Frank Luntz went on CNN this fall and admitted something the network's audience probably didn't want to hear. The pollster who built his career reading focus groups and tracking voter sentiment told anchor Audie Cornish that he "got 2016 wrong" because he "believed in the exit polls" — and that the same blind spot is still baked into the data heading into the 2026 midterms.
"We know that they don't respond to phone calls," Luntz said of Trump voters.
That's not a minor technical footnote. Current polling projects a 53-47 Democratic Senate majority and a 235-226 Democratic House edge. A three-point shift — exactly the kind of margin pollsters have missed in the last three cycles — flips the Senate to 54-46 Republican. And Luntz, who teaches at West Point and has spent decades inside the machinery of opinion research, is saying out loud that the instruments measuring this electorate are still broken.
The numbers tell a specific story. Luntz told CNN he sees "a turnout differential that advantages the Democrats by a couple of points." But he immediately undercut that projection with a fact the models keep failing to price in: "When he is on the ballot, his people come out. They participate. They did in the on-year elections, 2016 and 2024. And they did not, in 2018."
Trump isn't on the ballot in 2026. But Luntz pointed to something pollsters haven't accounted for — Trump's gravitational pull on races where his name doesn't appear. "Donald Trump does have more influence on more races than any other Republican, even than Ronald Reagan," Luntz said. "I've never seen a Republican as engaged and involved as this guy."
He named Texas specifically. "He needs his people to vote in a place like Texas. If the Trump voters participate, the Republican nominee in Texas is elected." Iowa, Alaska, and Ohio got the same treatment — states where the polling looks manageable for Democrats right now but where the Trump voter question turns comfortable margins into coin flips.
The objection from the polling industry is that they've corrected for this. They adjusted their likely-voter screens after 2016. They weighted for education after 2020. They rebuilt their models after 2024. And yet here's Frank Luntz — not a MAGA populist, not a Republican operative, but the establishment's own go-to pollster — saying the same voters are still invisible. "Republicans are less likely to participate. They're less likely to believe the polls are happening," he told CNN. "But I'm most interested in turnout."
That last line is the one Democrats should lose sleep over. Turnout is where every polling miss since 2016 has lived. Not in what voters said they'd do, but in which voters showed up that the models didn't expect. Male union voters broke for Trump at historic levels in 2024. Voters under 30 gave him more support than any recent Republican. Latino support shifted. These aren't projections — they're results. "The votes are the votes," Luntz said. "The results are the results."
Earlier this year, after Wisconsin primary results blew up the projections again, Luntz called for a "complete shutdown of primary polling in the Midwestern United States until we can figure out what the hell is going on." That wasn't hyperbole from a cable news segment. That was a pollster looking at his own industry's data and concluding the product is defective.
The pattern is worth stating plainly. In 2016, the polls missed Trump voters and he won. In 2020, the polls overstated Biden's margins in every battleground. In 2024, the polls underestimated Trump again, and he won again. Now in 2026, the polls show Democrats ahead — and the most prominent pollster in the country is on CNN telling everyone the same gap still exists.
"But if he turns out his people, who still like him, still believe in the Trump message and the Trump agenda, we can have an upset," Luntz said.
Three cycles. Same blind spot. Same warning. The polls have a 53-47 Democratic Senate on the board. Luntz is telling you the scoreboard is broken.
