Recently this paper was published purporting to show that staying at home does not prevent COVID-19 deaths I don't think the evidence provided supports that at all! Some peer-review on twitter 1/nhttps://twitter.com/jhnhellstrom/status/1368585541462208519 …
-
Show this thread
-
Replying to @GidMK
The biggest problem with this study is that it ignores time. When one evaluates the data over time one sees a clear pattern: a rise in infection rates drives people to stay at home which then drives a reduction in infection rates.
1 reply 0 retweets 16 likes -
Replying to @Hopozitzia @GidMK
Hmm, as far as I understand, they actually try to incorporate time by doing a regression over two time series (mobility, mortality). That's the new method they propose. I think that
@GidMK slightly mischaracterizes the author's method by not mentioning this.2 replies 0 retweets 1 like -
Replying to @RaphaelWimmer @GidMK
Their regression compares mobility and the *absolute* morbidity rate *at the same time*, which is irrelevant data. If you want to disprove the hypothesis that that staying at home reduces infection, you need to check...
1 reply 0 retweets 7 likes -
mobility against the *change* in the morbidity rate *after a period of time (~3 weeks)*. On general, to test a hypothesis you first need to understand it. The writers of this paper clearly did not.
3 replies 0 retweets 5 likes -
It's like a study that check if car breaks work by evaluating the correlation between the speed of a car at a given time and the fact that breaks are being applied. What you need to evaluate is if there is a *reduction* in speed *a few seconds after* breaks are applied.
1 reply 2 retweets 5 likes -
Replying to @Hopozitzia @GidMK
Yeah. I sent an example of such a case (country A with lockdown and near-zero deaths, country B without lockdown and many deaths) to the lead author because applying their model would result in p=.6 (no effect of lockdown measurable) - which is obviously problematic.
1 reply 0 retweets 4 likes -
He did not explain under which assumptions their model would actually detect an effect (i.e. avoid false negatives) but sent back a different example data set for which their model would actually detect an effect of lockdowns. But this data set does not resemble any real data.pic.twitter.com/wxLST9uRH6
1 reply 0 retweets 3 likes
Yeh wow I'm not sure that second graph is even possible
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.