03-29-2020, 03:56 PM
It's a shame that they aren't reporting hospitalizations, because the only two really useful statistics are hospitalizations and deaths, and those are only useful when we can assess them over time. And for all data, we should be focusing on multi-day averages rather than day to day changes, because of processing and reporting delays, etc. And remember how inconsistent testing protocols are, even within jurisdictions. See these cogent Nate Silver tweets on this:
[tweet]https://twitter.com/NateSilver538/status/1243335453140553728?s=20[/tweet]
[tweet]https://twitter.com/NateSilver538/status/1243948150173380608?s=20[/tweet]
[tweet]https://twitter.com/NateSilver538/status/1243656263482949635?s=20[/tweet]
It's actually kind of horrifying to realize how blindly public health officials are having to fly, given limitations on testing. Unless and until "everyone who wants a test can get tested," the problems we confront to day will be those that were set in stone 3-4 weeks ago. The lag between exposure and hospitalization is generally at least two weeks, and exposure to death 3-4.
[tweet]https://twitter.com/NateSilver538/status/1243335453140553728?s=20[/tweet]
[tweet]https://twitter.com/NateSilver538/status/1243948150173380608?s=20[/tweet]
[tweet]https://twitter.com/NateSilver538/status/1243656263482949635?s=20[/tweet]
It's actually kind of horrifying to realize how blindly public health officials are having to fly, given limitations on testing. Unless and until "everyone who wants a test can get tested," the problems we confront to day will be those that were set in stone 3-4 weeks ago. The lag between exposure and hospitalization is generally at least two weeks, and exposure to death 3-4.
