Nate Silver

Nate Silver @NateSilver538

Pollsters are also using lots of new techniques and the results are decidedly... mixed. Some new methods are getting comparatively good results but some others just aren't; you have to go on a case-by-case basis. https://t.co/F8jvuuusPk — PolitiTweet.org

Nate Silver @NateSilver538

Overall, the average error in primary polls was fairly high ... it's always high in the primaries, but was toward the high end of high this year. However, the unusually rapid shifts in the race caused pollsters problems. They did relatively well outside of SC and Super Tuesday. https://t.co/19dn6bVd9r — PolitiTweet.org

Nate Silver @NateSilver538

And you can see the ratings themselves here: https://t.co/HBex5M5zFm — PolitiTweet.org

Nate Silver @NateSilver538

Hey folks! We've updated our pollster ratings with results from the 2020 primaries! Overview here: https://t.co/ees0U6j4L7 — PolitiTweet.org

Nate Silver @NateSilver538

@gelliottmorris "Entirely" is tricky because cases may be declining, but less quickly than they would be without the reopening. — PolitiTweet.org

Nate Silver @NateSilver538

One plausible future is that small-scale stuff (restaurants, parks, gyms, retail) doesn't actually cause *that* much spread but large-scale stuff (colleges, work, travel, mass gatherings) does, and people make incorrect inferences about the large stuff if the small stuff goes OK. — PolitiTweet.org

Nate Silver @NateSilver538

@mattyglesias There's also interregional variability. If dense urban areas have a much higher R0 and also cases so far are concentrated disproportionately in dense urban areas, then ~5% national seroprevalence could go slightly further than you'd think. It's probably not a huge effect, though. — PolitiTweet.org

Nate Silver @NateSilver538

With that said, there will come a point at which some folks may need to re-examine their priors. Are we there yet? I'd say "no". But I also don't think it's a bright line. A better approach might be to have relatively weak/flexible priors and accept a high degree of uncertainty. — PolitiTweet.org

Nate Silver @NateSilver538

Remember, all these data series are lagging, so we're always looking a couple of weeks into the past. If there are recent changes that have made things worse (i.e. less distancing) they may not be showing up in the data yet. — PolitiTweet.org

Nate Silver @NateSilver538

The 7-day rolling average, which should avoid issues stemming from day-of-week effects, is 1,362 deaths, as compared to 1,761 a week ago. The peak (so far) for the rolling average was 2,070 on April 21. — PolitiTweet.org

Nate Silver @NateSilver538

Mondays are typically quite slow for reporting deaths because of weekend lags, so tomorrow's numbers are very likely to be worse. With that said, 786 is the lowest number of reported deaths in a day since March 30. — PolitiTweet.org

Nate Silver @NateSilver538

US daily numbers via @COVID19Tracking: Newly reported deaths Today: 786 Yesterday: 839 One week ago (5/11): 837 Newly reported cases T: 20K Y: 21K 5/11: 18K Newly reported tests T: 346K Y: 411K 5/11: 388K Positive test rate T: 6% Y: 5% 5/11: 5% — PolitiTweet.org

Nate Silver @NateSilver538

@SethPartnow Is he... his generation's Klay Thompson 😬? Also a tough guy to rate. — PolitiTweet.org

Nate Silver @NateSilver538

@yeselson @mattyglesias While there are a *lot* of questions about how well they're tracking cases, the apparent absence of catastrophic outbreaks in India, Africa and Southeast Asia (Singapore somewhat excepted) makes weather seem as though it could be decently important. — PolitiTweet.org

Nate Silver @NateSilver538

@mattyglesias 5—Greater access to testing (and some degree of contact tracing) matters a lot 6—Limiting large public gatherings, travel and school/work matters a lot even as other things return — PolitiTweet.org

Nate Silver @NateSilver538

@mattyglesias Theories, all plausible but unproven: 1—People are being cautious and the types of activities they're engaging in are relatively safe 2—Masks matter a lot 3—Weather matters a lot 4—Something about immunity we don't totally understand (e.g. maybe super-spreaders already got it) — PolitiTweet.org

Nate Silver @NateSilver538

@mattyglesias That is indeed a more coherent version but like its less coherent cousins is likely at odds with why Trump actually behaves the way he does and attributes too much coherence to Trump. — PolitiTweet.org

Nate Silver @NateSilver538

I continue to doubt the implication of a certain influential parcel of the conventional wisdom that tens or hundreds of thousands of additional Americans dying would be good for Trump's re-election prospects. — PolitiTweet.org

Nate Silver @NateSilver538

RT @NYCMayorsOffice: ATTENTION NEW YORKERS: There is a blood shortage right now. Please donate if you can — it’s safe and will save lives.… — PolitiTweet.org

Nate Silver @NateSilver538

RT @thehowie: #NewYorkTough : New record low positive rate & record high (for Sunday) testing. NYC also with vast improvement. @David_Eaton… — PolitiTweet.org

Nate Silver @NateSilver538

It is still a bit too soon to be all that confident about how much relaxed social distancing (both formal relaxation of stay-at-home rules and people just being out and about more) will affect the numbers. — PolitiTweet.org

Nate Silver @NateSilver538

Sundays are typically pretty favorable days, with lower numbers of deaths reported than during midweek. But this is nonetheless an encouraging report. The most tests ever in a day! 2nd-lowest positive test rate in a day. 2nd-lowest deaths reported in a day since March. — PolitiTweet.org

Nate Silver @NateSilver538

US daily numbers via @COVID19Tracking: Newly reported deaths Today: 839 Yesterday: 1,286 One week ago (5/10): 979 Newly reported cases T: 21K Y: 25K 5/10: 22K Newly reported tests T: 422K Y: 357K 5/10: 274K Positive test rate T: 5% Y: 7% 5/10: 8% — PolitiTweet.org

Nate Silver @NateSilver538

The point is: this is a long game. COVID-19 will be with us for a while. Maybe if we're very smart/lucky, it will have become less of a problem by the fall or by next spring. But those are optimistic scenarios. It could take years. So maintaining reader trust will be essential. — PolitiTweet.org

Nate Silver @NateSilver538

Maybe those headlines will turn out to be true. There are lags, so let's definitely give it another week or two. But *so far* the evidence is mixed. And it's easy enough to see a path where those dramatic predictions/headlines backfire and wind up reducing support for distancing. — PolitiTweet.org

Nate Silver @NateSilver538

One salient example is dramatic headlines and predictions put out around re-opening plans in states such as GA and TX a few weeks ago, which sometimes seemed written to bolster support for maintaining distancing or to scare these states into shape. https://t.co/gYHOWqOc1V — PolitiTweet.org

Nate Silver @NateSilver538

But also, it's fairly common for attempts to influence behavior to backfire and sometimes to produce the opposite behavior from what was intended. Maybe I'll do a separate thread sometime this week with examples of this. — PolitiTweet.org

Nate Silver @NateSilver538

Or if you DO want to influence behavior, make that transparent. Nothing wrong IMO with writing an editorial saying "HERE'S WHY YOU SHOULD WEAR A MASK". Just make clear that's what you're doing, instead of infusing it into your reporting of straight news stories. — PolitiTweet.org

Nate Silver @NateSilver538

For one thing, that is simply the professional obligation of journalists, in my view. I'm a traditionalist in this respect. Report the facts. Pay attention to the quality of your writing (e.g. avoid confusing jargon). But don't try to influence readers' behavior. — PolitiTweet.org

Nate Silver @NateSilver538

3. Finally—I think the media spends too much time worrying about knock-off effects of their coverage, i.e. worrying about scaring people or about making them complacent, and should instead try to report the facts as straightforwardly as possible, including the many uncertainties. — PolitiTweet.org