Nate Silver

Nate Silver @NateSilver538

@youyanggu @COVID19Tracking I wish states broke testing down into different tranches, e.g. "in the hospital", "have symptoms", "someone I know is positive", "was just curious". — PolitiTweet.org

Nate Silver @NateSilver538

@youyanggu @COVID19Tracking But less clear when the availability of tests is ~constant and demand for testing may itself be a sign of increasing prevalence. — PolitiTweet.org

Nate Silver @NateSilver538

@youyanggu @COVID19Tracking Yeah, it's tricky. Ideally, we'd know more about who is getting tested. We built a toy model for some of this stuff here. I think clearly when there's a very rapid increase in testing (e.g. the US in March), you're better off looking at the + rate... https://t.co/VK7rCgBNMc — PolitiTweet.org

Nate Silver @NateSilver538

Trump not exactly on the side of public opinion on mail voting. https://t.co/03bD5d5ygS https://t.co/roNOZPIr1D — PolitiTweet.org

Nate Silver @NateSilver538

@youyanggu @COVID19Tracking I think there are a number of practical circumstances where calibrating off of the number of positives could meaningfully overestimate R. Though calibrating off the positive test rate may underestimate it in some of those cases, too. — PolitiTweet.org

Nate Silver @NateSilver538

@youyanggu @COVID19Tracking Yeah, worth watching. I'm still a big fan of the positive test rate vs. the raw cases number. But, it feels much safer/nicer when everything is pointing down. — PolitiTweet.org

Nate Silver @NateSilver538

So this report looks OK. But keep in mind that if re-opening erodes the progress we're seeing, it probably WON'T be with a big spike, at least not at first. It might start with declines becoming plateaus, and plateaus becoming slight increases. So follow the numbers carefully. — PolitiTweet.org

Nate Silver @NateSilver538

Not a ton to say about this one. We continue to see death tolls decline week over week. The number of cases has been flat lately instead of declining, however... but testing has increased a lot so the positive test rate remains in decline (though a bit higher today than Wed.) — PolitiTweet.org

Nate Silver @NateSilver538

US daily numbers via @COVID19Tracking Newly reported deaths Today: 1,513 Yesterday: 1,402 One week ago (5/14): 1,908 Newly reported cases T: 25K Y: 22K 5/14: 27K Newly reported tests T: 408K Y: 413K 5/14: 382K Positive test rate T: 6% Y: 5% 5/14: 7% — PolitiTweet.org

Nate Silver @NateSilver538

@conorsen Run Klobuchar on the ballot in the purple states and Warren in the blue states. — PolitiTweet.org

Nate Silver @NateSilver538

@conorsen And it culminates with Biden picking co-VPs. — PolitiTweet.org

Nate Silver @NateSilver538

@IChotiner Might agree with you ordinarily, but I think there may be more scrutiny on the VP pick this year given Biden's age + the fact that we're in the midst of a pandemic. — PolitiTweet.org

Nate Silver @NateSilver538

As a corollary, projections that model the impact of re-opening *without* accounting for the other factors (e.g. mask-wearing, weather, increased testing/tracing) may contribute to the problem if they wind up looking overly pessimistic later. So maybe be a bit careful with those. — PolitiTweet.org

Nate Silver @NateSilver538

One thing I worry about, if cases remain steady/declining, is people concluding "re-opening doesn't raise transmission" when the more likely conclusion is "re-opening does raise transmission, but is being counteracted by other factors that lower transmission". — PolitiTweet.org

Nate Silver @NateSilver538

@MarcusHUSA It depends on the state, but, yeah that's another ambiguity. Some states are counting tests and some are counting people. — PolitiTweet.org

Nate Silver @NateSilver538

So it's interesting that faced with high-stakes, long-term decisions, colleges are accounting for weather after all. And while there are many uncertainties, I share their worry that the late fall/winter could be a rude awakening if weather effects are strong. — PolitiTweet.org

Nate Silver @NateSilver538

As I've noted here before, the communication on weather has not been great. A lot of public health folks seem to downplay weather so as not to give people a false sense of security. But if you read the research it generally *does* show some decent evidence of weather effects. — PolitiTweet.org

Nate Silver @NateSilver538

A lot of colleges and universities seem to be expecting fairly strong weather/seasonal effects for COVID-19, often citing the potential for a second wave that might begin in November or December. https://t.co/zb3yMJYFKR https://t.co/2KdpTmnLq7 — PolitiTweet.org

Nate Silver @NateSilver538

It's likely that Sarah Palin was worth a lot to McCain in Alaska (this method would estimate ~6.7 points). Polls had actually shown a close-ish race there before she was chosen. But picking a VP from a big state like say CA or TX or FL likely won't give you that kind of bounce. — PolitiTweet.org

Nate Silver @NateSilver538

A good rule of thumb is that a VP is worth a net of 20/EV percentage points in their home state, where EV is the number of electoral votes. So, e.g. Pence is worth 20/11 = ~1.8 points to Trump in Indiana. Or Whitmer would be worth 20/16 = ~1.3 points to Biden in Michigan. — PolitiTweet.org

Nate Silver @NateSilver538

Re-doing some old research on VP picks and one thing that's pretty clear is that there's a bigger home-state effect if the VP comes from a smaller state. The same holds for presidential candidates themselves. — PolitiTweet.org

Nate Silver @NateSilver538

@SeanTrende Unless states are only listing their antibody negatives but not their antibody positives, which I don't *think* is the case in most places but may be the case in Georgia, which apparently treats antibody positives as probable cases. https://t.co/G0N86TR36r — PolitiTweet.org

Nate Silver @NateSilver538

@thecity2 Yeah, see the next tweet. There are potentially effects that run in opposite directions and it's not clear which one would win out. — PolitiTweet.org

Nate Silver @NateSilver538

Anyway, states and the CDC need to stop this. Just list viral tests and antibody tests separately. And guidelines for re-opening need to make clear that they're based on viral tests. — PolitiTweet.org

Nate Silver @NateSilver538

Complicating this is that antibody tests may be conducted among currently healthy people (though there may be some self-selection toward people who thought they were exposed in the past), whereas viral tests are mostly conducted among people who currently have symptoms/exposure. — PolitiTweet.org

Nate Silver @NateSilver538

@kmedved Yeah, I'm not convinced it skews the data *that* much especially if some states are including antibody positives as cases also. But it might be worth it for @COVID19Tracking to publish a separate summary that excludes cases/tests from states that don't separate them out. — PolitiTweet.org

Nate Silver @NateSilver538

@kmedved Yeah, that would be really bad if states are counting the negatives but not the positives. — PolitiTweet.org

Nate Silver @NateSilver538

It's ambiguous what including antibody tests will do to the positive test RATE. Since they measure, more or less, whether you EVER had the disease, they may produce a *higher* positive % than viral tests, which measure if you CURRENTLY have it. See e.g.: https://t.co/1oSb5eV4XL https://t.co/6f16rh8mpp — PolitiTweet.org

Nate Silver @NateSilver538

Note though that while including antibody tests will artificially inflate the number of tests, it will *also* artificially inflate the number of *cases*, since antibodies are a sign of someone who once had the disease but is recovering and is probably no longer contagious. — PolitiTweet.org

Nate Silver @NateSilver538

States shouldn't be mixing PCR/viral tests (tests of whether someone *currently* had the virus) with antibody tests (tests of whether they *previously* had it) in their data. That just makes it a lot harder to tell what's going on. Don't do this, states. https://t.co/RNfloQpmPI — PolitiTweet.org