In 36 Hours, Some People Will Be Very Right, and Some Will Be Very Wrong

In the final flurry of campaigning to decide which party controls the House and Senate for the next two years, polling has dropped as fast and furious as Democrats bailing out on Abdul El-Sayed in Michigan. 

Starting with the congressional generic polls, they're as widely arrayed as analyses of early voting and what it portends. Just in the last few days, Harvard/Harris had the race tightening to a D+4 model, well within the range of the Senate remaining in GOP control, and a much higher chance of the House staying pat. 

And then Atlas Intel, who was very accurate in 2024's presidential contest, dropped a D+14.1 poll yesterday. Even factoring in margin of error, that's two polling outfits with previous track records that seem to be sampling two separate countries right now. 

In 2024, polling's biggest loser, and there really wasn't a close second, was Ann Selzer's Iowa poll for the Des Moines Register. In it, she proudly proclaimed Kamala Harris would beat Donald Trump in the Hawkeye State, one of seven swing states that cycle, by 3 points. Life came at her really fast three days later when the final margin was Trump +13 on his way of sweeping all seven. She missed by 16 percentage points. 

Polling has only gotten harder since then, being virtually no one has landlines anymore, and precious few bother to answer calls from unknown numbers or respond to polling spam texts on their cellphones. Nevertheless, the polling industry soldiers on without any solid methodology, because Kalshi can't have all the fun. 

Yesterday in the New York Times, in partnership with Siena, Nate Cohn returned with a doozy.  

NEWS: The last Senate polls of 2026 from NYT/Siena: From me and @RuthIgielnik https://t.co/inA3u2Urv9 https://t.co/vXlfPOKes1

— Shane Goldmacher (@ShaneGoldmacher) November 2, 2026

Democrats are allegedly going to run the table of the swing Senate seats tomorrow, despite what early voting indicates. By the time North Carolina flips for Roy Cooper, Democrats will wake up not just with the House of Representatives, but a 54-46 majority in the Senate. 

Selzer called her results "a big miss," instead of what it was designed to be — a psyop to keep Republicans from voting on Election Day. It's hard not to view most polling these days through that lens, especially when polls tell you in the fine print they have no earthly idea how to contact enough people to get a representative sample. Take this disclaimer from the NYT/Siena poll:

I hate Nate Cohn so much https://t.co/gQgCOOwMo2

— Latinx Adjacent Doctor PhD (@TonerousHyus) November 2, 2026

It's quite a take for pollsters to assume that since Republicans won't answer their calls, take their surveys, or respond to texts, they are either not voting or voting for the Democrat instead. Let me give you one strong anecdotal piece of data — me. I went back through my call history for the past week. Between calls and texts, my combined ignorance to their pleas to respond topped a hundred between national, state, county, and even city mayoral candidates where I live. I have never missed voting in my life, and I have my absentee/contingency ballot already filled out in case God calls me home before I can vote on Tuesday. If I do get hit by that crosstown bus and cannot exercise my franchise, my instructions are for my vote to get dropped in the box before my body gets cold. I'm voting whether i'm here or not. On the propensity scale, I'm an 8/4 voter, and I wouldn't waste a second responding to a pollster. It's a lot more fun watching them fail. 

But since the New York Times is back to playing head games right before Election Day, let's run through the seven states to see if there's anything matching reality to back up their numbers. 

Michigan — El-Sayed 53, Mike Rogers 42 — D+11.

So much wrong with this, and that doesn't include the political earthquake dropped by Haley Stevens going to a Black church in Detroit with Mike Rogers, endorsing him in a selfie video, then doing media all day to explain why she's endorsing him. For a campaign that's allegedly up 11, El-Sayed and the Democrats did not react like her endorsement was no big deal. They were scrambling madly, and actually made the situation worse by appearing on 60 Minutes Sunday night with Ross Douthat. 

Abdul El-Sayed, running as a Democrat to represent Michigan in the Senate, discusses Palestinian and Israeli issues with 60 Minutes’ Ross Douthat. https://t.co/0tUeaUFShn

— 60 Minutes (@60Minutes) November 2, 2026

Palestine has the right to exist. "Yes". Full Stop. Same question about Israel. "Yes, if...". For the Jews, there are always conditions for whether they should be allowed to have a state. Haley Stevens really had only one request of El-Sayed. She wanted him to condemn the antisemitic ad and say antisemitism is bad. He couldn't do it. Instead, after her endorsement, he attacked her angrily for taking all that Jew-y AIPAC money. 

El-Sayed wasn't done. Imagine being the parent of an adolescent girl these days, and this guy wishing to be Michigan's next senator says this about men getting naked in the locker room next to your daughter. 

If El-Sayed numbers drop drastically and he loses the election (🙏), it won't be because of Haley Stevens, it won't be because of the ad, and it won't be because of AIPAC. It will be because of THIS 60 Minutes interview.

— Kosher Chutzpah (@KosherChutzpah) November 2, 2026

I am blessed to have been the step-father of two adolescent girls, and I would have come unglued if a politician said this about a policy so preposterous while they were in school. 

But as to the poll's claim that El-Sayed has this contest in the bag, let's look at what the makeup of their poll shows. 

• 18–29: 12 percent
• 30–44: 20 percent
• 45–64: 32 percent
• 65+: 33 percent

Again, nationally, 18-29 is usually a D+31 cohort, and the 65+ group starts at R+4 and gets redder the more rings around your trunk. How did Michigan's early vote, both in-person and mail-in, shake out by age group? 

• 65+: 57.6 percent
• 55–64: 17.0 percent
• 45–54: 9.2 percent
• 35–44: 7.1 percent
• 25–34: 5.4 percent
• 18–24: 3.8 percent

It's not so much a question if the kids are not alright. They apparently aren't. They're just not voting anywhere near the margins for which the poll gives them credit. And even the 30-44 range in the poll, which favor Democrats but by a much smaller margin, are not showing up, either. It's not 20 percent, it's about half that. And the seniors, which NYT/Siena has representing a third of the votes to be cast in real life, is just shy of double that in early voting. And if you add in the 55-64, which also skew more narrowly Republican, that's 75 percent of all early votes cast. 

There was no roll-off of seniors early voting as the days progressed. They were either consistent or gradually increased both in raw numbers and percentage up to the deadline. Democrats, however, saw a roll-off after a heavy first few days of voting both for mail-in and in-person. 

In 2024's early voting, Democrats, using modeling from primary propensity, carried an 18-point edge going into Election Day, with Donald Trump beating Kamala Harris by a point-and-a-half. The bank this time is about three points below that. Republicans vote heavily more than Democrats on Election Day. With the extra notoriety and national publicity around this contest, the circus from yesterday alone, and El-Sayed not able to lock down Blacks and Jewish blocs in his base, this could easily be one of the biggest misses of the night for polling. 

Alaska

In Alaska, this is another race that has gotten more than its share of attention in the national spotlight after Mary Peltola, Democrat, lied, cheated, and weaseled her way through the campaign. DNC lawyers and campaign staffers combed the state until they found some doofus named Dan Sullivan to put on the ballot as a decoy hopefully to siphon off votes from her actual opponent. What does NYT/Siena predict? 

Peltola 48 — Dan Sullivan (the real one) 46 — D+2. 

How do they get there? 

• 18–29 - 12 percent
  
• 30–44 - 22 percent
• 45–64 - 33 percent
• 65+ — 30 percent
  

• 65+: 44.0 percent
• 55–64: 17.5 percent
• 45–54: 10.4 percent
• 35–44: 7.7 percent
• 25–34: 2.6 percent
• 18–24: 0.1 percent
• Unknown: 17.7 percent

What can I tell you about the unknowns? It's Alaska. There could be 17 percent of the state that actually has no idea how old they are. But among those who do, take a look at that 18-24 range. 0.1 percent. That's only a tick above the Blutarsky GPA in Delta House at Faber College. 0.1 is akin to ordering an old Radio Shack do-it-yourself AM transmitter you can still find on eBay and launching your own station. You'd still get a 0.1 in the Arbitron book for just being there. 

Less than 3 percent of the youth vote has turned out, and seniors are over-performing yet again. Republicans take between an 8-and-9-point early vote bank into Election Day, and hold between 10 and 15 points in registration advantage. Assuming independents do not suddenly embrace cheating and fraud in wholesale fashion, Sullivan (still the real one) will win this. The only caveat is ranked choice voting, which means only God and Lisa Murkowski know how it'll actually end up. 

Iowa

I wrote up the state of the Iowa Senate race last week, so I won't revisit all of it here, but let's take NYT/Siena's topline. 

Turek — 48, Ashley Hinson — 47 — D+1

To get there, this is the age group they see for their sample. 

• 18–29 - 12 percent
  
• 30–44 - 20 percent
  
• 45–64 - 32 percent
  
• 65+ - 33 percent
  

Iowa's actual early voting period, which got off to a fast Democrat start before rolling off like in so many other states, looks a lot different. 

• 18–29: roughly 6–7 percent (18–24 ~4 percent + part of 25–34)
• 30–44: roughly 11–12 percent (remainder of 25–34 + 35–44)
• 45–64: roughly 26 percent (45–54 + 55–64)
• 65+: roughly 58 percent

Half the kids they thought would vote are in, combined with 25 percent more seniors than they expect. 

In 2022, Democrats took a 64,000-vote bank into Election Day, only to run into a Republican-leaning electorate that resulted in a 148,000-vote win in the Senate race. This year, Democrats only take 53,000 votes with them as a bank going into tomorrow. The difference between then and now? About 100,000 more registered Republicans than Democrats. If Republicans vote Tuesday in roughly the same turnout as they did all through the early voting period, that bank of Democrat votes will be zeroed out by lunchtime. 

Kansas

In Kansas, NYT/Siena has Hamilton up five points over Roger Marshall — 49-44. As with the other states, the devil is in the crosstabs. 

• 18–29 - 11 percent
  
• 30–44 - 20 percent
  
• 45–64 - 32 percent
  
• 65+ — 34 percent
  
  

Once again, actual early voting in the Sunflower State is something radically different. 

• 65+: 49.0 percent
• 55–64: 18.7 percent
• 45–54: 12.2 percent
• 35–44: 9.4 percent
• 25–34: 6.3 percent
• 18–24: 4.3 percent

Half the young'uns, 50 percent more seniors who really do not wish to see Medicare for all become Medicare for none if the socialists grab control and break the system. 

Republicans carry a 10,000-vote advantage in early voting into Election Day, which is significantly below 2022's 366,000 edge and 2024's 130,000 bank. How did Kansas vote on Election Day in both cycles? In 2022, Election Day was R+12. In 2024, it was R+18.7. 

In 2022, the GOP had an advantage in registered Republicans over Democrats by 360,000. Today, it's around 380,000. The only way Roger Marshall loses is Kansas Republicans don't vote on Election Day. That's not likely given their previous propensity.

Maine

NYT/Siena gives the replacement schlub, Troy Jackson, a 6-point edge over Susan Collins, 51-45. 

This is what the poll thinks the electorate is going to look like. 

• 18–29 - 9 percent
  
• 30–44 - 19 percent
  
• 45–64 - 31 percent
  
• 65+ - 36 percent
  

Early voting is now in the books. What has reality been so far? 

• 65+: 60.9 percent
• 55–64: 17.2 percent
• 45–54: 8.4 percent
• 35–44: 7.0 percent
• 25–34: 4.3 percent
• 18–24: 2.3 percent

I'm sensing a pattern here. Are you? That 61 percent of the votes in? Those are Collins people. The youth vote is not turning out. They still might all over the country, but so far, they're just not. These are voters that Democrats and pollsters need in order to make the math work. 

The bank of early votes by Democrats this cycle is on par with the bank they took into Election Day in 2022. But on Election Day, Republicans had a 10-point edge, and Susan Collins beat the odds, and the pollsters. Will she again this year? Again, it's a ranked choice vote state, so it's very hard to predict. But as has been the case in a lot of other states, Democrats used up their high-propensity voters in the first part of early voting, and then their numbers tailed off near the end compared to Republican-leaning early voters which increased steadily going into Election Day. How much did Democrats front-load? We'll find out tomorrow, but I expect Collins to win until someone beats her. 

Ohio

Old, tiring Sherrod brown allegedly had a five-point lead over Jon Husted in Ohio, according to the poll. Their projection of what the electorate of the Buckeye State is a similar pattern. 

• 18–29 - 12 percent
  
• 30–44 - 20 percent
  
• 45–64 - 33 percent
  
• 65+ - 31 percent
  

Ohio's early voting was extensive and detailed. And it didn't look like this sample at all. 

• 65+: 54.3–55.4 percent
• 55–64: 18.4–18.6 percent
• 45–54: ~9.8–10.1 percent
• 35–44: ~7.3–7.5 percent
• 25–34: ~5.4–5.7 percent
• 18–24: ~3.7–3.8 percent

Our Townhall friend and colleague calls this the Silver wave instead of a blue wave. The odds that seniors all over the country are overperforming in early voting by numbers 50-100 percent over polling methodology so that they can finally embrace Medicare for All and give socialism a chance is not very high. And if socialism was selling, you'd see college-aged kids hitting their numbers. They're not, and Ohio is no exception. Democrats started out very hot in early voting, but in virtually every county, exhausted their highest propensity voters, and by the end of early voting, Republicans cleaned up daily wins.

The bank Democrats take into Election Day is around 120,000 votes. In 2022, it was only 75,000, and in 2024, Donald Trump had a quarter million votes in his bank going into tomorrow. One early voting analyst sees Ohio shaking out this way. 

Prediction Announcement: Ohio Gov and Senate This is another tough call but here I believe that based on the early voting and overall red lean of the state, both Republicans will win. The early voting started off solid for Democrats, but has slowly become more Republican. Most of the electorate will come out on Election Day, which I think will keep Republican candidates over the edge. Husted is a strong candidate facing a strong Democrat in Brown, but he already failed two years ago and I think he does so again. In the Governor's race, Ohioans have quite the choice but I think they select Vivek over Acton. This will be closer than what they are used to, but Republicans should keep both seats.

— CA ET Nerd (@earlyvotedata) November 2, 2026

Why? Because Democrats probably have exhausted most of their highest-propensity voters, and Republicans still have plenty in the tank that tend to turn out on Election Day. 

Texas

Finally, in Texas, it's another state with a lot of data that doesn't really give you what you want — party ID. You have to extrapolate using two methods — either past primary participation, or past precinct voting tendencies and how high or low the turnout is by precinct. In the poll, this is what they think is going to happen. 

Talarico - 50, Ken Paxton — 43 — D+7

Their age group makeup? 

• 18–29 - 13 percent
  
• 30–44 - 22 percent
  
• 45–64 - 34 percent
  
• 65+ — 29 percent
  

Early voting thus far? 

• 18–24  ~5.7–5.9 percent
  
• 25–34 ~9.7–10.0 percent
  
• 35–44 ~12.3–12.6 percent
  
• 45–54 ~15.2–15.4 percent
  
• 55–64 ~19.6–19.7 percent
  
• 65+ ~34.7–37.5 percent
  

This is as close to the poll's crosstabs as any of the seven states, and they're still over-sampling the front edge of the voting demo and under-sampling the older set. 

Based on prior voting in a partisan primary, Republicans are probably taking 130,000 net votes as a bank going into Election Day. That's about 1.5-2 percent compared to a 9-point edge in 2022, and an 11-point edge in 2024. Republican-leaners beat Democrats on Election Day, 2022 by 310,000, and by 90,000 two years later. Republican turnout is a little softer than 2022, while Democrat turnout is a little stronger. But if Republicans turn out for a normal midterm cycle on Tuesday, especially with Democrats exhausting their highest propensity voters in early voting, the outcome of this will be closer than it ought to be, but probably not that close. 

I won't spend a lot of time on this, but it should be said about another datapoint in the NYT/Siena poll's methodology. Of all their respondents across seven states, they asked whom they voted for in the 2024 presidential election. This is the call-back question. Here's the result compared to how each of the seven states actually voted. 

Editor’s Note: The 2026 Midterms will determine the fate of President Trump’s America First agenda. Republicans must maintain control of both chambers of Congress.

Help HotAir continue to report on the Democrats’ radicalism and inform voters as our nation faces a crossroads. Join HotAir VIP and use promo code FIGHT to receive 60% off your membership.