It is sad. @DrJBhattarcharya is the worst example I have personally seen of someone who was previously a scholar but who now engages in repeated misrepresentation of scientific results to serve a partisan agenda.
https://twitter.com/gregggonsalves/status/1455238042646626313
I recently testified pro bono in a trial about masking in schools (in my view, a complex question) for the sole purpose of explaining that the court should not trust @DrJBhattarcharya because he is deliberately misleading people about our study and others.
The judge disallowed him as an expert witness because of his repeated misrepresentations:
https://twitter.com/UniversalMaski2/status/1452430888130777095
He now summarizes our study in Newsweek by saying, "In a study in Bangladesh, the 95 percent confidence interval showed that masks reduced transmission between 0 percent and 18 percent. Hence, masks are either of zero or limited benefit."
This is an absurd mischaracterization of our results as Jay is well aware. What does it get wrong?
1) We can rule out a zero effect w/ a p-value of .032 in the specification to which Jay refers
2) This is the *reduced form* from an increase in masking from 13% to 42%; so universal masking might be several times as effective
2) This is the *reduced form* from an increase in masking from 13% to 42%; so universal masking might be several times as effective
3) The CI for Covid symptoms is a 7-17% reduction (from our 30 pp increase in masking).
4) Our standard errors for symptomatic seropositivity drop by half under alternative (equally plausible) ways of imputing missing values, w/ a CI of 6-20%.
4) Our standard errors for symptomatic seropositivity drop by half under alternative (equally plausible) ways of imputing missing values, w/ a CI of 6-20%.
5) We have precise estimates for symptoms and symptomatic seropositivity at older age groups, esp. for surgical masks, where the CI rules out less than a 15% reduction (and includes a 55% reduction)
6) We have cross-sectional evidence showing that villages with larger increases in mask-wearing had larger reductions in symptoms and symptomatic seropositivity, in line w/ our other estimates.
Jay knows all of this since I messaged him privately to correct his misrepresentations, and he saw it all again when I testified in the trial to correct his misrepresentations but he keeps repeating his misleading statements in public forums.
Despite admitting to me in private correspondence that focusing on the absolute reduction in risk was misleading relative to the proportional reduction, he did *exactly this* in the court case before he knew I would be there to correct him.
His assessment of the DANMASK trial as showing that masks are ineffective is also equally misleading as I explained in my declaration to the court. 
This experience casts in a new and more sinister light Jay's past policy mistakes, such as his claim that Covid mortality was substantially lower than epidemiologists believed and so Covid would lead to fewer deaths than a typical flu season: wsj.com/articles/is-th…
Policy issues are complex and difficult -- some of the best scientists (e.g. @ProfEmilyOster) look at the data for themselves and reach conclusions that are at odds with many epidemiologists. This work is essential.
@DrJBhattarcharya's behavior is especially disgraceful because he has the credentials and appearance of a person doing this type of science, when in practice, he is engaged in sophistry to justify his predetermined conclusions.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
