Republican Pollster Frank Luntz Calls Trump's Midterm Message 'Dangerous' as Approval Sits at 38.3%

Since Trump told a crowd at the Republican National Committee's midterm convention in Dallas earlier this month to "pretend that I'm on the ballot," Republican strategists have started saying out loud what some had only whispered: that message might backfire in November.

Frank Luntz, the longtime GOP pollster, put it bluntly on CNN This Morning with Audie Cornish. He called Trump's strategy "dangerous."

"There's always been a historical advantage to the party out of power in the midterms, particularly in the first midterms, and we see that now, that there's an energy challenge for GOP voters, and that's what the president is trying to address," Luntz said. He noted the pattern isn't new: Trump's base turns out when Trump is literally on the ballot, as in 2016 and 2024, but stayed home in 2018 and largely sat out 2022 even under Joe Biden.

"His numbers, his support is not as strong today as it has been in the past, so it's a very dangerous strategy. But he needs his people to vote in a place like Texas," Luntz said.

The Numbers Behind the Warning

Luntz isn't speculating in a vacuum. A Decision Desk HQ polling average put Trump's approval at 38.3% and disapproval at 59.8% as of Tuesday morning. That's a gap of more than 21 points.

CNN's own reporting, drawing on CNN/SSRS polling, adds detail to where the erosion is coming from. Only 27% of respondents expressed confidence in Trump's handling of the economy, and a similarly small share said they think economic conditions are good, even as two-thirds of Americans polled called the economy "extremely important" to their vote in November. Mortgage rates crossed 7% last week, according to CNN, adding pressure on top of grocery and housing costs voters already say feel too high.

On Iran, where the conflict is approaching its eighth month, CNN's polling found only 26% approve of Trump's handling of the war, and three-quarters of respondents don't believe he has an exit plan. At the UN General Assembly, Trump told the crowd he was weighing whether to "annihilate" Iran or cut a deal, and predicted the war would end sometime after the midterms, a timeline CNN describes as reassuring for no one.

On artificial intelligence, CNN's polling found 7 in 10 Americans want more government regulation, a bloc square against Trump's public position that he wants AI policy "left exactly where it is," a position he's stated directly on social media.

The Case for Trump's Approach

Luntz's own framing offers the strongest defense of what Trump is doing, even as Luntz calls it risky. Trump's coalition, by Luntz's own account, is not a base that shows up out of habitual civic duty. It shows up when it believes Trump himself has skin in the game. That's why the president is telling voters in states like Texas to act as though he's personally on the ballot, even though he isn't.

There's also a broader point, one CNN's own analysis concedes even while criticizing Trump: presidents don't have to chase polling to justify a decision. Holding a line on immigration enforcement, tariffs, or AI policy against public opinion can reflect genuine conviction rather than political miscalculation, and voters have rewarded conviction before. Whether that applies here is a matter of interpretation, not fact, and reasonable people can disagree on which of Trump's positions reflect principle versus a misread of the electorate.

Republicans are trying to defend House and Senate majorities in an environment where the party out of power historically gains ground, a dynamic Luntz says is already visible and one that has applied in every first midterm going back decades regardless of who holds the White House.

What's Unresolved

The Hill reported it reached out to the White House for comment on Luntz's remarks and had not received a response as of publication. Trump's economic messaging, including his claim that he has driven food prices down, is disputed by CNN's reporting, though CNN did not cite a specific government inflation dataset in that segment to quantify the gap between Trump's claim and measured price levels.

The open question is turnout math, not sentiment. Luntz's data shows Trump loyalists come out for Trump, not necessarily for down-ballot Republicans. With five weeks until Election Day, whether "pretend I'm on the ballot" moves enough of that bloc to the polls, against a 38.3% approval rating, is something no poll can answer in advance.