an 87 percent chance of taking over the House, according to the Deluxe version of our model. That’s far from certain, but Democrats are fighting the odds: Their 13 percent chances are equivalent to tossing a coin and having it come up tails three times in a row.
That’s not to say House control will be a matter of luck, exactly. A change in the political environment could have ripple effects. For instance, sometimes one party wins nearly all the toss-up races, as Republicans did in 2020. However, even if Democrats were to win all 13 races that our model currently designates as “toss-ups” (meaning that no party has more than a 60 percent chance of winning), plus hold on to all the seats in which they’re favored, they would still wind up with only 208 seats, 10 short of the number they need for a majority.
Instead, Democrats will also have to win some seats where Republican candidates are currently favored, and that requires the national political environment in November to be more favorable for Democrats than our model is currently expecting.
On the one hand, the task isn’t that daunting for Democrats. Our model calculates that Democrats would be favored to keep the House if they win the House popular vote — or lose it by less than 0.7 percentage points — something that Democrats did in both 2018 and 2020.
Moreover, Democrats are down by only about 2 points in our current average of generic-ballot polls. Given the inherent error in polling, and how much time there is between now and November, it isn’t hard to turn a 2-point deficit in the polls into a 1-point win.
However, in important ways, that 2-point deficit understates the degree of trouble that Democrats are in. One reason is because many of those polls are conducted among registered voters rather than likely voters, and the electorate that turns out in November is likely to be more Republican than the broader universe of all registered voters. Historically, the patterns in midterm elections are that: 1) Republicans turn out more than Democrats, and 2) voters for whichever party doesn’t control the presidency are more enthusiastic and turn out more. In 2018, those factors canceled one another out. Democrats, not controlling the presidency, were the more enthusiastic party, helping to neutralize the Republicans’ historical turnout advantage. This year, though, they both work in the favor of Republicans.
Thus, the model adjusts those registered-voter polls based on its estimate of what likely-voter polls would show, and when it does that, the Republicans’ generic-ballot lead is really more like 4 points than 2 points. I should note that this adjustment is not rigid in the model. Although the model uses historical turnout patterns as its baseline assumption, it will override that based on polls. In other words, if polls come out showing Democrats holding their own among likely voters — such as because of increased Democratic enthusiasm in the wake of Roe being overturned — the model will adjust to reflect that. Put another way, a very strong turnout would give Democrats a fighting chance of keeping the House.
But also, the generic ballot isn’t the only input that the model considers, and some of the other factors look worse for Democrats than the generic ballot does. Based on the historical tendency for the president’s party to lose seats in the midterms and Biden’s poor approval rating, for instance, the situation is more likely to get worse for Democrats than better. The model also evaluates factors such as polling and fundraising data in individual races.
Overall, the Deluxe forecast expects Democrats to eventually lose the popular vote for the House by closer to 6 points, about the margin that they lost it by in 2014. And it expects Republicans to wind up with 237 seats in an average outcome, a gain of 24 seats from the 213 they had at the start of the current Congress.2
As I mentioned, this analysis is based on the Deluxe version of our model, which accounts for polling, “fundamentals” — or factors such as fundraising and incumbency — and expert race ratings such as those put out by the Cook Political Report. The Classic version of our model, which leaves out the expert ratings — sacrificing the additional accuracy they add but sticking to purely quantitative factors — tells a similar story, with Democrats also having a 12 percent chance of keeping the House. The Lite version of our model, meanwhile, which tries to forecast as much as it can based on polls alone, does paint a more optimistic picture for Democrats, giving them a 22 percent chance of keeping the House. But that version leaves out a lot of useful information, especially given that there isn’t much polling in a number of competitive House races.
Democratic hopes of keeping the Senate are much more viable, however. Part of this, as I mentioned, is because they appear to have stronger candidates in a handful of key races. Pennsylvania, for instance — which is an open seat after the retirement of Republican Sen. Pat Toomey — is ordinarily the sort of seat that you’d expect Republicans to win since Pennsylvania is a purple state in a Republican year. However, the Democratic candidate, Lt. Gov. John Fetterman, is ahead of Republican Mehmet Oz, the doctor and TV personality, in every poll conducted so far. The model, though, is trained to be a bit skeptical given the fundamentals of the race, so it hedges against those polls and, at this point, has determined that Pennsylvania is best thought of as a toss-up. Still, that means Democrats have roughly a 50-50 chance of gaining a GOP-held Senate seat, offsetting potential losses elsewhere.
Indeed, our forecast sees the overall Senate landscape to be about as competitive as it gets. The Deluxe forecast literally has Senate control as a 50-50 tossup. The Classic and Lite forecasts show Democrats as very slight favorites to keep the Senate, meanwhile, with a 59 and a 62 percent chance, respectively.
Part of this is because Senate terms last for six years, and so most of these seats were last contested in 2016,3 a mediocre year for Democrats in which they lost the popular vote for the House and also lost Senate races in swing states such as Pennsylvania, Wisconsin and Arizona. Of the 35 Senate seats up for grabs in November, 21 are currently held by Republicans. True, most of these are not competitive, but in addition to their chances to gain a GOP-held seat in Pennsylvania, Democrats also have credible chances in Wisconsin and North Carolina (and outside chances in Ohio and Florida, although those are a stretch given how GOP-leaning both states have become).
Republicans don’t have any surefire pickups, meanwhile. Our model regards their best chances as being in Georgia, but that race is rated as a toss-up. And the races in Arizona and New Hampshire merely lean toward the Democratic incumbent, meaning they are still highly plausible GOP pickup opportunities.
Still, the picture isn’t as bad as you might expect for Democrats. If the political environment really deteriorates for them, they’ll be in trouble, lose most of the competitive races and even blue states like Colorado could come into play. But if things are merely pretty bad for Democrats instead of catastrophic, the outcome of the Senate will remain uncertain enough that stronger candidates could make the difference for them.
The Governors’ Mansions
It’s hard to talk about gubernatorial races on a systematic basis since there’s no particular prize for winning a majority of governorships. But, for the record, our model does run these calculations, and the Deluxe version estimates that there’s an 83 percent chance that Republicans end up with a majority of governorships following this November’s elections, compared with a 7 percent chance for Democrats. (There is an 10 percent chance that neither party has a majority.) However, a lot of these governorships are in smaller, lower-population states, and the model thinks there’s a 73 percent chance that the majority of the U.S. population will reside in states run by Democratic governors.
Overall, though, gubernatorial contests take the theme from the Senate a step further: Individual candidates can matter a lot. Indeed, partisanship matters less in gubernatorial races than in races for Congress, even if it matters more than it once did. Consider, for instance, that there are currently Democratic governors in Kansas and Louisiana and Republican ones in Massachusetts and Maryland, although several of those seats could flip parties this year.
However, incumbency is a powerful force in gubernatorial races. For instance, even though Michigan is a slightly red-leaning state, its incumbent Democratic governor, Gretchen Whitmer, is a clear favorite against a Republican field marred by fraudulent attempts to access the ballot and the arrest of a leading candidate for his participation in the Jan. 6 insurrection. Meanwhile, Wisconsin’s Tony Evers, also a Democratic incumbent, is a favorite against a Republican field where the most likely nominee is Rebecca Kleefisch, the former lieutenant governor. This is the sort of race where abortion could matter: Technically, Wisconsin’s 173-year-old abortion ban — which outlaws all abortions, except in cases to save the life of the mother — is now in effect following the Dobbs decision, although the Democratic Attorney General Josh Kaul has said he won’t enforce the ban.4 But Kleefisch has said she opposes abortion even in cases of rape and incest.
Perhaps the most important gubernatorial race is in Pennsylvania, an open-seat race as the Democratic incumbent governor, Tom Wolf, is term-limited. There, the Republican nominee, Doug Mastriano, was present outside the Capitol during the Jan. 6 insurrection and worked to overturn Biden’s win in Pennsylvania, potentially yielding a constitutional crisis if he’s governor in Pennsylvania and the election outcome is close again there in 2024. But Mastriano is an underdog against Democrat Josh Shapiro, the Pennsylvania attorney general.
Overall, we’re happy with our congressional and gubernatorial forecasts, which last underwent a major revision before the 2018 elections. They performed very well in 2018 and fairly well in 2020 (despite a challenging year for the polls in 2020; it helped that our model also considers a number of other factors in addition to the polling). Therefore, the overall methodology is largely the same. However, after assessing the performance of the models, we did make a few changes around the margins:
In evaluating fundraising for congressional candidates, the model now places more emphasis on contributions received within the candidate’s state. Fundraising is a highly nationalized activity these days; Democrats in California and New York regularly contribute to Senate campaigns like that of Democrat Jaime Harrison of South Carolina, who raised a record amount of money in 2020but nonetheless lost to Republican Sen. Lindsey Graham by 10 percentage points. As such, our analysis finds that a dollar received from within a candidate’s state is about five times as valuable as a dollar received outside of it in predicting the eventual election outcome.5 That’s why the model now applies that ratio, multiplying any funds raised from voters within the candidate’s state by 5, when assessing which candidate has the fundraising edge.
The model now puts more emphasis on basic partisanship as opposed to other fundamentals. The fundamentals component of our model, used in the Classic and Deluxe versions, evaluates a series of variables such as fundraising, incumbency, candidate experience, congressional voting records and even scandals to make a forecast. The formula is fairly complicated, so we tested a simple alternative in an era of high partisanship. Specifically, we looked at what would happen if you made a forecast based solely on the partisan lean of a state or district, plus generic-ballot results. For instance, if the generic ballot favors Republicans by 4 percentage points, and a state leans Republican by 3 points, you might expect the Republican to win there by 7 points. In testing this on 2018 and 2020 races, we found that the more complicated version of our fundamentals formula does outperform this simple alternative. But it’s pretty close, and a blend of the complicated and simple formulas does better than either one taken alone.6 What this means is that state partisanship and the generic ballot will have more influence on our model going forward and the other fundamentals factors less, although they certainly matter some.
We assume errors are more correlated from race to race. In presidential races, forecast and polling errors are highly correlated from state to state. It wasn’t a surprise that Trump won Wisconsin and Pennsylvania in 2016 given that he also won Michigan, for example, since the states are fairly similar demographically. (Accounting for this properly was a big reason that FiveThirtyEight’s model gave Trump a better chance than others in 2016.) This also holds in congressional races, though to a considerably lesser degree since there are different candidates on the ballot in each state. A scandal affecting the Republican candidate in Georgia won’t necessarily have any impact in Arizona, for instance. However, our analysis suggests that polling errors are becoming more correlated in congressional races, too. Certainly, 2020 was a prominent example given that Republicans won nearly every toss-up race for the House and that polls for Congress had a strong overall bias toward Democrats. Therefore, we’ve increased the degree to which race-by-race errors are assumed to be correlated in the model.7
We’ve removed a series of changes in the 2020 model that were specific to the COVID-19 pandemic. Yes, I know the pandemic isn’t over yet, but voting will take place under far more normal conditions than in 2020.
Finally, a couple pieces of housekeeping. A number of states haven’t held their primaries yet, so in those cases, we guess at the most likely nominees based on polling, fundraising and other factors. These “presumed nominees” are designated with an asterisk in the interactive. If you see anything egregiously wrong — such as a candidate listed as a presumptive nominee when they’ve dropped out of the race — please drop us a line.
We’re also still thinking about how best to handle Alaska, which has a new system in place this year in which the top-four finishers in the primary advance to the general election regardless of political party, and then the general election outcome is determined by ranked-choice voting (or, as some call it, an instant runoff) if no candidate receives a majority. This is not entirely dissimilar to the way elections are conducted in Louisiana, in which all candidates from all parties appear on the ballot in a blanket nonpartisan primary in November, and then there’s a runoff later between the top-two candidates if no candidate gets 50 percent of the vote. In fact, we’re currently taking a bit of a shortcut by using the Louisiana code for Alaska, essentially treating the instant runoff as though it’s an actual runoff where voters go to the polls again.
We may revisit this assumption later, but it does avoid one potential pitfall. In Alaska’s House, Senate and gubernatorial races, it’s fairly likely that we’ll end up with one Democratic candidate but two or three Republican candidates following the Aug. 16 primaries. If the Republican vote is divided two or three ways, it may well be that the Democrat initially receives the plurality of the vote. However, this lead is unlikely to survive the instant-runoff process assuming voters for one Republican rank the other Republicans ahead of any Democrat. The process we use for Louisiana assumes that votes mostly tend to stay within the same party in the event of a runoff, and this same assumption is in place in Alaska. Thus, we have Republicans as fairly heavy favorites in the Alaska races, although the new system introduces some uncertainty.
The forecast itself will update continuously whenever new polls or other information are added to our database. We’ll also publish a written update to the forecast once per week or so, usually on Fridays, to review new data and other changes in the landscape, before upping the frequency as the election draws closer. We hope you’ll regularly visit FiveThirtyEight as part of your midterms rotation.