Posted on 10/21/2026 2:58:50 PM PDT by MinorityRepublican
Nate Silver's midterm elections forecast shifted slightly toward Republicans after he removed a series of fake polls that he said had been roughly 1 point more favorable to Democrats than the broader consensus.
The famed polling analyst said the biggest change came in Michigan, where Democratic U.S. Senate nominee Abdul El-Sayed's lead fell from 3.6 points to 3. Silver noted that the movement also reflected the addition of a Marist poll to the model that showed El-Sayed 2 points ahead of Republican former Representative Mike Rogers.
"Overall, our forecast is little changed, though that's partly because some strong generic ballot polls entered this AM for Democrats," Silver posted on X on Wednesday morning.
The fake polls came from Wedgewood Polling, which revealed on Tuesday that it had fabricated most of its polls. Newsweek has corrected and retracted stories that included data from Wedgewood Polls.
(Excerpt) Read more at newsweek.com ...
Dear FRiends,
We need your continuing support to keep FR funded. Your donations are our sole source of funding. No sugar daddies, no advertisers, no paid memberships, no commercial sales, no gimmicks, no tax subsidies. No spam, no pop-ups, no ad trackers.
If you enjoy using FR and agree it's a worthwhile endeavor, please consider making a contribution today:
Click here: to donate by Credit Card
Or here: to donate by PayPal
Or by mail to: Free Republic, LLC - PO Box 9771 - Fresno, CA 93794
Thank you very much and God bless you,
Jim
Your source says TIME not Newsweek.
Removing fake polls from the group is like the words people used to write on particularly dusty and dirty cars and trucks in parking lots:
Don’t wash off the dirt.
It’s all that’s holding this thing together. 🛻🚗
Nate, so you removed some fake polls from your analysis. Good. Has it dawned on you that maybe your analysis STILL contains other fake polls?
Nate’s a pure data guy - and I actually know him a bit because I’m also a big sabermetrics fan (and Nate was doing advanced metrics in baseball long before he entered the political realm).
The data is the data. Lots of people like to give him crap over 2016 — but the reality is that his modeling gave DJT a better chance than anyone else. I can’t help people who simply don’t have a foundation in statistics — but a “28% chance”? Well, fall back to baseball — if you think a 28% of getting a hit means nobody ever gets a hit? You just don’t understand statistics and probabilities.
Despite the fact he is a lefty - I knew that long ago, before he got into politics (I’ve exchanged with him a long time ago, though more on baseball analytics) - he’s good at statistical modeling.
If you still know how to get in touch with Nate, can you ask him how he can be sure there aren’t more fake polls that are still included in his analysis. Thank you!
But that’s the point!
28.6% was certainly higher than any other analyst or predictor gave him.
And harkening back to baseball - a .286 batting average would be an extreme, high-offense outlier (let’s set aside that OBP matters more for a minute).
The model wasn’t wrong - because it’s not a BINARY thing. And the model was also predicated on 50 different probabilities, and drift inevitably stacks.
Elections are one and done. And they only happen - for POTUS - every 4 years.
This is simply how statistics work. People do bowl 300 games. Hitters do hit .400 - or even .350... and even Tony Gwynn or Wade Boggs or Ty Cobb often went for 0-4 (or even 0-5). Even Mario Mendoza - of the “Mendoza line” - had a 4-4 game.
That's still the case today.
Wrong, Wedgwood admitted to literally fabricating data, the problem with polls is they are entirely propaganda for a democrat narrative.
Most of the other polls are doing the exact same thing
OMG. You’re arguing he’s better because he went as high as 28%.
He was still...wait for it...
WRONG!!!!!!!
I don’t need to ask - he’s written about it.
Not just for politics - but more broadly to understand statistics and probabilities, I highly recommend his book “The Signal and the Noise: Why So Many Predictions Fail—but Some Don’t”.
Nearly 15 years later, I still think it’s the best book to understand statistics, probabilities - and why “probabilities” are just probabilities, NOT certainties.
Forget politics - I’ve personally found it really valuable in investing.
But in any case - I *know* he’d say that ‘polls’ are just input data points. If a pollster has a long history? You can measure their success. If they don’t? Add them on the pile, but weight them lower. Bad data is certainly a GLARING problem - and FAKE data (like the wedgewood thing) is worse than bad. But - to properly model? You weight. And you might STILL get a bad number.
Specific to “fake” polls? That’s how the “wedgewood” scam was discovered — it was a different analyst (and yeah, another lefty - G Elliott Morris/50+1) that sniffed it out. Raw data is messy - and that fake poll was just “too smooth”. Data doesn’t work that way. You model, you normalize, you smooth.... but the reason the wedgewood scammers (and it looks like kalshi betting gambit) got caught is because it the underlying raw data pre-modeling (everybody does turnout/LV vs RV etc modeling) was just too... normal.
Just for example - “herding” is a common thing in polling. Pollsters aren’t “faking” data - rather, if they’re an outlier? They adjust their modeling to match the “herd”.
In years past - he’s pointed out that some firms (Atlas, for example) have proven better at modeling because they avoid herding. As the minority conservative who follows and reads - I hear plenty of the inverse.
Absolutely.
But modeling from REAL polling firms does *try* to adjust.
Look at 2022 - they overcompensated for “shy/no response”. We flipped the House, but it was closer than expected and I thought we’d flip the Senate, too.
As a Republican? My big fear: 2016, 2020, and 2024 are markedly different than 2018 and 2022 when you look at polling + probability modeling vs results.
I DO hope 2026 breaks the trend, but it’s a bet against history.
I lay no claim to liking DJT - he’s always gotten my vote by being “better than the alternative”. But - for a decade now? He doesn’t have coattails, especially when he’s not actually on the ballot.
NC will be a bloodbath we could lose 2 seats
Remove ALL the fake polls and there’d be nothing left.
If I had to bet I would agree with you, but the data says Republicans are historically way ahead and the Democrats historically behind on early voting. But there’s still two weeks to go...
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.