Could the polls be wrong? Why some analysts are urging caution

Follow along to our live coverage to the 2026 US midterm elections here. 

The polls leading into the US midterm elections are increasingly clear: Democrats are on track for a significant, possibly landslide, victory.

What began as a consensus that Republicans would lose their slim majority in the House of Representatives but likely retain the Senate has become a growing expectation that Democrats will win control of both chambers.

However, some pollsters are nervous about the accuracy of those surveys, and reluctant to predict a “blue wave” will eventuate.

Why? In short, because they are pointing to results that would be unprecedented in some cases, and because they’ve been wrong before.

The most ambitious outcomes currently predicted by the polls are Democrats winning a Senate seat in Texas, turning Ohio, Iowa and Alaska blue, and potentially squeaking home in Kansas – a state that has not sent a Democrat to the US Senate in 94 years.

“The Senate is closer to a tossup than most people think,” GOP pollster Patrick Ruffini at Echelon Insights told The New York Times. “The prospect of another red-state polling error looms – particularly in the Midwest.”

Polls have chronically overrated the Democrats’ chances in some of those states. As Donald Trump loves to remind people, they predicted Hillary Clinton would beat him handily. And in 2024, a shock poll by veteran pollster Ann Selzer had liberals aflutter with the prediction that Kamala Harris would triumph in Iowa. Trump ended up winning by an even larger margin.

‘No matter how much error there is in the polls, I don’t think they will miss by that big a margin.’

Rob Shapiro, a political scientist at Columbia University

But those clangers have generally happened when Trump himself is on the ballot, because of polling companies’ failure to reach rural and working-class voters, and peculiar attributes of the president’s followers.

“Trump voters don’t talk to the pollsters, and that can’t be adjusted for through adjusting demographic characteristics,” says Rob Shapiro, a political scientist at Columbia University specialising in public opinion polling and statistical methods.

“They’re basically personality and psychological profiles that have to be adjusted for, for which we don’t have a measure to adjust.”

Shapiro points out that pollsters have a better track record with midterm elections. They may overestimate the Democrats’ gains, but they are unlikely to be in the wrong direction entirely.

“No matter how much error there is in the polls, I don’t think they will miss by that big a margin,” Shapiro says. “I wouldn’t adjust them by more than three points towards the Republicans. Even if you do that, these elections will still lead to a Democratic Congress.”

Nobody seriously believes Republicans will retain their majority in the House, even though most analysts think massive gerrymandering has reduced the potential Democratic pickup to about 20 or 25 seats.

But broadly, Trump’s approval ratings – in the 30s – are just too low. Henry Olsen of the Ethics and Public Policy Centre monitors how midterm election outcomes track presidential job approval in this piece for The Washington Post.

“The Dems are going to take back the house, there’s no doubt about it,” says Douglas Heye, a Republican strategist and party operative who worked in the Bush administration.

“There’s no realistic thought that [polls] can be wrong when it comes to the House of Representatives or the Dems gaining Senate seats. The question is how many. It’s very hard to predict what’s going to happen in Michigan or Maine or Alaska, and so forth.”

Heye says Republican Michael Whatley will not be able to defend the Senate seat in North Carolina, where incumbent Thom Tillis is retiring. The party long ago moved money and resources out of that contest.

Meanwhile, Georgia is a “shoo-in” for Democratic incumbent Jon Ossoff. “We’re talking about South Carolina. We’re not talking about North Carolina and Georgia. That’s insanity.”

As well as Trump’s terrible approval ratings, Republicans have been wounded in critical Senate contests by poor candidate selection, Heye says. He points to Darline Graham in South Carolina – the inexperienced sister of late Republican senator Lindsey Graham – and Texas attorney-general Ken Paxton, who carries enormous baggage.

Shapiro believes a Senate tied at 50-50, with Vice President JD Vance holding the deciding vote, is still a live outcome.

But some pollsters who have urged caution up to this point are willing to make bolder calls with election day now upon us. Nate Silver, based on a rigorous analysis of the final polls from Siena and The New York Times, believes Democrats will win the Senate, “perhaps soundly”. He says the party’s chances of controlling that chamber are somewhere around 80 to 90 per cent.

Read more on US politics: