A study in the Lancet suggests 766,611 deaths in the US from Covid as of May 29.
I think their methodology makes sense. Maybe a better mathematician could help me assess it.
thelancet.com/journals/lanam…
I think their methodology makes sense. Maybe a better mathematician could help me assess it.
thelancet.com/journals/lanam…
What they're trying to do is improve upon the standard methods of calculating the death toll--either counting the death certificates or counting excess deaths.
They rightly note the limitations of both methods.
They rightly note the limitations of both methods.
(In the first case, Covid19 may not have been properly diagnosed; in the second, excess mortality might be owed to something other than the virus, e.g., lack of access to medical care owing to lockdowns or over-full ICUs.)
So they've tried to model it by matching death certificates to viral surveillance data, which makes sense.
I'm just not sure I understand the regression model they're using. Does it look right to people who do this sort of thing more often than I do?
I'm just not sure I understand the regression model they're using. Does it look right to people who do this sort of thing more often than I do?
If they're right, it's appalling.
Though once you get into these numbers, Stalin's remark about tragedy and statistics applies.
Though once you get into these numbers, Stalin's remark about tragedy and statistics applies.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
