2/n The study is here: https://dash.harvard.edu/handle/1/42669767 … It is...not great
-
-
Show this thread
-
3/n Basically, the authors looked at internet search trends for "cough" and "diarrhea" and some satellite photos of hospital parking lots, and concluded that higher-than-usual activity began BEFORE the COVID-19 outbreak officially happened Thus, COVID-19 started early!pic.twitter.com/TmFk49AEsp
Show this thread -
4/n This has spurred conspiracist fools (and some media sites...ahem) to declare that the Chinese government is hiding data, because obviously this ROBUST SCIENCE has proven things
Show this thread -
5/n But even THE MOST CURSORY EXAMINATION of the paper reveals that it is pretty hugely flawed Firstly, the magnitude of the findings How much higher was the activity in these hospital parking lots?
Show this thread -
6/n If you look at the regression curve, you can see that this huge increase in activity was actually not that impressive. It's a touch higher than the baseline, but not orders of magnitude biggerpic.twitter.com/ImrJX1l1Jm
Show this thread -
7/n Similarly, the vast increase in search terms for diarrhea only seems to start...well after the outbreak was officially acknowledgedpic.twitter.com/hDJPPgIXvy
Show this thread -
8/n So, at face value, there doesn't really appear to be much going on here But it gets worse The authors make this statement, and reference this paper It's one of the foundational arguments of their paper - diarrhea is a major symptom of COVID-19pic.twitter.com/6Dv3x8hIpd
Show this thread -
9/n But if you actually follow that link and check the paper out, you'll find that it showed that 35/204 COVID-19 patients had diarrhea, which is about 17%https://pubmed.ncbi.nlm.nih.gov/32287140/
Show this thread -
10/n More broadly, the current best evidence suggests that diarrhea is a RARE symptom of COVID-19, which makes tracking internet search terms for it a bit...problematicpic.twitter.com/sTKwz7jTot
Show this thread -
11/n It's also worth noting that these are really just vague correlations - there are SO MANY THINGS that could cause a few more cars in a select group of hospital parking lots and a few dozen extra queries about diarrhea on Baidu
Show this thread -
12/n It's hard to know what to make of the paper really, because all the authors are doing is drawing incredibly vague correlations and then suggesting a very unlikely outcome from them Not great, that
Show this thread -
13/n Really, what this study shows is that if you have enough time you can correlate basically anything It's definitely not proof of much
Show this thread -
14/n I'd say that the correlations found at Tyler Vigen's website are more likely to be accurate reflections of the truth than these very cursory connections https://www.tylervigen.com/spurious-correlations …
Show this thread
End of conversation
New conversation -
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.