An attendee wears a hat in the colors of the U.S. flag during U.S. President Donald Trump's visit to Verst Logistics in Hebron, Kentucky, U.S., March 11, 2026. REUTERS Kevin Lamarque
Data analyst G. Elliot Morris fears the GOP may be a dead-party-walking as they head into the mid-terms — and polling analytics are setting the GOP up for terrible disappointment.
“The polls are [possibly] understating support For Democrats, not Republicans, as you might have heard,” said Morris, speaking on the data generating process with The Downballot Podcast host David Nir.
“If you pay any attention to politics and the polls over the past couple of cycles, then you will probably be triggered by large apparent misfires in public opinion surveys in the direction of the Democrats over the last four election cycles, basically every cycle since 2016, except for the midterms,” said Morris, referencing President Donald Trump’s unexpected wins in 2024 and 2016.
However, “once you look under the hood,” said Morris, it appears pollsters are accidentally creating the potential for bias in the other direction and setting Republicans up for even more disappointment than they’re already expecting. “Pollsters now are increasingly weighting or adjusting their samples by political variables, in particular, the results of the last election.”
The problem with that, said Morris, is that survey respondents don’t always admit they voted for Trump when they did.
“Maybe you don’t want to say you voted for that guy. Maybe you were like a soft Trump voter in 2024 and a pollster is going to ask you, ‘hey, who did you vote for last time? And you were under no obligation to the pollster's statistical methodology to tell them the truth. Maybe you say, ‘oh, I didn't vote.’”
And according to follow-up surveys, respondents appeared to be changing their claims of earlier support, not wanting to admit they went Trump the last time around for whatever reason.
“Those sort of soft members of the Trump coalition now won't admit that they voted for him. So that's not necessarily a problem unless you are equalizing your sample by past vote,” said Morris. “… [T]his level of recalled vote error could be increasing support for Republicans in the generic ballot by a little less than one percentage point in terms of vote margin. So, in other words, if there wasn't this level of hypothetical bias we're exploring, maybe Democrats are winning the generic ballot by like eight points instead of seven, which is what our average shows today.”
This, said Morris, is going to be a problem for Republicans in very close state and district races, especially if they were expecting that single digit to get them over the finish line.
