Nate Silver

Nate Silver @NateSilver538

Dems' generic ballot lead has climbed from 6.1 points before the first debate to 7.3 points now. May not sound like a lot, but it's probably worth a half-dozen House seats and 1 Senate seat or something on that order. https://t.co/7mfucyRAFG — PolitiTweet.org

Nate Silver @NateSilver538

It's not great to reverse-cherry-pick by throwing out data that doesn't match the trend. On the other hand, there's this group of quasi-partisan pollsters that plainly behave like partisan polls even though they may or may not officially get classified that way by our model. — PolitiTweet.org

Nate Silver @NateSilver538

On average, Biden has gained 0.5 points in post-debate polls (median field date of Friday or later) vs. the previous editions of those same polls. However, he's gained 1.5 points if you exclude Rasmussen's national poll and Insider Advantage's PA poll. — PolitiTweet.org

Nate Silver @NateSilver538

@WesPegden Yeah. One interesting wrinkle is that Biden is actually a favorite conditional on losing PA but winning MI, WI and MN, where his lead is larger. He'd have to make up for PA with FL, AZ or NC (or more exotically, GA or TX). — PolitiTweet.org

Nate Silver @NateSilver538

FWIW, the current implementation of our model is sparing with how it uses national polls. You could argue for moving Biden's state polls upward slightly so that they better match the national polls. Our 2012 & 2016 models did that to some degree, but this year's does not really. — PolitiTweet.org

Nate Silver @NateSilver538

Why this state/national poll gap exists is an interesting question. Also there have been points in the year where it seemed to run in the opposite direction, e.g. before the first debate, our model thought Biden led by ~8 points from state polls vs. a ~7 point national poll lead. — PolitiTweet.org

Nate Silver @NateSilver538

Why is Biden +9.1 in our national poll average but +7.8 in our forecast? * The forecast still assumes just a teensy bit of tightening (about 0.4 points toward Trump) * The forecast is mostly based on state polls, which have been more consistent with an ~8 point lead than 9-10. — PolitiTweet.org

Nate Silver @NateSilver538

I would note that the gap is a little bit narrower in our *forecast*: Biden +5.1 in Pennsylvania vs. +7.8 in the national popular vote. https://t.co/ajG88SznSA — PolitiTweet.org

Nate Silver @NateSilver538

Our current trifecta chances: https://t.co/He5gzzliWS — PolitiTweet.org

Nate Silver @NateSilver538

@IChotiner In the abstract, I guess that's right (early vote = less potential for late swings) but I also have no idea if pollsters are accounting for the early vote properly which creates some additional uncertainty. — PolitiTweet.org

Nate Silver @NateSilver538

By contrast, our fancy model puts Trump's chances at 13%, in part because it also uses data from pre-1980 when polling errors were more common. But that's not so different than 10% and it's good to pass a sanity check like this. — PolitiTweet.org

Nate Silver @NateSilver538

Model gut check: * Biden leads by 5 points in PA, the most likely tipping point * In 2 of the past 10 elections, there's been a 5+ point swing between polls now and the final result. * Assume a 50/50 chance such a swing would favor Trump. * That gives Trump a 10% chance. — PolitiTweet.org

Nate Silver @NateSilver538

I guess that doesn't totally answer your question @kmedved but in general, the firms that are doing good enough work to diverge from the consensus *and have reason to think they're right about it* are the firms with good reputations that may not want to risk them. — PolitiTweet.org

Nate Silver @NateSilver538

3. Some high-prestige academic and media pollsters may be scared to publish a perceived outlier very late in the race, when they think it could hurt their reputation. For most of the year, these pollsters are the ones you trust NOT to herd. But sometimes their final polls are 🤔. — PolitiTweet.org

Nate Silver @NateSilver538

2. Some partisan and quasi-partisan pollsters seem to play a lot of games with the 538 and RCP averages. They don't want to stray *too* far from the average, but they'll often show results that are like the 538 or RCP average shifted by 2 points toward their side. — PolitiTweet.org

Nate Silver @NateSilver538

I think there are basically 3 groups of polls that herd. 1) Some (certainly not most or all) online or IVR polls with crappy raw data seem to look to live-caller polls for guidance. They tend to stay pretty close to the averages throughout the year. — PolitiTweet.org

Nate Silver @NateSilver538

Trump lead back up to 1.4 points in Texas after a couple of polls showing him ahead there today. https://t.co/nXOH1SNdLP https://t.co/CTT644rIB3 — PolitiTweet.org

Nate Silver @NateSilver538

@DanRosenheck Yeah that seems like a smart tactic. — PolitiTweet.org

Nate Silver @NateSilver538

What I'm saying is that we're probably too conservative in what we classify as a partisan poll. Or maybe we need different categories, e.g. "soft" partisan (say, PPP or Rasmussen not done for any outside client) and "hard" partisan (a campaign internal). — PolitiTweet.org

Nate Silver @NateSilver538

Because I don't want to make a bunch of ad-hoc, subjective decisions about which polls to include and how to include them when we're ingesting 40 polls a day. I want to have a good set of rules in place. https://t.co/gbeX593YCJ — PolitiTweet.org

Nate Silver @NateSilver538

Usually, manipulating the media narrative is what (some) partisan pollsters do. We include partisan polls in our averages but they're handled differently and weighed less. But there's a gray zone of firms that act like partisans even though they technically aren't by our rules. — PolitiTweet.org

Nate Silver @NateSilver538

And it's really the good faith part that matters. A firm that has a middling track record but I can trust to do honest work ... at least you can get some sense of direction from the poll. Different when the pollster seems to be engaged in a game to manipulate the media narrative. — PolitiTweet.org

Nate Silver @NateSilver538

I don't think it's really worth talking about individual polls from firms that have a poor track record and I wouldn't particularly trust to act in good faith. — PolitiTweet.org

Nate Silver @NateSilver538

RT @ameliatd: One finding from our survey worth highlighting: 54% of Black and 35% of Hispanic ppl think the Republicans don't want people… — PolitiTweet.org

Nate Silver @NateSilver538

What about the millions of Americans who *don't* vote? We think you'll learn a ton from our deep dive about them here: — PolitiTweet.org

Nate Silver @NateSilver538

Although it seemed like the race was tightening pre-debate that's more based on national polls than state polls. State polls didn't show the big 3/4 point bounce after the first debate that national polls did (maybe 1/2 pt instead). But they also showed fewer signs of tightening. — PolitiTweet.org

Nate Silver @NateSilver538

The consolation for Trump is that a fair bit of it predates the debate, although there's no particular reason to think the debate helped him. — PolitiTweet.org

Nate Silver @NateSilver538

A lot of good state polling for Biden over the past 48 hours, including YouGov & Gravis in the Midwest as well as state polls showing him (very narrowly) ahead in GA and TX. https://t.co/9AeTKX5o0O — PolitiTweet.org

Nate Silver @NateSilver538

@ASDem Yeah, it's the relentlessness of the tension that makes hockey great in person. Same is true for soccer I guess although my sample size of attending high-stakes soccer matches is small. — PolitiTweet.org

Nate Silver @NateSilver538

@ASDem Not saying it isn't good on TV! Just that I've never regretted going to a playoff hockey game in person, even with the worst seats imaginable, where there are some MLB and NBA playoff games where I've paid a lot for bad seats and thought I'd rather be at home. — PolitiTweet.org