Okay, the polls are getting a little ridiculous

Hey — I'm Isaac Saul, the founder and executive editor of Tangle. This is my new semi-regular newsletter A Grain of Saul, where I dive into some topics we don't get to in our main newsletter.

One of two things is currently happening:

1. President Trump is burning down a powerful political coalition in historic time, or
2. The polls are very misleading.

YouGov recently ran a poll on the Senate race in Texas. Among the respondents, 52% said they had voted for Donald Trump in 2024. 40% said they had voted for Kamala Harris. 41% self-identified as Republicans, while just 26% self-identified as Democrats.

And yet, in that poll, 50% of the respondents said they were planning to vote for the Democratic Senate candidate, James Talarico, and 44% said they were planning to vote for the Republican, Trump-endorsed candidate Ken Paxton. Of the people who responded that they were voting for Talarico, 97% said their mind was made up, while just 3% said there was still a chance they’d switch to Paxton.

Again: James Talarico, the man President Trump has dubbed “James TalaFREAKO,” is polling six points ahead of the president’s preferred candidate, in Texas, among a group of respondents who broke 52–40 in the president’s favor less than two years ago.

It’s not just in Texas, either. In Kansas, a state Trump won by 16 points in 2024, the Senate race is now a toss-up. A New York Times/Siena College poll published earlier this month shows the race tied. Kansas hasn’t elected a Democrat to the Senate since 1932. 

Fox News found that government corruption was the issue voters were most concerned about, ahead of inflation, immigration, health care, and the war in Iran.

Data journalist G. Elliott Morris found that low-information voters have turned most sharply against Trump, a reversal of 2024, when Americans who didn’t consume any news at all intended to vote for Trump over Harris by a 60% to 40% margin.

Texas districts that President Trump and Republicans pushed to gerrymander to their advantage are now in play for Democrats because the Hispanic vote has swung so quickly back to the left. Exit polls showed Trump winning the Hispanic vote in the state by about 10 points in 2024. As Nick Miroff noted in The Atlantic, a survey this summer of Hispanic Texans likely to vote in the midterms showed them breaking for Democrats by a 2-to-1 margin. 

Even in Arkansas’ Senate race, which Cook Political has rated as “Solid R” (meaning it’s not competitive), a Hendrix College poll of likely voters from August produced a result with the Democrat Hallie Shoffner at 47%, incumbent Republican Sen. Tom Cotton at 44%, and Libertarian candidate Jeff Wadlin at 3%. This month, that same pollster found Cotton was nine points ahead of Shoffner (50 to 41), but the fact the race was ever a point of discussion is pretty shocking. 

The New York Times, CNN, and Strength In Numbers/Verasight all have Democrats with a 10+ point lead over Republicans on generic ballots, as of this month.

To emphasize: Donald Trump won the popular vote and swept all seven swing states in an election less than two years ago.

If this polling sample is representative, Democrats won’t just get a shock Senate win in Texas — they’ll win nearly every competitive House and Senate race across the country, and they’ll redraw the electoral map that we’ve become accustomed to since the turn of the century.

So… I’m not really sure what to do with polls like this. 

It doesn’t take much effort to buy that Trump is burning down his coalition. Core to Trump’s electoral appeal was not not wasting money abroad and not starting new wars. The Iran war broke all of those promises in spectacular fashion. Americans have now paid for the war in blood and treasure, and no end is in sight. Trump has insisted it’s over, and we won; but it’s not, and we haven’t. And most of his supporters seem to recognize that.

Trump also promised to swiftly address the inflation and cost-of-living issues that peaked in the middle of President Biden’s term. But inflation broadly persists. ABC News reported this week that “filling up a Ford F-150 truck now costs about $100, up $30 from a year ago. Mortgage rates have jumped to levels that have depressed home sales. And for the past 12 months, consumer prices have risen faster than paychecks, making the average family poorer.” So, we’re two years into Trump’s second term, and things are trending in the wrong direction. And unlike other presidencies, these economic forces can be tied directly to Trump’s policies: the war in Iran and the constant tariffs. 

Even Trump’s signature issue, immigration, has inflicted some political pain upon his base. In exchange for a more secure southern border, those worried about unchecked immigration have now also had to endure unchecked armed law enforcement in their neighborhoods. DHS agents have killed multiple American citizens (on video, no less) while arresting, detaining or assaulting hundreds more, resulting in collapsing support for ICE across the country. There are stories upon stories of Trump voters experiencing the mass deportation effort and deciding they didn’t much like it. 

All the while, the “Drain the Swamp” president is also very clearly self-dealing, enriching himself and his family and pardoning all manner of scummy white-collar criminals across the country (usually friends, political allies, or donors). 

Incredibly, Trump isn’t even playing defense on these facts. He seems preoccupied with Washington, D.C. construction projects and the stock market. Just last weekend, he actually told a crowd not to worry about how much they were paying at the pump because Big Oil companies were making hundreds of billions of dollars. As Punchbowl News reporter Jake Sherman said upon seeing the video of Trump’s remarks, “needed video to believe he actually said this 9 days before an election.”

Meanwhile, other promises have died on the vine. There is no SAVE America Act, no healthcare reform, and no lasting immigration reform.

So, if you want to tell me that all of the above has been a disappointment for both diehard and reluctant Trump voters, I’d believe you. Maybe we’re now seeing the political price for all of this. Maybe, as is typical, the sequel isn’t as good as the original. Maybe all the voters who swung to Biden in 2020 and back to Trump in 2024 are remembering why they left Trump in the first place. Maybe Trump voters are somewhere between apathetic and furious, and the polling is showing us that. Maybe this is why Republicans running in these races are scrubbing Trump from their websites while figures like Tucker Carlson, Nick Fuentes, and Marjorie Taylor Greene proclaim their regret for supporting the president.

If this is all true, then a Democratic tsunami is surely coming: A healthy House majority is guaranteed, and control of the Senate might come with a 3–5 seat margin (an outcome that was literally unthinkable two years ago). 

Personally, right now, I’m more confident in this theory than the “polls are wrong” theory, which is why I’ve started to predict Democrats will take back the Senate on election night. 

But am I confident in that prediction?

I can’t say I am.

Because these unbelievable polls are just, well, unbelievable. By now, you probably know that pollsters are trying to solve for the reality that conservatives are less likely to answer surveys than Democrats. They attempt this through various methods like weighting results for the expected electorate or screening for likely voters. In basic terms, that means if a pollster thinks an upcoming election’s voters will be made up of 40% Republicans, but their survey sample is made up of 30% Republicans, they’ll weight the results a little bit toward Republicans.

It’s a science, but it’s also messy. And while it’s a bit of a misnomer that the polls are always wrong (we’ve had some big polling misses, but also some very accurate polling), it is true that polls can only tell us so much. They represent what respondents say, but they can’t tell us who will actually vote, or how late-deciders will break. 

To that end, I just don’t think James Talarico is actually six points ahead of Ken Paxton with Texas voters who broke 52–40 for Trump. I don’t think the Arkansas Senate race will actually be remotely competitive. I don’t believe government corruption is actually the #1 issue for voters. This is all instinct and based a lot on anecdotal experience, of course, but it’s also a response to being burnt by media hype and polling in the past.

Take Maine: Susan Collins is trailing Troy Jackson in Maine; Jackson is the replacement candidate for Graham Platner. Jackson is not an extraordinarily captivating person (he certainly doesn’t have the “juice” that Platner did, pre-scandal). The last time Susan Collins was on the ballot in Maine, in 2020, she trailed state Rep. Sara Gideon in nearly every single poll. You probably don’t remember Gideon, because Collins trounced her: Collins won by 8.6 points in an election where Biden won the state by nine points. 

So, do I trust the polling in Maine? I don’t think I can. All signs point to a Jackson win, but I’m deeply unsure of the signs themselves. 

On the other hand, maybe I’m only doubtful because I’ve been burned before, which isn’t exactly a logical chain of thinking. Texas is a state that polls pretty well. If anything, in this environment, we have good reason to believe Texas polls are undercounting Democratic support. In 2018, the last midterm election under Trump, Beto O’Rourke came within 2.6 points of flipping Ted Cruz’s Senate seat despite polls consistently showing him three to 10 points behind. If the polls are off in that direction, Talarico might win the Senate race by four or five points — or more. 

I’m not suggesting the polls are fake (though, apparently, we do need to worry about that). I’m also not suggesting that Democrats are going to bungle the House and Republicans will keep the Senate. Trump did so well in 2024 partly because being an incumbent is extremely difficult post-Covid; now his party owns the incumbency, and he’s not even on the ballot. The GOP really might get walloped. 

I am suggesting that the tsunami may “only” be a rather normal wave. A combination of media hype and pollsters trying to correct for past misses could be setting up the delusional belief that Democrats are pulling off a historic electoral comeback, but maybe it’ll be more like the average midterm blowback. Take just one example: NPR reported this week that “just 42% of likely voters under 45, 53% of non-white voters and 56% of white women with college degrees said they are ‘very enthusiastic’ to vote,” while the percentages for “MAGA voters (68%), white women without degrees (66%) and white evangelical voters (61%)” were significantly higher. Given that midterm elections are always a game of turnout, these numbers should dampen some of the hype — but they really haven’t gotten much purchase in the press.

Nobody can tell you what will happen on Election Day, and certainly not me. I’m still predicting Democrats take back the Senate, which would be a remarkable turnaround from 2024. But if we witness a very normal marginal midterm blowback, which results in a divided Congress, I won’t be particularly surprised, either.