The answer to at least one of the questions everyone has been asking will be made clear by Tuesday’s midterm results: Just how accurate were the polls?
Campaigns, donors and outside groups across the country made decisions about how to spend millions based on polls. Whether it was worth it — whether they even needed to — will soon be apparent. But because of polls, the battlegrounds that both parties are focused on are spread across far more states than at the start of the cycle.
This year’s polls have heavily favored Democratic candidates. Because of past missteps across the polling industry, polling firms have been approaching the significant leads with a lot of caution and skepticism, said Adam Carlson, a Democratic pollster. They’re also functioning in an environment with less data: there are fewer polls, and response rates have dropped.
“We’re relying a lot more on vibes,” Carlson told The Washington Sun.
Trending
He added that some pollsters are so worried about being wildly off with this year’s polls — like how Democrats were inaccurately favored in 2016 — that they’re adjusting their data analysis to prevent too extreme of a tilt toward Democrats in competitive races.
“There’s a bigger reputational risk of being wrong again, underestimating Republicans or overestimating Democrats,” Carlson said.
On Monday, House Speaker Mike Johnson appeared on political commentator Scott Jennings’ radio show, where he said this cycle’s polls inaccurately favor Democrats, as they have in the past.
“The polling is wrong, OK,” Johnson said. “Everybody understands the polls have been off,” adding that polling “underrepresents Republicans.”
He said pollsters “blame Republicans” for inaccurate results.
“They say, ‘Conservatives don’t answer polls anymore,’” Johnson continued. “But they’ve learned not to trust — they don’t know who that person is on the other side of the phone, and they know it’s a psyop. They’re trying to create this idea of a blue wave.”
One of the issues facing pollsters is the challenge of reaching certain voters, especially those who only show up to vote in elections in which President Donald Trump is on the ballot. The Washington Post found that most Republican voters in Kansas didn’t want to speak with a pollster, while Democrats were more willing.
“They don’t like the media, and they don’t like pollsters,” said Charles Franklin, the director of Marquette Law School Poll. “That’s a recipe that makes it especially hard to reach the very people that you most want to talk to.”
No one has to reach very far back for examples of just how wrong the polls can be. The 2026 primaries had plenty of examples: Polling found Francesca Hong, the progressive candidate in Wisconsin’s Democratic gubernatorial primary, ahead by as much as 20 percentage points over the course of her campaign.
David Crowley beat her in the primary by less than 1 percentage point, wildly overperforming expectations — which were set by the polls.
The number of public polls conducted nationwide has steadily declined. According to a FiftyPlusOne analysis conducted in mid-October, there was only polling data in October for 54% of Senate races in 2026, compared to 80% of Senate races in 2018. One reason for fewer polls is because of the increasing cost to use reliable methodologies, Republican pollster Chris Wilson explained.
“Campaigns and donors and outside groups make real decisions based on those numbers,” Wilson said. “A bad poll … can move millions of dollars into a race that doesn’t need it, or away from a race that does.”
Surprisingly narrow margins in this year’s polls have shifted millions toward races previously seen as safe Republican wins in the Senate, such as Mississippi and Kansas. The Democrat-aligned Senate Majority PAC dropped at least $5 million to back Adam Hamilton in Kansas and more than $1 million for Scott Colom in Mississippi.
Similarly, Republicans have stepped back from spending big in places like North Carolina and Georgia, where polls show comfortable leads for Democrats. The Senate Leadership Fund, a major Republican super PAC, originally planned to spend $18 million in North Carolina in September but entered October having spent $32 million to back Senate candidate Michael Whatley — which is when Semafor reported the super PAC planned to pull back spending. Whatley’s first ad in nearly four weeks aired only a week before Election Day.
Two firms have admitted to fabricating polling data this year, including Wedgewood Polls, which admitted online to faking polls in the Michigan Senate primary race and the Wisconsin gubernatorial race.
Michigan Democrat Mallory McMorrow, who spoke with The Washington Sun after Wedgewood revealed its fraudulent polls, said one of the reasons she suspended her primary campaign for Senate was because Wedgewood showed her with only 9% of the vote.
McMorrow said polls are closely watched by donors and national political committees. Shortly after the fake June poll, the state lawmaker said she had major donors recommending she drop out of the race, which started the “death spiral” for her campaign.
“I think the biggest thing that I want to remind people now is we shouldn’t trust any polls,” McMorrow said. “I hope this teaches us a lesson.”
Prediction market companies like Kalshi and Polymarket want to capitalize on the polling industry’s shortcomings. Rather than measuring actual voter preference data, these platforms allow users to wager money on the odds of an election result — but political operatives have warned campaigns not to build a strategy based on prediction market odds.
Pollsters should be thinking ahead to 2028, Wilson told The Sun, and figuring out ways to get the data right in two years.
“The lesson of 2026 isn’t that public opinion is unmeasurable,” Wilson said. “It’s that measuring accurately matters more than ever before.”
Author
