4 ms·

> According to the article, in the only large randomized studies that have been performed, masking showed little, if any benefit. From the first one [1] they

> According to the article, in the only large randomized studies that have been performed, masking showed little, if any benefit. From the first one [1] they linked: > odds ratio, 0.82 [CI, 0.54 to 1.23] … Although the difference observed was not statistically significant, the 95% CIs are compatible with a 46% reduction to a 23% increase in infection. … Sounds like the issue is lack of statistical power. We basically can't conclude much at all from that study, but taken "at face value", that was an ~18% reduction in transmission. Presumably this would compound if done for longer or among more people due to exponential spread. Compound an 18% reduction for three generations (about ~15 days) of viral spread and that's about a 45% reduction in case load. Of course it's more complicated than that because a faster-spreading virus will also reach saturation of the susceptible population more quickly, but it does mean lowering peak hospital load, as well as lowering R_0 which means the point of saturation is smaller, etc. From the second one [2] they linked: > Proper mask-wearing increased from 13.3% in the control group to 42.3% in the intervention arm (adjusted percentage point difference = 0.29 [0.26, 0.31]). The intervention reduced symptomatic seroprevalence (adjusted prevalence ratio (aPR) = 0.91 [0.82, 1.00]) So that's a 9% reduction (again, with a wide confidence interval), as a result of increasing mask usage by only 29 percentage points. So if the signal's real, then presumably increasing mask usage further, say up to 85%, would add a very meaningful further reduction. Again, we can assume that would compound if done for longer or among a larger population. ** Furthermore, there have been plenty of lab experiments on masks which show them to be effective at filtering water droplet-bound Covid particles. And Covid is not the first virus we've ever encountered. In the absence of lots of data specific to Covid, we can apply background scientific knowledge about the effectiveness of masks in slowing the spread of respiratory disease. There are also costs to wearing masks, as well as to mask mandates, which I'm not taking into account in this post. But I wouldn't characterize the evidence as you have. ** [1] https://www.acpjournals.org/doi/full/10.7326/M20-6817 https://www.acpjournals.org/doi/full/10.7326/M20-6817 [2] https://www.poverty-action.org/publication/impact-community-masking-covid-19-cluster-randomized-trial-bangladesh https://www.poverty-action.org/publication/impact-community-...

claytongulick 5y ago
I agree that we interpret the results of those studies very differently. I'd also suggest that we only regard lab studies with a bit of intellectual curiosity, rather than a meaningful focus. Lab studies are very different than the real world where my son trades his batman mask for a friend's spiderman mask because it's cooler :D
- DavidSJ 5y ago
  In biological sciences we have to integrate multiple kinds of evidence — in vitro (or in larva? — in the mask) lab experiments, animal models, randomized controlled trials in humans, descriptive statistics of human populations, in silico simulation, etc. — because they all have their advantages and disadvantages. Lab/in vitro models are great because they allow for tightly controlled conditions and are clear tests of mechanisms, but they might not represent real-world conditions in living human beings in our social context, with all of the attending quirks and nuances. Animal models are great because they're a complete organism upon which we can still do a whole range of controlled experiments which we can't do in people. But they're not human beings, so there could be differences there. And we can't control them as closely as e.g. an experiment with a few cells or a few viral particles. etc. etc. When all the evidence seems to be pointing in roughly the same direction, I think it's fair to say we know something. That appears to be the case with masks limiting the spread of Covid, even if the confidence intervals on the randomized controlled trials are large. That's very different from "little, if any benefit".
  - claytongulick 5y ago
    Margery Smelkinson (the author), Ph.D., infections disease expert focused on Covid 19 and affiliated with NIH, disagrees with your interpretation.
    - DavidSJ 5y ago
      I think we should either evaluate arguments for ourselves, or rely on a broad consensus of experts in the field in question. If we resort to argument from authority, but cherry pick our authorities to be those which agree with our preexisting views, then that won't get us very far epistemically. In this case, I think both the broad consensus of experts as well as the balance of evidence/argumentation supports the efficacy of masks.

claytongulick 5y ago

I agree that we interpret the results of those studies very differently. I'd also suggest that we only regard lab studies with a bit of intellectual curiosity, rather than a meaningful focus. Lab studies are very different than the real world where my son trades his batman mask for a friend's spiderman mask because it's cooler :D

DavidSJ 5y ago
In biological sciences we have to integrate multiple kinds of evidence — in vitro (or in larva? — in the mask) lab experiments, animal models, randomized controlled trials in humans, descriptive statistics of human populations, in silico simulation, etc. — because they all have their advantages and disadvantages. Lab/in vitro models are great because they allow for tightly controlled conditions and are clear tests of mechanisms, but they might not represent real-world conditions in living human beings in our social context, with all of the attending quirks and nuances. Animal models are great because they're a complete organism upon which we can still do a whole range of controlled experiments which we can't do in people. But they're not human beings, so there could be differences there. And we can't control them as closely as e.g. an experiment with a few cells or a few viral particles. etc. etc. When all the evidence seems to be pointing in roughly the same direction, I think it's fair to say we know something. That appears to be the case with masks limiting the spread of Covid, even if the confidence intervals on the randomized controlled trials are large. That's very different from "little, if any benefit".
- claytongulick 5y ago
  Margery Smelkinson (the author), Ph.D., infections disease expert focused on Covid 19 and affiliated with NIH, disagrees with your interpretation.
  - DavidSJ 5y ago
    I think we should either evaluate arguments for ourselves, or rely on a broad consensus of experts in the field in question. If we resort to argument from authority, but cherry pick our authorities to be those which agree with our preexisting views, then that won't get us very far epistemically. In this case, I think both the broad consensus of experts as well as the balance of evidence/argumentation supports the efficacy of masks.