Read on Twitter
The 95% confidence levels of the UW model are so wide as to be worthless for planning, even if one buys its Procrustean guesses for assumptions about the effectiveness of social distancing policies with inconsistent compliance and differing efficacy in different contexts./1 https://twitter.com/alicia_smith19/status/1247140408125202433
With minimal knowledge, you can tell the IHME model is questionable simply because it pretends to predict deaths to FIVE significant digits, even though its range of possibilities isn’t even within a single order of magnitude. /2
Worst of all is that its assumptions seem to predict a virtual end to deaths in early May. Does that seem remotely plausible for an infection with asymptomatic contagion when we’re not implementing half the controls of a South Korea and there are hundreds of thousands infected?/3
@NateSilver538 isn’t an epidemiologist, but he surely sees these flaws? Why isn’t anyone pointing them out? We’d see through this nonsense immediately if IHME was making any other kind of prediction inconsistent w/ empirical data. Or is this Pilot Matt Damon’s “half hour” trick?
The IHME model is dangerously wrong. https://twitter.com/popwellzach/status/1247188932762251264?s=21 https://twitter.com/popwellzach/status/1247188932762251264
A friend much smarter than me points me to this explanation from IHME: https://www.medrxiv.org/content/10.1101/2020.03.27.20043752v1
"The cumulative death rate for each location is assumed to follow a parametrized Gaussian error function." Oy. /6
"The cumulative death rate for each location is assumed to follow a parametrized Gaussian error function." Oy. /6
The IHME model is assuming the lockdown in the US is as effective as Wuhan. Two things wrong with that:
1) It’s clearly not. Look at Orthodox Jewish gatherings in NYC or block parties in LA County.
2) The Wuhan statistics are likely fake. /7
1) It’s clearly not. Look at Orthodox Jewish gatherings in NYC or block parties in LA County.
2) The Wuhan statistics are likely fake. /7
Yeah, the Wuhan lockdown didn’t take Wuhan cases to zero—just their censored statistics. Why do we trust the IHME model that says that our (substantially less onerous) lockdown will give us the same results as fake Chinese statistics? https://twitter.com/onlyyoontv/status/1247839615668686850?s=21 https://twitter.com/onlyyoontv/status/1247839615668686850
