I know, I feel like we ask this question every election cycle. But, each time, it’s a fair and legitimate question. In 2024, when polls claimed that Kamala Harris was beating Trump, I knew better than to trust those numbers at face value. I looked at the crosstabs and found data that undermined the topline results. Should we be just as skeptical of 2026 midterm polling?
Maybe. Let me explain.
On Tuesday, CNN's Harry Enten looked at some recent generic ballot polling, and what he revealed (probably without meaning to) is a pretty compelling argument for why you should take every single one of these numbers with a fistful of salt.
ICYMI: Gutfeld Delivers an Epic Autopsy of the Left’s Dying Race Card
Here's the situation. Democrats are leading Republicans on the generic congressional ballot. That much, most polls agree on. But by how much? Well, that depends entirely on which poll you're reading. The New York Times has Democrats up by 10. Marist matches that. The Washington Post puts it at five. Fox News is also on at five. CNN's own poll has Democrats up by just three points. That's a seven-point swing between the lowest and highest readings, and every one of these outfits is measuring the same electorate at roughly the same moment in time.
Of course, we know this happens, and Enten pointed out this isn’t actually a problem. "This is a normal thing. This is what happens when you have margins of error that are going out there. You'd expect an average, but then you have a range of possible results around that average. And we're not necessarily sure which one is right if the election were held today."
That’s all fine and good, except there’s a huge difference in the outcomes produced by a three-point Democratic lead and a ten-point Democratic lead, ranging from Republicans holding the majority to Democrats having a blowout victory as big as 30 seats.
"All of that is within the margin of error at this point," Enten acknowledged, "especially given that we're months from the election."
So the polls are simultaneously telling us Democrats might not take the House and that Democrats might absolutely demolish Republicans in the House. Both are "within the margin of error." Both are being reported. Both are shaping narratives. This is the state of political polling in 2026.
The answer to this problem, of course, is to do what RealClearPolitics has been doing for about 20 years now: take all the polls and average them together. And the polls cited by Enten give us an average of +7 points for Democrats, up from a four-point advantage on Election Day in November 2025. That average, Enten said, "would result very much assuredly a clear Democratic majority." He also pointed to a Kalshi prediction market giving Democrats a 69% chance of flipping the House.
"Democrats are still in the catbird seat here," he said. "They're likely to gain the House, likely to gain the Senate, but it's by no means a guarantee."
Not a lot of trust in the data, is there? Which is basically where I’m heading now. Let’s go back to 2024, when a whole bunch of national polls showed Kamala Harris beating Donald Trump. And then she lost every single swing state.
Many of us questioned those polls, and with good reason. Buried inside those same polls, in the crosstabs that most people skip past, were signs that made it clear that there was no way Kamala was winning. The crosstabs were showing record support for Trump among minority voters. They were showing unusual movement toward Trump among young people. The top-line numbers were screaming "Harris wins" while the internal data was quietly assembling a very different picture. The pollsters and the pundits focused on the headline number. Voters went ahead and re-elected Trump anyway.
None of this means polls are worthless. Enten is right that averages smooth out noise, and right that Democrats currently hold an advantage in the available data. But the range of outcomes he described — from "Republicans hold the House" to "Democrats pick up 30 seats" — isn't really a polling story. It's a confession that the polling industry is essentially giving us a number and hoping for the best.
If the polls today were those on E-Day, a very wide range of House possibilities would exist.
— (((Harry Enten))) (@ForecasterEnten) May 19, 2026
GOP could hold the House with redistricting (see CNN poll) to Dems blowing the GOP of the water with 20-30+ seat gain (see NYT poll).
Average: Dems favored & gaining nationally. pic.twitter.com/sMiGY2SQSf
So, should you be paying attention to the 2026 national polls right now? You can. I do. Just remember, there is a much bigger picture to look at outside of one or even a few polls.
