Read on Twitter
this is not even an analysis.
it& #39;s a compilation of data tricks and mistaking models for reality.
it uses cases without adjusting for testing levels.
but worse, it uses a made up baseline. to see just how badly that baseline misleads, here are 3 US states
one hosted sturgis. https://twitter.com/SDSUCHEPS/status/1302480031638147074">https://twitter.com/SDSUCHEPS...
it& #39;s a compilation of data tricks and mistaking models for reality.
it uses cases without adjusting for testing levels.
but worse, it uses a made up baseline. to see just how badly that baseline misleads, here are 3 US states
one hosted sturgis. https://twitter.com/SDSUCHEPS/status/1302480031638147074">https://twitter.com/SDSUCHEPS...
can you guess which one it was?
all have similar latitude, population density, etc.
i have used "people currently in hospital" because cases are, frankly, a joke as a metric. false positives are high, non viable virus is mistaken for live, and low risk cases are a benefit.
all have similar latitude, population density, etc.
i have used "people currently in hospital" because cases are, frankly, a joke as a metric. false positives are high, non viable virus is mistaken for live, and low risk cases are a benefit.
and after all, this is supposed to be about preventing bad outcomes, sparing hospitals, and stopping deaths.
yet all these states saw hospitalizations rise. all had them rising even before sturgis.
so what& #39;s the claim that this "baseline" figure is meaningful?
yet all these states saw hospitalizations rise. all had them rising even before sturgis.
so what& #39;s the claim that this "baseline" figure is meaningful?
it seems to fall apart when faced with a real control group.
the blue and the green are nearly identical. they are also quite low. the US as a whole has 97 hospitalizations per million right now
NY state peaked at 966.
yellow shows the post sturgis spike and leads the others
the blue and the green are nearly identical. they are also quite low. the US as a whole has 97 hospitalizations per million right now
NY state peaked at 966.
yellow shows the post sturgis spike and leads the others
these 3 states combined have 309 hospitalizations. the US as a whole has 31,999.
SD has 78 people in hospital.
calling this a "superspread" is like mistaking the geico lizard for godzilla.
deaths from cov in SD were 46 in the month prior to sturgis and have been 29 since.
SD has 78 people in hospital.
calling this a "superspread" is like mistaking the geico lizard for godzilla.
deaths from cov in SD were 46 in the month prior to sturgis and have been 29 since.
this whole paper is just a collection of statistical tricks.
it& #39;s case data unadjusted for testing levels (despite a known round of testing in meade etc afterwards) used against a "synthetic control" which is utterly made up and fails to conform to like states
it& #39;s case data unadjusted for testing levels (despite a known round of testing in meade etc afterwards) used against a "synthetic control" which is utterly made up and fails to conform to like states
they snuck a model in as if it were data.
that& #39;s a dirty trick.
don& #39;t fall for it.
here& #39;s the key.
fun fact: ND and SD had the same number of hosp/mm on 8/8 and on 9/7.
and MT was the big loser, up 97% vs 66% in SD.
"synthetic control" loses to "real control"
that& #39;s a dirty trick.
don& #39;t fall for it.
here& #39;s the key.
fun fact: ND and SD had the same number of hosp/mm on 8/8 and on 9/7.
and MT was the big loser, up 97% vs 66% in SD.
"synthetic control" loses to "real control"
and for the love all all that is econometric, please spare me the "see, this was sturgis superspread!!!"
montana clearly led the way here and all were rising before sturgis even started, much less could have caused hospitalizations.
montana clearly led the way here and all were rising before sturgis even started, much less could have caused hospitalizations.
