This is total BS. The study DOES NOT show masks have no significant effect.

It tells you: "wearing a mask in your garden (but not a home) will not prevent you from catching COVID at home."

Statistically, the study is junk science. It wasn't easy to publish because of the flaw.
Actually Heneghan DID post false information: the study DO NOT make that claim.
I am very surprised that someone can be a professor of "evidence b" & be ignorant basic statistical methodolody plus have a reading comprehension problem.


See for yourself: @carlHeneghan, "evidence based prof", is SHAMELESSLY misrepresenting the paper, which explicitly claims it has no result. (Aside from severe flaws in the paper itself).

I agree the authors (unlike the propagandists) hedged their claims. So, aside from the fact that infections tend to happen at home,
here is my junk science accusation: Can an infection rate of ~2% be compatible w/properties below?

He was my student at Courant (Math Finance). And he is a math virtuoso.

He is also the nephew of the French translator of Dino Buzzati's Deserto. So I may be biased.
The only thing I would warn against is the false positive paradox, when the prevalence of the disease is lower than than the false positive rate. One of those nice Bayesian counterintuitive thingys.

• • •

Missing some Tweet in this thread? You can try to force a refresh

Keep Current with Nassim Nicholas Taleb

Nassim Nicholas Taleb Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!


Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @nntaleb

23 Nov
Preliminary comments on the DANISH STUDY.
Tells you nothing about masks.
Tells you something about methodology using fortune cookie randomized control studies methodology. ImageImage
Some math tinkering this morning
Correcting the study results as it itself claims. Not counting the true odds ratio off outside infections as comparative numbers. Image
Add to that the p-value hacking.
David Salazar has a great explanation.

Read 4 tweets
19 Nov
I am in favor of masks & intransigent w/Twitter unrigorous babble; but one must ethically try to have a fresh look when facing a new study.

Let's do an objective analysis of this study to see if has any merit, if it can change one's mind, etc.
2) For starters the paper does'nt make the claim by "James Todaro, MD", that masks were NOT effective.

Worse, there is in fact a big problem w/the study: it is ONLY saying:

"If you have exposure to COVID inside your house, wearing a mask in the garden won't protect you".
3) Aside from some statistical problems, this is junk.

A proper study should be:
"What is the risk reduction for those catching COVID OUTSIDE their house" by eliminating at home infections, which is the BULK of the spreading.
Read 4 tweets
10 Nov

Bendford's law requires unconditionality. Conditional on an outcome (win), the first digit must be less frequent than otherwise.

Conditional on the average human being ~ 5+ f, the first digit (1 ft) needs to be less frequent than 5f.

Be careful. be very careful.
BENFORD-NEWCOMB'S LAW. Be careful. Probability is not a fortune cookie enterprise! Image
Sorry, typo in previous (changed the variance not the mean but STILL illustrative). This should be more convincing. Image
Read 4 tweets
9 Nov
Chris too is (was) an option trader. To divert from 45/55 you need a looooooooooot of certainties not met with such elections.
@NateSilver538, to repeat, does not understand probability. Statistics is not his thing.
2) This explains the BS with
3) And Silver understood his own BS on election night

Read 4 tweets
2 Nov
There are a lot of discussions circulating about our problem with 538's and Silver's computation of "probabilities". Most fail to get the point. This is OUR presentation.

This comment by Edward Yu is WRONG: I am saying that:
"It is inconsistent to say: my forecast is 85% AND I may change my mind IF new information comes in."

Yu is also clueless abt Silver's performance & the very definition of Bayesian inference.
And old Black Swan Man... I've lost 21 lbs since!

Read 4 tweets
1 Nov
Ce n'est pas vrai, monsieur le président: j'ai d'ailleurs fait la même erreur.
Armeniens-Mao-Staline-Khmers-Hitler-Corée du N.
Le nationalisme laïc tue beaucoup, beaucoup plus que la religion.
Et la Vendée, bien sur, très proche de l'origine du régime actuel.
Sauf le respect que je vous dois, bien entendu.
Read 5 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!