Nate Silver

Nate Silver @NateSilver538

Maine is pretty similar in certain respects to Wisconsin/Minnesota, so not surprising to see Biden polling well there given his strong polling in WI/MN lately. — PolitiTweet.org

Nate Silver @NateSilver538

So, a bit of a mixed bag here. Trump's lead expanded to 20 points in KY from 9 in Quinnipiac's previous poll. But South Carolina is largely unchanged (Trump +6 vs +5). Biden has a huge lead in Maine (+21 ?!?) meanwhile. https://t.co/xc3vBTTr7v — PolitiTweet.org

Nate Silver @NateSilver538

Anyway, Quinnipiac should have polls of SC and KY out later today which can help us to test this theory. Again, I don't think it's super likely, but there's a lot of "dark matter" when comparing national and state polls because ~2/3 of Americans don't live in swing states. — PolitiTweet.org

Nate Silver @NateSilver538

...one alternative explanation is if Trump is making gains in red states, perhaps consolidating his base especially in states with no Biden ads running, which we wouldn't know very much about because most pollsters aren't polling, say, Idaho very much. — PolitiTweet.org

Nate Silver @NateSilver538

So, recently Trump has been getting some fairly awful results in swing state polls while national polls have tightened a bit. It is fairly likely that is noise, i.e. that Trump is getting a little unlucky in state polls but lucky in national polls and things will even out. But... — PolitiTweet.org

Nate Silver @NateSilver538

@BenjySarlin Trafalgar at least has some sort of "shy Trump" theory, but they haven't explained it too well and even if it holds, I'm not sure it's appropriate to apply to an IVR poll. Rasmussen just honestly doesn't give a shit, I don't think. — PolitiTweet.org

Nate Silver @NateSilver538

Of course, there is a market for explicitly Trump-leaning polls because of 2016 (despite the fact that their performance was um, iffy, in 2018). That's one reason why it's hard to predict in which direction the polls might be biased; polls are responsive to the previous election. — PolitiTweet.org

Nate Silver @NateSilver538

I'm really not into policing polls with strong house effects (that consistently lean toward one party). Good to have diverse approaches. But I do think it's bad if, rather than explaining why they have a house effect, the firms accuse anyone who points it out of being a shill. — PolitiTweet.org

Nate Silver @NateSilver538

You'll Never Believe Which Polling Firm Showed Trump Ahead In A National Poll — PolitiTweet.org

Nate Silver @NateSilver538

@kmedved Yeah, there's no special treatment. I agree that the panels violate some of the assumptions we make about how polls behave, although for practical purposes I don't think it should have much of an effect. — PolitiTweet.org

Nate Silver @NateSilver538

The USC poll, which was decently "bouncy" in both 2016 and so far this year despite constantly surveying the same panel of voters, is an interesting counterpoint to the "all polling movement is just nonresponse bias" crowd. — PolitiTweet.org

Nate Silver @NateSilver538

@mattyglesias @conorsen To bring this full circle, I think there are things it may make sense for governments to do for shorter periods of time that wouldn't make sense to do for an indefinite period, so the notion that "we'll still have to do this after a vaccine" can weaken the case for restrictions. — PolitiTweet.org

Nate Silver @NateSilver538

@conorsen @mattyglesias I think there are uncertainties that span an order of magnitude or so especially with respect to the rate of long-term complications short of death. But that's partly why I'm wary of the implication that people are being irrational. — PolitiTweet.org

Nate Silver @NateSilver538

@conorsen @mattyglesias Mo hot take is the modal American is being fairly rational and has internalized many of the major public health messages (outdoors safer, older people more at risk) without wanting to forgo certain things completely. Though individual rationality may not be optimal, societally. — PolitiTweet.org

Nate Silver @NateSilver538

Minnesota's position on the list of tipping-point states had already become a bit tenuous and this poll isn't exactly going to help its case. https://t.co/Elltwk6q0Q — PolitiTweet.org

Nate Silver @NateSilver538

@miatkowski @2AvSagas Maybe a high proportion of commutes are people who would once have taken the subway/Metro-North/LIRR but are now driving to work? — PolitiTweet.org

Nate Silver @NateSilver538

Perhaps more important, Biden could survive a 2016 (3-ish point in battleground states) polling error and probably hold on. Take 3 points off Biden's current polling margin in each state, and he's still ahead in PA/AZ (narrowly) and MI/WI/MN (more comfortably) and FL is a tossup. — PolitiTweet.org

Nate Silver @NateSilver538

@chrislhayes Yeah. There was that stretch in September, another in July around the time of the RNC, and then of course it tightened at the end. — PolitiTweet.org

Nate Silver @NateSilver538

Granted, you saw fewer of those leads in MI and WI than some others (they also weren't polled a ton). But still quite different from this year, where you'll get the occasional tie but it's rare to see Trump leads from high-quality polls in the half-dozen or so key battlegrounds. — PolitiTweet.org

Nate Silver @NateSilver538

People forget that there were several streches in 2016 (one of them was in mid-September) where the presidency polled as a very close race including polls that showed Trump leads in the tipping-point states fairly often. — PolitiTweet.org

Nate Silver @NateSilver538

@chrislhayes 1a. Pennsylvania's historically been redder than MI/WI/MN and so it's mean-reversion. 1b. More non-college white voters in MI/WI/MN than PA, probably Biden's greatest strength relative to Clinton (well, unless polls are wrong again). — PolitiTweet.org

Nate Silver @NateSilver538

* Though importantly, Europe is having far fewer deaths than in the spring, though it may just be a matter of time before that changes. — PolitiTweet.org

Nate Silver @NateSilver538

The fact that Europe is having a fairly big second COVID wave* and being pretty nonchalant about it should maybe cause people to revise their priors on how long Americans will tolerate restrictions. https://t.co/Au1AsbZzvY — PolitiTweet.org

Nate Silver @NateSilver538

We've had years (notably including 2016) where our model was more bullish on Republicans than the consensus, and others where it was more so on Democrats. With your polls, by contrast, it's always easy to predict what they'll say ahead of time. That's a really bad sign. — PolitiTweet.org

Nate Silver @NateSilver538

I was saying the polls are being sponsored by a partisan organization, not accusing *you* of being a partisan pollster, but it's a huge tell that your mind instantly goes to "agenda driven polling". — PolitiTweet.org

Nate Silver @NateSilver538

Internally the model also calculates a national snapshot which is basically where it thinks the race currently stands based mostly on *state* polling. We don't publish that number (maybe we should) but it looks like it's closer to Biden +8 than Biden +7 right now. — PolitiTweet.org

Nate Silver @NateSilver538

FWIW our forecast doesn't really use national polls for much. Their main purpose is for detecting a trend line, which is used to adjust polling averages in states with little recent polling. But with swing states having lots of polling lately, that adjustment doesn't matter much. — PolitiTweet.org

Nate Silver @NateSilver538

Biden's lead has slipped a bit in national polls to +7.0, but you'd be hard-pressed to find the same slippage in state polls and that's where most of the action has been lately. https://t.co/cy51vc5isJ — PolitiTweet.org

Nate Silver @NateSilver538

Yes, and this is often also true for lots of other things (see e.g. "cancel culture" or say views on Israel); pundits express concern on behalf of some hypothetical swing voter when it's actually their own concern. — PolitiTweet.org

Nate Silver @NateSilver538

The lagging state for Biden remains Pennsylvania. His polling has been pretty strong in MI, MN and WI since the early summer. But Pennsylvania is keeping the tipping point a couple of points to the right of his national numbers. — PolitiTweet.org