Nate Silver

Nate Silver @NateSilver538

@SethS_D I suspect some non-COVID deaths may also be underreported because various parts of the system are closed or overwhelmed. — PolitiTweet.org

Nate Silver @NateSilver538

@SethS_D Less spread of influenza and likely other contagious diseases. — PolitiTweet.org

Nate Silver @NateSilver538

That study is definitely on the high end of the consensus range. But the consensus is that there's a *lot* of underreporting in Western countries and that the UK, which hasn't done much testing, is among the countries with the biggest underreporting problems. — PolitiTweet.org

Nate Silver @NateSilver538

A study today from Imperial College suggested that the UK has 35X to 160X more cases than it claims, which would be worse than what the UK alleges about China. https://t.co/IauogGfsED — PolitiTweet.org

Nate Silver @NateSilver538

@jbarro I think issue polling suggested that Biden was closer to the median Democrat on the issues, though. — PolitiTweet.org

Nate Silver @NateSilver538

If you're one of the epidemiologists I follow, send me a DM with your e-mail address and I will send you a copy. Will want your feedback though! — PolitiTweet.org

Nate Silver @NateSilver538

This was supposed to be a simple model (simple enough to do in EXCEL) but it's clear that you can get some extremely funky behavior where actual and detected cases don't line up very well depending on what assumptions you make about testing. — PolitiTweet.org

Nate Silver @NateSilver538

Working on something where you can model the number of *detected* cases of a disease as a function of the number of *actual* cases and various assumptions about how/how many tests are conducted. https://t.co/fFv2P4L6TO — PolitiTweet.org

Nate Silver @NateSilver538

Biden leads Sanders by 19 points in our national poll average. This is basically unchanged over the past several weeks. https://t.co/NRHApT7FSF https://t.co/VnplpxpRrK — PolitiTweet.org

Nate Silver @NateSilver538

RT @ClareMalone: New podcast about where we are in the Covid curve, why Trump’s approval rating is up and WTF is the DPS supposed to do htt… — PolitiTweet.org

Nate Silver @NateSilver538

Charts showing the number of deaths are likely useful on balance, even though deaths may be underreported or misreported. But charts showing the number of detected *cases*? No, not without a *lot* of context, when epidemiologists say reporting varies by 20x across countries. — PolitiTweet.org

Nate Silver @NateSilver538

To be blunter than Matt, some of these charts are fantastic examples of data visualization and pretty dubious examples of data science. — PolitiTweet.org

Nate Silver @NateSilver538

There's like a pretty strong correlation between the people who tweet about how the networks shouldn't carry Trump's press conferences are the people who tweet incessantly about Trump's press conferences. — PolitiTweet.org

Nate Silver @NateSilver538

@jeffhauser I'm not confident, but I would guess that the problems in underreporting deaths, while significant, are less bad than in underreporting total cases. — PolitiTweet.org

Nate Silver @NateSilver538

One caution to all the above is that reporting can sometimes be a little slower on weekends and coming out of weekends (i.e. on Mondays). Certainly an issue in some European countries. Less clear if it's true in the US. But would be good to wait until midweek. — PolitiTweet.org

Nate Silver @NateSilver538

New York is reporting fewer new cases than yesterday—and they reported fewer yesterday than Saturday—but their testing numbers have stagnated a bit and otherwise been a little weird lately. Mon: 6,984 Sun: 7,195 Sat: 7,683 Fri: 7,377 Thu: 6,447 https://t.co/FgB4yxFnGG — PolitiTweet.org

Nate Silver @NateSilver538

Deaths are not leveling off yet, unfortunately, but the rate of increase is slowing to doubling about every 4 days instead of every 3. Mon: 535 Sun: 465 Sat: 435 Fri: 367 Thu: 263 — PolitiTweet.org

Nate Silver @NateSilver538

Same pattern now for several days. Both reported cases and tests leveling off. New US detected cases: Mon: 21,469 Sun: 20,827 Sat: 18,821 Fri: 18,678 Thu: 16,807 New US reported tests: Mon: 114K Sun: 96K Sat: 109K Fri: 107K Thu: 98K — PolitiTweet.org

Nate Silver @NateSilver538

It's because people have to make up pretend reasons why the primary is still competitive. This would be 10x worse if not for coronavirus. — PolitiTweet.org

Nate Silver @NateSilver538

@JamesSurowiecki Also, interestingly enough Washington state seemed to handle things reasonably well in the first wave. It was the rest of the country that wasn't prepared enough for near-inevitable community spread. — PolitiTweet.org

Nate Silver @NateSilver538

@JamesSurowiecki I think we learned from Italy and Spain about the risk of overloaded hospital systems. — PolitiTweet.org

Nate Silver @NateSilver538

A better critique is: the US had a week or two more lead time than Europe and we were pathetically unable to take advantage of it. But strictly in terms of outcomes, it's not clear where we'll end up relative to say, other NATO countries. — PolitiTweet.org

Nate Silver @NateSilver538

It's not at all beyond question and people who say it's beyond question don't know what they're taking about. https://t.co/XyAyxYG2jm — PolitiTweet.org

Nate Silver @NateSilver538

Yep yep yep — PolitiTweet.org

Nate Silver @NateSilver538

Here's a different paper that suggests less under-reprtoing, though still a lot of it, and also with very wide variations from country to country (i.e. some countries are 20x better at case detection than others). https://t.co/aSy7gNw0EJ — PolitiTweet.org

Nate Silver @NateSilver538

This is somewhat encouraging to see in Louisiana / New Orleans. It's been very warm down there so this might also be interesting data for people studying weather effects. — PolitiTweet.org

Nate Silver @NateSilver538

Or put some uncertainty bars on the estimates and see how pretty your graphs look then LOL. — PolitiTweet.org

Nate Silver @NateSilver538

Don't make pretty graphs with the data either. That just obscures how messy it is. — PolitiTweet.org

Nate Silver @NateSilver538

If detection rates vary that much than it is NOT at all reliable to use the number of detected cases to compare countries (or states). Y'all really need to stop doing that until/unless you have a sense for how effective different places are at testing/detection. — PolitiTweet.org

Nate Silver @NateSilver538

Also: these folks estimate that European countries are variously detecting anywhere between 1.2% (the UK and Spain) and 20% (Norway) of their actual cases. That is, there is nearly a 20-fold difference in detection rates. — PolitiTweet.org