November 12, 2026
At THP, our final generic ballot had Democrats up 13 points among likely voters, 54 to 41, after earlier readings of 11, 9, and 12 points. Although votes in various House races are still being counted, that 13-point number essentially matches the final vote totals. It would be the largest midterm margin since 1974.
Patting oneself on the back too vigorously can be unseemly, but as Howard Baker said in 1980 while running for president, “He who doth not toot his own trumpet – his trumpet shall not be tooteth.”
In that spirit, I’d point out that two years ago the national tracking at Rasmussen Reports, where I was director of polling for five years, showed Trump winning the popular vote for the entire 2024 cycle, with Rasmussen’s fall tracking showing Trump up 2 to 3 points through the final stretch. My polling had six of seven main battlegrounds correct, missing the seventh, a tossup, inside the margin of error. Across all 14 of the battleground states we polled, Rasmussen Reports’ average error was only 0.7 points too favorable to Harris, and the average absolute error was only 2 points.
The more important story is how badly the rest of the industry misread the electoral outcome and the mood of the country. This should not have been a hard election to predict, with the final result being one of the top three margin shifts in the last 150 years. I am not the only one who produced accurate national data over the past year. The list is small, but it includes Atlas Intel, Emerson, NYTimes/Siena, and the Big Data Poll, run by Rich Baris, who caught the double-digit Democratic lead early and kept publicly warning people about it.
Most of the big media and academic pollsters, missed it, just as they did in 2024. Most polling shops were clustered at D+7 to D+9. There are hints that they didn’t trust their own eyes. YouGov actually had D+15 in September, then backed off to D+9 just in time to be on the low side, and outside the margin of error.
Sixty-four percent of the polls in the final RCP Generic Poll Average landed between D+7 and D+10, clustered tightly around the final D+8.8 average. These firms couldn’t find the Trump voter in 2024, and this time they somehow missed one of the biggest “blue tsunamis” imaginable. What are we even doing here?
Trafalgar and InsiderAdvantage, two firms which have had a very solid track record during the Trump era, hurt their accuracy by promoting that a “silver wave” powered by seniors would help Republicans close the gap with Democrats. This narrative, amplified in the press, proved to be wishful thinking. Older voters did turn out, but they split evenly between Republicans and Democrats Meanwhile, independents broke for Democrats by 20 points in the NBC exit poll, and by 26 among independents with no partisan lean – a total 180 degrees from 2024.
A relative polling newcomer, Quantus, dismissed the signs of Republican defection that Baris and I were finding, while showing a mid-single-digit Democratic lead before finally converging on the aggregate. Reuters, Harvard-Harris, Rasmussen, Echelon, Morning Consult and others all put out polling throughout the cycle showing the race far more competitive than it was. Even left-wing G. Elliot Morris’ Strength in Numbers national poll was only showing 6-7 point Democrat leads in June, July, and August.
Republican polling firm McLaughlin & Associates went further. Its data seems to have encouraged House Speaker Mike Johnson into making what in hindsight is a comically bad forecast – that the GOP would keep control of the House. Johnson based his prediction on a simple formulation: “The polls are wrong.” This was true, but in the opposite direction. The day before the election, John McLaughlin told Benny Johnson’s audience that “we should hold the Senate like 52, 53 seats … So the Senate should be ours,” after Benny introduced him as “the most accurate pollster in America.”
Democrats now look to have flipped nine Senate seats. McLaughlin’s last publicly available generic ballot had Democrats up five, in September. Fox News pollster Chris Anderson now says you “can’t model for a seismic election,” but the signs were all there that this election was going to be seismic long before the numbers began to show up in generic-ballot polling.
Anyone attempting to walk in George Gallup’s footsteps should have seen this coming for about a year and a half. The evidence was sitting in plain sight. Across 110 special elections since January 2025, Democrats beat Harris’ margin by an average of 12.5 points. Abigail Spanberger won Virginia by 15 and Mikie Sherrill won New Jersey by 14 last November. Democratic primary turnout was lopsided enough to point to a D+13 House vote. Prediction markets gave Democrats a 78% chance at the House in December 2025.
The online right was in open revolt over things like the Epstein files, the affordability “hoax,” Dow 50K, and of course, President Trump’s military campaign against Iran. And Trump-supporting Democrats and independents were constantly being shed as early as May 2025. Iran was a factor, but it’s simplistic to blame this shellacking on one issue. In my view, the collapse of the GOP on Election Day was caused by incoherent and lackadaisical governance and a complete failure to deliver an ounce of economic populism in a way that made Americans, now in a stealth recession, confident enough to keep voting Republican. The unpopular war of choice was the added ingredient that made the losses historic.
The foreboding GOP disaster was even more of an information bubble on the right because unwelcome numbers were “fake polls.” Early-vote trackers peddled registration math with no regard to shifting sentiment, and anyone offering honest polling feedback got treated as an enemy. In May 2025, my polling showed the first Republican generic-ballot loss in years. In July 2025 I wrote that “Republicans Are Going to LOSE The HOUSE.”
People who just two years prior had binged on my content now comforted themselves by deriding me as some kind of traitor, while ignoring the fact that I had been invited to the White House to provide guidance, early enough that it would have been effective – though it went mostly unheeded.
We’ve had almost a century of a formal public opinion industry in this country, and it feels that voters are farther than ever from being represented. The electorate deserves better.
Last Tuesday wasn’t a surprise to me. It shouldn’t have been a surprise to anyone.
This article was originally published by RealClearPolitics and made available via RealClearWire.
