Polls saw the Democratic wave. Here’s why they missed the tsunami

The polls had a decent night Tuesday in predicting Democratic wins but fell short of projecting the blue tsunami that ultimately swept larger-than-expected Democratic majorities into power in the House and Senate.

Democrats exceeded even their own expectations on election night, winning Senate races in states such as Kansas and South Carolina, along with contested House races across the country.

Polling widely suggested ahead of Election Day that a blue wave could be coming, as Democrats led or were at least competitive in all battlegrounds they were focused on flipping.

But the polls underestimated the extent of just how much Democrats would dominate the midterms.

That’s a far cry from what has felt like the norm since President Trump’s first election in 2016.

“What we did get is a 2020-level polling miss, thereabouts, but the inverse where the Democrats were underestimated in polling,” said Chaz Nuttycombe, the executive director of State Navigate.

Nuttycombe pointed to a couple factors to explain what happened. Chiefly, he pointed to how pollsters weigh their results.

Instead of simply polling respondents and presenting those results as their finding, pollsters use weighting to project what they expect the electorate to look like in an election, based on certain key demographics like education, race and party identification.

Ahead of the election, State Navigate put together a report expressing that polls might be underestimating Democrats. Nuttycombe said many pollsters weighted their results based on expecting an electorate similar to 2022, when former President Biden was in office.

He said looking at the 2018 electorate, the first midterm year for Trump, was a better parallel.

Nuttycombe also pointed to an issue with total recall bias, when results are impacted by flaws in how respondents reported they voted in a past election.

“You had a whole bunch of people who were lying, who are definitely Trump voters who said that they weren’t,” he said.

When voters either choose to lie to a pollster about who they voted for, or potentially don’t remember correctly, it can skew results when it comes to weighting a poll. Fewer people who reported voting for Trump in 2024 may cause a pollster to believe those voters are being undercounted.

Nuttycombe dismissed the idea of people being ashamed of their vote but instead pointed to voter frustration with the president, illustrated by Trump’s low approval rating.

“This is them being frustrated and them saying in their vote choice, ‘I did not vote for this. I’m p—ed off, and I’m going to go vote,’” he said. “Mostly, these people are voting for Democrats.”

In terms of the bottom-line results on who won, the polls certainly pointed in the right direction.

The final average from The Hill’s election partner Decision Desk HQ, which aggregates surveys from across nonpartisan, left-leaning and right-leaning professional pollsters, showed Democratic candidates ahead in most of the key Senate races. They led by varying margins in Georgia, Maine, Michigan, Minnesota, New Hampshire, North Carolina, Ohio and Texas — all of which the Democratic candidate won.

Even Democrat Annie Andrews in South Carolina led narrowly by 0.1 point in the average over Sen. Darline Graham (R) based on the limited public polling available.

The average also showed Democrat Mary Peltola leading by 1.4 points. She is locked in a close race with Sen. Dan Sullivan (R) as the final ballots in Alaska’s Senate race are tallied.

But margins were underestimated almost across the board, with significant implications for gubernatorial and downballot races.

Some of the largest overperformers of the night included Sen. Jon Ossoff (D) in Georgia, who won by 18 points despite a 10-point lead in the average, and James Talarico (D) in Texas, who led Republican opponent Ken Paxton by nearly 12 points despite holding just a 3.5-point lead in the average which is a gap within the margin of error for most polls.

Democrats also pulled off impressive wins in gubernatorial races.

In arguably the most notable flip of the evening, Texas Gov. Greg Abbott (R) lost his reelection bid to state Rep. Gina Hinojosa (D), who led by 7 points. The polling average showed Abbott ahead by 1.3 points, well within the margin of error but still didn’t indicate Hinojosa could win so comfortably.

In another example, Arizona Gov. Katie Hobbs (D) led on average in polling by just under 7 points, but she’s currently ahead by 13 points.

Still, pollsters did have some successes from the results, as many outcomes at least signaled most states that were seriously in contention so the public had an idea of where they should pay attention.

Alec Tyson, the senior vice president and lead pollster for Ipsos, argued that polls did well in illustrating “the arc of the election.” There’s more to evaluating the electorate that’s coming than just the horserace polls that often dominate people’s attention as Election Day approaches.

“A lot of things were true for the past decade in politics until they weren’t this year, and one of the things polls did a good job of is documenting the erosion of Trump’s approval,” he said.

Tyson said many observers and pollsters for years have seen Trump as having a low ceiling and a high floor for his approval rating, keeping it in a narrow range that often stayed in the mid-to-low 40s. Polls effectively showed just how much Trump’s approval had dropped over the past year, a major indicator that a clear rebuke of the president was coming.

They also showed that the cost of living was the top issue on voters’ minds as inflation remained high. They showed Democrats taking the lead in the question of who voters trust more to handle the economy — a striking change that hadn’t been observed in more than a decade .

“It’s worth noting that the average of the generic ballot, even going into election night, was saying this has a chance to be a historic night,” Tyson said.

RealClearPolitics showed Democrats’ lead in the generic ballot to be 8.8 points, larger than any midterm cycle since 2006 , another major Democratic wave.

Tyson acknowledged pollsters still have more to learn and will need to adjust based on the new information of this year, just as they have during past cycles when Trump outperformed.

“Direction polls did a great job painting or kind of outlining the story of the election, and even so, we have much to learn,” he said. “And I think we’re wise to assess each election on its own context and see what we can learn and carry forward.”