Nate Silver

Nate Silver @NateSilver538

Still no nonpartisan polls of Nebraska's 2nd Congressional District, which could be important under certain scenarios (e.g. if Biden wins Arizona but loses Pennsylvania). Another D internal was just published though showing it Biden +6. https://t.co/1aV5LyfBG0 — PolitiTweet.org

Nate Silver @NateSilver538

@jbarro @conorsen @EconBerger One interesting theory is that house-district polls of the presidential race might be less subject to herding, because nobody gives a shit if you get the answer right (unless it's ME-2 or NE-2) so there are no perverse incentives. — PolitiTweet.org

Nate Silver @NateSilver538

RT @jonathanvswan: exactly. polling errors only tend to get penalized in one direction https://t.co/8bB8YG3V50 — PolitiTweet.org

Nate Silver @NateSilver538

At the risk of being slightly redundant... it's mid-September and Biden leads by 5+ points in states totaling 290 electoral votes (he needs 270 to win). That's not a foolproof position by any means, but it's a pretty decent one. There's a bit of give there. — PolitiTweet.org

Nate Silver @NateSilver538

@PollsterPatrick I mean literally the first sentence of that article is wrong. https://t.co/suvjQ8aDAu — PolitiTweet.org

Nate Silver @NateSilver538

Yeah, maybe I should have been clearer about that. Biden does have a little wiggle room here. These states give him 290 EV. So, e.g., he could lose PA (as long as he holds everything else) or lose MI + NH or WI + MN. https://t.co/JwQ8eqabNN — PolitiTweet.org

Nate Silver @NateSilver538

Anyway, I'm fine with almost every Trump Still Has A Chance take. (Because... he *does*). That was basically the headline on our feature when we launched our model! But don't pretend it's some sort of bold, contrarian take! It may be the most conventional opinion of all time. — PolitiTweet.org

Nate Silver @NateSilver538

But for the time being, most of Trump's not-inconsiderable hopes of winning re-election stem from tightening the race. Either tightening it to the point where the swing states are actually a toss-up, or at least enough that a *small* polling error (~2 points) would do the trick. — PolitiTweet.org

Nate Silver @NateSilver538

Biden could probably also survive a 3-point error, depending on where it's distributed. If it came in the *exact* same places as in 2016, it would be close, but it probably wouldn't be exactly the same. With a 4-5 point error (also possible! but less likely) that gets very tight. — PolitiTweet.org

Nate Silver @NateSilver538

So let's say by the time we get to Election Day, there's a 40% chance that polls are off by 3 or more points. It's probably somewhere in that vicinity. But half the time, they're off in Biden's direction, not Trump's. Thus, there's a 20% chance of a big error that helps Trump. — PolitiTweet.org

Nate Silver @NateSilver538

A point that Poll Guys (and Gals) could make a bit more forcefully is that while there indeed is reason to be concerned about polling error (and perhaps to think it could be higher this year) it's *very* hard to predict the direction of that error. It could favor Trump…or Biden. — PolitiTweet.org

Nate Silver @NateSilver538

Can't believe it took until Sept. 18 for someone to realize that the polls could be wrong again and Trump could win. Quite the canny observation. Completely undercuts both the mainstream media narrative and the Poll Guys, both of whom [checks notes] bring this up constantly. — PolitiTweet.org

Nate Silver @NateSilver538

So if you take that at face value, Biden's m.o. is basically "find 270 EV from Clinton states + MI + WI + PA + AZ + Nebraska-2 and anything else you win is a bonus". Not at all a bad position, but it does diminish the importance of FL and NC a bit. — PolitiTweet.org

Nate Silver @NateSilver538

The other feature of the polling landscape right now is the big gap between PA/AZ and Biden +5 and FL/NC at Biden +2/+1 with nothing in between. — PolitiTweet.org

Nate Silver @NateSilver538

Overall, our model is still showing a discrepancy here. Biden has lost a point or two in our national polling average since the conventions. But his state polling is roughly as strong as pre-convention. https://t.co/cy51vc5isJ https://t.co/g3LxlXk3PZ — PolitiTweet.org

Nate Silver @NateSilver538

Kind of an interesting mix of national polls today. On the surface, looks pretty good for Biden, but the AP-NORC poll is an exception (though note it's among adults, not registered/likely voters) and it's from a group of pollsters that have had somewhat pro-Biden house effects. https://t.co/UPkhXJcqpp — PolitiTweet.org

Nate Silver @NateSilver538

Once it became clear that COVID-19 was something we were going to be living with for *at least* a year or so, a lot of the public health advice needed to evolve to recognize that and it never really did, at least in the US. So people are left to figure out a lot on their own. — PolitiTweet.org

Nate Silver @NateSilver538

Listening the Ezra Klein podcast with Julia Marcus now and it's really good. The sort of conversation about managing COVID-19 risk, trade-offs, uncertainty, etc. in your everyday life that's been badly, badly lacking. https://t.co/iXo5hJRZT5 — PolitiTweet.org

Nate Silver @NateSilver538

@chrislhayes If Biden were polling better in PA, then maybe there’s a story where actually things are just going terribly for Trump everywhere but trends in FL happen to favor him. So maybe it’s PA that’s the hard one to explain. Then again, Biden was polling well in FL earlier this year. — PolitiTweet.org

Nate Silver @NateSilver538

“Biden’s doing super well in AZ *and* in WI/MN/MI but *not* FL or PA” is a story that’s a bit hard to reconcile if there are uniform swings among demographic groups. So either there’s some secular stuff going on in these states or the polling averages are noisy/wrong. — PolitiTweet.org

Nate Silver @NateSilver538

It’s definitely a little weird. Maybe there are secular trends in both states where AZ is going to do that CO/VA thing where it goes from red to blue without spending much time at purple, and maybe in FL the people who self-select to retire there are increasingly MAGA. But weird. — PolitiTweet.org

Nate Silver @NateSilver538

Obviously there's a lot going on here—other countries may have less COVID and/or better non-lockdown strategies (e.g. testing/tracing)—but there's sort of this myth that the US is uniquely cavalier when actually our control measures are on the stricter side vs. the average. — PolitiTweet.org

Nate Silver @NateSilver538

If you look at the 5 largest counties in Western Europe + the 2 that border the US, most had stricter lockdowns in March/April but they've mostly had *less* strict ones since then. https://t.co/1IbYVVMhl5 https://t.co/26Us2SAEyB — PolitiTweet.org

Nate Silver @NateSilver538

This gap had gotten as low as 1.4 points at points in June and July when Biden's polling was stronger in Florida. Although Biden has had plenty of strong state polls lately, they mostly haven't been in FL or PA, which are pretty important for this calculation. — PolitiTweet.org

Nate Silver @NateSilver538

Is the Electoral College / popular vote gap decreasing? Maybe a *bit* but our model still shows a 2.3-point gap (favoring Trump) between the popular vote and the tipping-point state margin in our average simulation. Still, that's smaller than 2.9 points earlier this month. — PolitiTweet.org

Nate Silver @NateSilver538

With Arizona polls looking strong for Biden lately and Florida polls looking weak (though he maintains a narrow lead), a person does wonder which camp Texas would fall into. — PolitiTweet.org

Nate Silver @NateSilver538

RT @onyxfish: Hey it's actually Friday this time instead of one of the many faux Friday's we've been having lately. We celebrated by launch… — PolitiTweet.org

Nate Silver @NateSilver538

It's *possible*, I suppose, that something different is happening in the swing states than elsewhere, perhaps because of Biden's advertising advantage. But ... usually this stuff is noise or reflects the mix of pollsters that happen to be active at any given time. — PolitiTweet.org

Nate Silver @NateSilver538

We don't want to over-emphasize small changes in our forecast and a Trump second term remains plausible, but Biden's win probability has crept up a bit. That reflects the combination of the passage of time and a steady stream of mostly-strong-for-Biden state polls. https://t.co/3md8BsSyaq — PolitiTweet.org

Nate Silver @NateSilver538

I don't want to encourage anyone to do national polls at the expense of state polls, but the national polls are a fairly weird mix right now. Trump has pulled within 6.8 points in our national polling average but his swing state polls have mostly been bad. https://t.co/cy51vc5isJ https://t.co/zJb0NaHEkT — PolitiTweet.org