Nate Silver Tweets

The other weird thing about that Marist poll in Maine is that their LV version is 10 points off from the polling average in the *governor's* race. And they show Pingree doing worse than Jackson, which no other poll does; the whole Q in Maine is about Susan Collins's strength. Original post

FLIPR update. A couple of contradictory trends: 1) D edge on the likely voter version of the generic ballot is down a bit and is now basically aligned with the default version. That partly reflects some pollsters, like Marist, showing better #s for the GOP among LVs than RVs. Original post

Repost by @NateSilver538 Really excited to announce that @TheArgumentMag, Silver Bulletin (@NateSilver538) and @galendruke will be hosting an election night watch party, with real-time analysis and reactions as results roll in! Come join us! Click to see original repost

Within Original post

FLIPR update. Model starting to get a bit less skeptical about Osborn's odds in Nebraska, otherwise not a ton of new high-quality polling today. Original post

There's some difference of opinion re: Nov. 3, but less than most cycles in the past. Even the most optimistic non-insane Rs are like, "All these Senate races are toss-ups." Even the most optimistic non-insane Ds are like "We like our odds, but we'll believe it when we see it." Original post

I'm in the NYT with some other smart folks talking about the midterms. Original post

I'm not sure what it would take to de-entrench the presumption that polls systematically underrate conservative parties or candidates. It's not true historically, but it's a cheapie take that will never really get you criticized. Original post

Evan I thought your book was OK, and I'm certainly very receptive to the idea of D organizations having become too woke/progressive/etc but this Cheesecake Factory stuff makes you look like a humorless reactionary idiot take the fucking L. Repost of @evanwch: Original post

The gap between FLIPR and prediction markets is closing, but I thought some of you would appreciate a column that goes into detail about the disagreements. It comes down to state-by-state differences of opinion, rather than anything overly "macro". Original post

Repost by @NateSilver538 The midterms are upon us. Will Democrats capture both the House and Senate? Are the polls trustworthy? Will Republican gerrymandering backfire? I spoke with @NateSilver538 about the midterm elections: Click to see original repost

The people complaining about Obama’s Cheesecake Factory line sound like performative humorless scolds, maybe this is Woke 2.0. Repost of @SteveKrak: Original post

Repost by @NateSilver538 ELWAY is updated with yesterday's game. We still like New England quite a bit, but they're not quite an inner circle contender anymore... Click to see original repost

What should we make of races like Michigan, Texas, and Georgia where there's a big split in the polls between Senate and gubernatorial races? Do these gaps tend to close by Election Day? Answers in today's Model Talk: Original post

If everything is super correlated with partisanship but there's a D +10 map, meaning a 12-point shift from 2024, then all the OH/TX/IA/AK tranche of states that Trump won by ~12 points are toss-ups anyway. And partisanship mattering more would help Ds in ME, MI and MN. Original post

FLIPR Senate odds have been quite flat over the past week or two and yet Dems are surging on prediction markets. I guess everything just converges or maybe there are some perceptions being baked in about early voting data. Original post

Back in the day, one of the most reliable ways to spot fake pollsters was large volumes of telephone "polling" that were implausibly expensive given the firm's business model, or lack thereof. This heuristic doesn't work as well given the presence of cheap online panels. Original post

A few points about Wedgewood. 🧵 1. They're totally full of shit when they say there isn't any vetting going on. The NYT, whose polling feed we use at Silver Bulletin, extensively vets new pollsters. And FiftyPlusOne caught them. I've caught fake pollsters before, too. Original post

Repost by @NateSilver538 .@NateSilver538: This is the most the Democrats have been getting along in a long time. Click to see original repost

IDK, it seems like certain pollsters are concerned their samples are too D, maybe for good reasons and maybe not, and are using their likely voter models to curb that. Original post

Repost by @NateSilver538 For those interested in seeing the second term Trump and second term George W. Bush net approval ratings in one figure: Repost of @NateSilver538: Click to see original repost

Trump approval starting to look like a dead ringer for Bush's second term. Original post

We removed Wedgewood "polls" from our database, will write more about that tonight/tomorrow. These polls tended to be ~1 point more favorable for Ds than the consensus. Biggest change in MI (El-Sayed goes from +3.6 to +3.0) but that also reflects the Marist AES +2 being added. Original post

Are you familiar with the American Midwest (Mɪˈdwɛst), a region associated with agriculture and industrial production that includes states such as Ohio, Iowa, Kansas, Michigan and Nebraska? Repost of @RyanGirdusky: Original post

If the generic ballot/national environment is really D +10 or something, that reflects a 12-point shift from 2024, so a whole tranche of states (IA, TX, OH, AK, FL) that Trump won by ~12 points behave as toss-ups. Original post

What annoys me about the polling error debate is that there's a genre of commentators who seem to assume that everyone in the industry has been in a coma since 2014. Original post

Last week, it looked like there might be the beginnings of a Republican rebound on the generic ballot, but it's back to D +9 now. Original post

The thing that's annoying about election analysis is that you'll never be criticized for saying "thing that happened last time will happen again" which makes people kind of boring and risk-averse and also leads to a lot of overfitting from n=1 or n=2 samples. Original post

Who's the best quarterback in the NFL? Patrick Mahomes? Josh Allen? Lamar Jackson? Well, what if it's Brock Purdy? Original post

I started following politics seriously in 2006 and since then there's only been one election cycle, 2012, that went well for the president's party. 2022 probably a distant second but closer to "not badly" than "well". Original post

It's a day that ends in "y", which means that today's Model Talk is about systematic polling error. Right now, if you go purely by polls, the tipping-point state is Maine, where Democrats lead by 2.8 points. Original post

Today's FLIPR update. And a Model Talk coming soon. Original post

Hmm convicted felon seems jealous he's not interesting enough to get the Nathan Fielder treatment Repost of @MichaelAvenatti: Original post

Big runup in oil futures over the past week. Gas prices have stabilized/declined a bit but are lagging relative to oil. This is what may be facing voters as they go to the polls on Nov. 3. Original post

Michigan is core Midwest, absolutely as Midwest as it gets. Repost of @JoshuaBarzon: Original post

My mental model of the election is kind of asymmetric at the moment. Like, probably the median is a bit worse for Ds than FLIPR shows; it's easy to see mild D underperformance. But then there's also a tail of crazy D overperformance. Original post

Repost by @NateSilver538 .@NateSilver538: "I get annoyed, by the way, when people are like 'Well it's good that people are mad at Trump, but they're mad at him for the wrong reasons!'" Click to see original repost

In theory, you should learn more as the election approaches. But in practice, the discussion often gets dumber in the final 3 weeks. Particularly in elections when the polls aren't changing much. Original post

Friends don't let friends extrapolate from early voting data. Original post

OK for just this week, gonna share our projected NFC standings from behind the paywall because they're kind of hilarious. *No NFC East or NFC South team projected for a winning record. *Every* NFC North and NFC West team projected for a winning record except the Cardinals. Original post

Repost by @NateSilver538 The most informative chart I've ever assembled regarding the midterms isn't a map — it's a comparison of Trump net approval and the average gas price. Repost of @NateSilver538: Click to see original repost

Eli and I ran the numbers on how big Republican redistricting gains will be in practice. Answer: IF current polls/forecasts hold, not very big at all. Maybe just a net gain of a seat or two after subtracting California and Utah. Original post

Just a slight tick of a rebound for Republicans on the generic ballot and Trump approval. Gas prices are down 15c from the recent peak, though oil futures are up again, so we'll see if that holds. Original post

Yeah, everything the labs are saying is, "We're not ready for RSI." It's not even clear they think we're ready for just slightly better models short of RSI. But also, they're trying to do RSI! It's inherently absurd. And dangerous. Repost of @DKThomp: Original post

Talarcio story right up there with "the Iran War will be over in a week" and "Joe Biden is sharp as a tack" for surfacing the most gullible people in the Milky Way galaxy. Original post

We need a total and complete shutdown of the NFC East until we can figure out what's going on. Original post

FLIPR update. The polls might well be wrong. There is, however, no persuasive evidence of a shift back toward Republicans at the moment. Original post

"Super Intelligence" is to Artifical Intelligence as "Latinx" is to Latino or Hispanic. Original post

Kind of striking how persistent the decline in Trump's approval rating has been. There's been mean reversion around news events and economic data, but to a declining rather than steady baseline as in the 1st term. Original post

The contrast between Peggy Flanagan and El-Sayed is interesting. They're both up ~3.5 points in our polling average. But since Minnesota is bluer, Flanagan is actually underperforming the state's partisan benchmark by more. Original post