The United States goes to the polls tonight in the year’s largest election and one of its most consequential. Donald Trump is not personally on the ballot, but the results and their aftermath will help determine the character of the remaining two years of his term. Voters will choose the House of Representatives, more than a third of the Senate, governors in 36 states, almost all of the state legislatures and a multitude of other state and local offices.
The House of Representatives need not detain us for long. It has 435 members, elected for two-year terms from single-member districts, first-past-the-post. In 2024 the Republicans won a narrow majority, 220-215, off 51.3% of the two-party vote. A uniform swing of 0.5% from that result would have given the Democrats the three additional seats required; that figure has since increased slightly due to gerrymandering from both sides. But since the polls are showing a swing of the order of 6%, there is no realistic doubt that the Democrats will regain control of the House.
The Senate is a quite different beast. Its members, two per state, serve six-year terms, with a third of them up for election each time. To the 33 seats being contested this year must be added two by-elections, in Florida and Ohio, for a total of 35: 22 Republican and 13 Democrat. The Republicans currently have a 53-47 majority in the Senate, so to win control the Democrats need to pick up (net) four seats, since a tie would be resolved against them by the casting vote of the vice-president.
The most marginal Republican seat in the Senate is North Carolina, on a margin of 0.9%. (Note that I express all margins in the Australian fashion, as the swing required to change hands, factoring out any third-party vote – figures from American sources will all look a bit different.) You can pocket that one straight away; the Democrats are about ten points ahead in the polls and the Republicans have stopped trying.
That means the Democrats need another three, assuming they hold all their own seats. As it happens, one Democrat-held seat is in some danger: Michigan, also on a margin of 0.9%, where the incumbent is retiring and the party has chosen left-winger Abdul Al-Sayed to replace him. Polls have generally put Al-Sayed in the lead, but it’s too close for comfort: if he should lose, the target would rise to four more Republican seats.
A uniform swing of 7% to the Democrats would bring them, in addition to North Carolina, another nine seats: in order, Ohio, Iowa, Maine, Texas, Montana, Mississippi, South Carolina, Kansas and Alaska. Swings are never uniform, especially in the Senate, where personal votes (whether positive or negative) are typically substantial. Nonetheless, party-line voting has been on the increase for years now, so in fact the pendulum isn’t a bad guide.
Only one of those nine, Mississippi (5.1%), is clearly out of reach. Its place on the list could be taken by Nebraska, which at 17.1% is much safer on paper but has moved into range after the Democrats withdrew in favor of a sympathetic independent, Dan Osborn. He is now running neck-and-neck in the polls with Republican incumbent Pete Ricketts.
The best Democrat prospects at this stage, perhaps roughly in order, look to be Ohio (3.1%), Texas (4.9%), Maine (4.6%), Kansas (6.0%) and Alaska (6.7%). Nate Silver’s polling averages currently show Democrat leads in all of them of between two and a half and four points. Three gains out of those five (if they hold Michigan; four if they don’t) will deliver Senate control. It’s no certainty, but it now looks much more likely than not.
It’s also not just about winning control. Because senators often go they own way on particular issues, every seat matters: a Senate with a 55-45 majority – say, if Osborn wins Nebraska and the Democrats reel in Iowa (3.4%) as well as the above targets – would be very different from one at 51-49. Particularly if it’s coupled with big wins in the House and in some major state races, a Democrat landslide in the Senate would induce panic in the Republican Party and possibly some high-level desertions from the Trump camp.
Two other points are worth noting. Firstly, the discussion has assumed that the elections happen in more or less the normal way, the actual results are tabulated and certified and the winners are allowed to take office. In Trump’s America those are no longer sure assumptions; the administration has been working on a number of lines to subvert the process. But none of them have won favor in the courts, and if the results are as clear as the polls are saying then they will be very hard to undo.
Secondly, it’s important to remember that the predictive power of midterm elections isn’t very good. The Republicans won big majorities in midterm elections in 1994 but still lost the presidential election two years later; then they repeated the pattern in 2010-12. The Democrats swept the House in 2018, but could only convert that into a narrow win for the presidency in 2020.
The midterms are set to deliver a setback to Trumpism, but it will not be a knockout blow. That, if it comes at all, will have to wait until 2028.
