Nate Silver

Nate Silver @NateSilver538

RT @jbview: This is the big one. Very hard to know what the information environment will be like whenever this happens - including who will… — PolitiTweet.org

Nate Silver @NateSilver538

@StevenTDennis I mean in the long run, I hope the distancing is *not* a long-term change and we can stop it as soon as possible. But, yeah, in the medium term, I hope we're still doing a lot of mask wearing and some limited but precisely targeted distancing around high-risk activities. — PolitiTweet.org

Nate Silver @NateSilver538

Among other things, the combination of some people being vaccinated, some people having had COVID previously (depending on how long immunity lasts) and perhaps some people having a degree of preexisting immunity because of other coronavirus would all count toward that threshold. — PolitiTweet.org

Nate Silver @NateSilver538

5. I think it's best to see a vaccine in combination with other interventions rather than a magic bullet. And I'd also keep in mind that the math on when we hit herd immunity is a bit complicated. https://t.co/2YOsC0idji — PolitiTweet.org

Nate Silver @NateSilver538

I'm not sure about all of the political and legal implications here. But if people see a vaccine as a ticket back to normal life, not just for society but also in a more literal way for themselves personally, that could encourage more vaccination. — PolitiTweet.org

Nate Silver @NateSilver538

4. It seems likely to me that many activities will *require* a vaccine, including many schools and some employers. Also many leisure activities: Want to attend a Knicks game? Fly to Cancun? Enter a nursing home? Show proof of a vaccine or a recent negative test. — PolitiTweet.org

Nate Silver @NateSilver538

3. In polls like these, people sometimes aren't very good at predicting their own future behavior under what for now is a hypothetical scenario. So it might be good look at empirical rates of vaccination instead to predict what % will ultimately be vaccinated. — PolitiTweet.org

Nate Silver @NateSilver538

2. Headlines on polls like these can sometimes be misleading by treating "not sure" responses as "no". You'll have a poll result like e.g. 40% yes, 20% no, 40% "need to learn more", and the headline will frame it as "only 40% will get vaccinated!" when really it's 40-80%. — PolitiTweet.org

Nate Silver @NateSilver538

That is to say, if we get to the point where **everyone who wants a vaccine can get one in reasonably short order**, then we can start thinking about how to persuade more people to be vaccinated. But getting to that first point may not be easy. — PolitiTweet.org

Nate Silver @NateSilver538

I have a few thoughts on this and similar polls that ask people if they'll get vaccinated. 1. In the medium run, the limiting factor is likely to be access and availability to the vaccine. — PolitiTweet.org

Nate Silver @NateSilver538

RT @jazzmyth: Now that sports are back, you can dive into this EXCELLENT @FiveThirtyEight interactive by @ryanabest diving into NBA player… — PolitiTweet.org

Nate Silver @NateSilver538

Well, that's a nice week-over-week decline in cases! Though deaths are still up from last Tuesday. I wouldn't take too much for granted yet. — PolitiTweet.org

Nate Silver @NateSilver538

US daily numbers via @COVID19Tracking: Newly reported deaths Today: 1,121 Yesterday: 428 One week ago (7/21): 1,036 Newly reported cases T: 54K Y: 55K 7/21: 63K Newly reported tests T: 733K Y: 761K 7/21: 750K Positive test rate T: 7.3% Y: 7.3% 7/21: 8.5% — PolitiTweet.org

Nate Silver @NateSilver538

I think part of it, too, is the pandemic warps people's sense of time (it certainly warps mine). November feels weirdly soon if you don't have the usual milestones and rhythms to ground yourself. But in political time, the final 99 days of the campaign are often an eternity. — PolitiTweet.org

Nate Silver @NateSilver538

So while Biden has a robust polling lead now, and a number of other things going for him, it's a little early to be all that confident about what is going to happen. — PolitiTweet.org

Nate Silver @NateSilver538

At the end of the day, it's only July, there are a lot of unprecedented events happening, polling is more accurate than its critics think but probably *not* as accurate as it was in say 2004-2012—and if we wind up in a photo finish, Trump likely wins b/c of the Electoral College. — PolitiTweet.org

Nate Silver @NateSilver538

.@aedwardslevy already stole my joke about likely boater models. — PolitiTweet.org

Nate Silver @NateSilver538

So we do show an Electoral College vs. popular vote gap that works in Trump's favor, although it's about half as large as in 2016. — PolitiTweet.org

Nate Silver @NateSilver538

In our snapshot, the tipping-point state is Pennsylvania (Biden +7.2). By comparison, we show Biden up 8.6 points nationally based on our snapshot. (Similar to his margin in national polls, though the snapshot mostly uses state polls.) — PolitiTweet.org

Nate Silver @NateSilver538

For comparison, here's our current *snapshot* in each state (our estimate of the result in an election held today) which combines the regression technique described below with polls, plus some mean reversion based on a state's partisan baseline. https://t.co/rugxLX9y7V — PolitiTweet.org

Nate Silver @NateSilver538

@gelliottmorris It's not about the implementation. I'm saying that MRP likely produces at best only marginal gains over techniques that rely only upon topline numbers. It just doesn't squeeze that much additional information out of your dataset. And it has a lot of drawbacks. — PolitiTweet.org

Nate Silver @NateSilver538

@gelliottmorris I am willing to bet that using topline polling data from *many* pollsters is *much* more accurate than using microdata from *one* pollster. And this will likely also lead to fewer issues with correlated errors, etc. — PolitiTweet.org

Nate Silver @NateSilver538

@gelliottmorris In fact a big issue with MRP is that it's liable to lead to highly correlated out-of-sample errors from state to state, which likely leads to overconfident probabilistic estimates unless you carefully account for that. — PolitiTweet.org

Nate Silver @NateSilver538

@gelliottmorris It was "less biased than the polls in 201"6 based on one pollster running numbers **only after the fact once the results were known** and that probably just reflected the fact that the one poll happened to do a better job of capturing Trump's % of the white working-class vote. — PolitiTweet.org

Nate Silver @NateSilver538

@gelliottmorris It was "less biased than the polls in 201"6 based on one pollster running numbers **only after the fact once the results were known** and that probably just reflected the fact that the one poll happened to do a better job of capturing Trump's are of the white working-class vote. — PolitiTweet.org

Nate Silver @NateSilver538

@gelliottmorris No, it's absolutely right. You can try to true-up a pollster's estimates if you don't think they accurately reflect the population, I suppose. But it's much easier and more robust to use an average of different polls, which MRP doesn't permit unless you have all the microdata. — PolitiTweet.org

Nate Silver @NateSilver538

Our estimates also reflect changes in voting access laws since 2016. That helps Biden in states like Virginia, Michigan, Ohio and Arizona but hurts him in Iowa, Georgia, Texas and Maine, among others. — PolitiTweet.org

Nate Silver @NateSilver538

This ^^ is also an issue for techniques like MRP (in fact, even more so, since MRP generally uses just one pollster's data). Still, it's fun to look at and does have *some* predictive accuracy. Among other things, clear that CO and VA are pretty solidly into blue territory now. — PolitiTweet.org

Nate Silver @NateSilver538

As you can see, this matches polling data well. Then again, it's supposed to do so. In fact one limitation of the this technique is that if there are biases in the polling (say, they underestimate Trump's standing with a certain group) they'll also be reflected in your estimates. — PolitiTweet.org

Nate Silver @NateSilver538

FWIW, here's a regression-based estimate of Biden's current standing in each state. This works by taking our state polling averages, and trying to figure out what differences there are from past elections based on regional or demographic variables (without overfitting the model). https://t.co/OEEGtBrQK6 — PolitiTweet.org