More evidence that politics has taken over “science.”
Yale economists claimed a small (~10%) statistically significant effect of masks on symptomatic COVID illness in Bangladesh. Their paper was published in the journal Science – the supposed gold standard of scientific research.
The authors finally published their dataset. The difference in cases between the 300 villages that masked and 300 villages that did not was… 20 cases. Out of many tens of thousands of people, this is obviously NOT a statistically significant result.
An excellent re-analysis from Ben Recht, a computer science professor at Berkeley, and math professors Wes Pegden of CMU, and Maria Chikina of Pitt shows that sampling biases explain the entire difference (which is a tiny effect size anyway): https://arxiv.org/abs/2112.
Professor Recht did an interview with the Accad and Koka Report explaining these results that is long but highly entertaining: