Nate Silver

Nate Silver @NateSilver538

Maybe not, but I think Texas depends to some degree on what's happening in Florida, with which it's correlated in some ways but not others (much younger). You can imagine a world (perhaps not this one!) in which TX is a better target than FL for a Dem. https://t.co/4xjUeBRD98 — PolitiTweet.org

Nate Silver @NateSilver538

Still, you need to swallow your pride and prepare for which states matter if the election tightens. For Trump, that means conceding the map could look real bad if Biden's current margin holds. For Biden, it means not taking swingy tipping-point states like MI and WI for granted. — PolitiTweet.org

Nate Silver @NateSilver538

After all, any election where the race is competitive in November requires a substantial change from today, with Biden up +2 or +3 (or less) instead of +9. If there's a big shift back toward Trump, it may not be uniform across states. — PolitiTweet.org

Nate Silver @NateSilver538

Maybe Texas is different because it's *so* many Electoral Votes and it's a bit hard to predict (historically low turnout but that may be changing, plus the COVID situation there) that it's worth taking a flier on. And I do think campaigns need plan for a multitude of scenarios. — PolitiTweet.org

Nate Silver @NateSilver538

So Biden can win GA. But in the scenarios where he wins GA, he'll probably already have won FL and/or NC, meaning the Electoral College may already have become a lost cause for Trump. — PolitiTweet.org

Nate Silver @NateSilver538

Georgia is close right now, as are TX, OH and IA. But, those states aren't Trump's biggest problems. Instead, it's that he's way behind in MI, WI, PA and FL and to a lesser extent AZ and NC. Those states give Biden plenty of winning options. https://t.co/VYrvEGxHXu https://t.co/4If3fkFpiS — PolitiTweet.org

Nate Silver @NateSilver538

One urge the campaigns will need to resist is investing in the CLOSEST STATES right now (given that Biden is well ahead currently) as opposed to the TIPPING POINT STATES that would be decisive if the election tightens. — PolitiTweet.org

Nate Silver @NateSilver538

If you're in the US and have gotten a COVID-19 test within the past month, how long did it take you to get the results? — PolitiTweet.org

Nate Silver @NateSilver538

@joshtpm @ibflorea Yeah, that's pretty hard to know. And also, there are uncertainties around both what true seropositivity rates are and exactly how large an effect they have. But in general I think it's good news that the same places haven't gotten hit hard twice and will be bad if that changes. — PolitiTweet.org

Nate Silver @NateSilver538

@joshtpm @ibflorea It's really a mix of everything at this point. But, medium-density cities that *didn't* get hit hard in March/April seem to be particularly having issues right now, and that describes a lot of places in Texas and the Sunbelt. — PolitiTweet.org

Nate Silver @NateSilver538

@laurabronner Yeah for sure. Fortunately, I know both the more common and the less common symptoms so that gives me more to be worried about! — PolitiTweet.org

Nate Silver @NateSilver538

RT @nataliexdean: @NateSilver538 We have a word for this. Specific (in contrast to non-specific symptoms). In surveillance, we talk about t… — PolitiTweet.org

Nate Silver @NateSilver538

@thecity2 "What symptom would have the highest coefficient in a logit model" basically. — PolitiTweet.org

Nate Silver @NateSilver538

e.g. Loss of smell (anosmia) isn't that common a symptom of COVID, but it's very distinctive, whereas a headache is common but not distinctive. So if you have one person with a headache and another with anosmia, the person with anosmia might be the more likely one to have COVID. — PolitiTweet.org

Nate Silver @NateSilver538

One thing that would be helpful here would be data on the most *distinctive* symptoms of COVID rather than the most *common* ones. — PolitiTweet.org

Nate Silver @NateSilver538

At the same time, there's some uncertainty here. The spread does seem to be among younger people than before, which could matter. I don't think it's worth speculating too much beyond that for now. — PolitiTweet.org

Nate Silver @NateSilver538

And the *really* big increases in case numbers have only come in the last several days, so I'd say to wait 2-3 weeks before we take too much for granted. — PolitiTweet.org

Nate Silver @NateSilver538

When might you expect to see increases in deaths? Earlier on in the pandemic, there was roughly a 7-day lag between spikes in cases and spikes in deaths. But, that lag may be longer now because there's more testing, meaning that cases are detected earlier. — PolitiTweet.org

Nate Silver @NateSilver538

On deaths, the number today (619) was similar to last Friday (647). Note that the big number reported yesterday (†) reflected a change in reporting standards in New Jersey that affected past cases rather than a sudden spike there. — PolitiTweet.org

Nate Silver @NateSilver538

Cases have increased from 23,481 two Fridays ago to a record 44,373 this Friday. That's a big-time increase, obviously. And it's mostly NOT attributable to more testing, with the positive test rate having increased from 4-5% before to more like 6-8% now. — PolitiTweet.org

Nate Silver @NateSilver538

US daily numbers via @COVID19Tracking: Newly reported deaths Today: 619 Yesterday: 646 +1854† One week ago (6/19): 647 Newly reported cases T: 44K* Y: 39K 6/19: 31K Newly reported tests T: 603K Y: 638K 6/19: 571K Positive test rate T: 7.4% Y: 6.1% 6/19: 5.4% * record — PolitiTweet.org

Nate Silver @NateSilver538

Pretty sharp rebound in concern around coronavirus, which I think has to count as good news if it makes people more careful going forward. — PolitiTweet.org

Nate Silver @NateSilver538

@VanessaSelbst It's also not super easy to find data on how much time people spend indoors vs. outdoors in different parts of the world. — PolitiTweet.org

Nate Silver @NateSilver538

@VanessaSelbst In fact, I suspect that some journal articles on weather got screwed up because they were mistaking weather effects for indoor/outdoor effects based on looking at data in March/April. Warmer weather seems to help to a degree, but being indoors b/c it's too hot may outweigh that. — PolitiTweet.org

Nate Silver @NateSilver538

@VanessaSelbst I've noodled around with some of this myself, and, yeah, there seems to be something there. Not aware of much peer-reviewed work, though, that's tried to disentangle indoor/outdoor effects from weather effects. — PolitiTweet.org

Nate Silver @NateSilver538

And second, the conversation around social distancing changed a lot in early June with the protests and Trump making plans to resume his rallies. And COVID was no longer the lead story. Not blaming anyone here. But the timing is pertinent if people felt like "lockdowns are over". — PolitiTweet.org

Nate Silver @NateSilver538

What other stuff? Two things seem worth pointing out. First, there seems to be some correlation with greater spread in states where it's hot and people are spending more time indoors with the AC on. That *is* a bit nonlinear; there's much more demand for AC in June than May. https://t.co/nf9lL92uUI — PolitiTweet.org

Nate Silver @NateSilver538

Obviously some of that gets to the nature of exponential growth. An R of 1.3 isn't *that* different than an R of 1.1, but played out over a few weeks, it makes a lot of difference. Still, a more complete story probably includes premature re-openings coupled with other stuff. — PolitiTweet.org

Nate Silver @NateSilver538

Something to think about: re-openings have been occurring gradually since late April in different states/counties. If you had a metric averaging out how open different states are, it would likely show a fairly linear pattern. So why is there a nonlinear increase in cases now? — PolitiTweet.org

Nate Silver @NateSilver538

This is misleading. As he says at the end of the thread, the spike in deaths yesterday was caused by New Jersey now listing "probable deaths" in its official totals. Outside of that deaths are still falling, though there's a good chance IMO they'll begin rising again later. — PolitiTweet.org