4 ms·

13% is a large effect size, you'd have to presuppose a systematic bias in the unreliability between the comparator districts. Also, this is the case where resea

13% is a large effect size, you'd have to presuppose a systematic bias in the unreliability between the comparator districts. Also, this is the case where research needs to be conducted but reality makes it hard but you can't just +not+ do the research. Field tests are always messy, and I'd posit that community reporting is probably more reliable thn politically motivated reporting from some governments.

timr 3y ago
> 13% is a large effect size, You can't just say that. Depends on what you're measuring, how you measure it, and how big the denominator is. The fact that they saw the same effect size in groups that didn't get the vaccine is a reason to doubt the results, regardless of effect size. It was weird/credulous that they called it out as some kind of mysterious woo-woo advantage ("maybe it helps their immune system somehow!"). When you see stuff like that in a paper, it makes you scrutinize the results. When you hear it in a conference talk, it makes you reserve judgment until you see the paper.
- roywiggins 3y ago
  > The fact that they saw the same effect size in groups that didn't get the vaccine is a reason to doubt the result Where does it say that in the article? The closest it gets is "The mortality benefit was documented even in the areas with the lowest RTS,S coverage" but the lowest coverage was 62% (the highest was 75%).
  - timr 3y ago
    Yep, but then they go on to make the credulous woo-woo argument. None of us know what the actual results are here (since this is a conference talk, translated by a reporter), so I can only work with what they say. In any well-done study with an effective treatment, you'd expect, a priori, that a reduction in intervention produces a reduction in effect. In other words, you don't benefit if you don't get the drug.
    - roywiggins 3y ago
      The article is actually silent on whether the mortality effect was reduced, or whether the study was powerful enough to reliably detect that the mortality effect was reduced in lower-vaccinated communities. It just says the effect "was documented" in those communities, not whether it was exactly as strong.

timr 3y ago

> 13% is a large effect size, You can't just say that. Depends on what you're measuring, how you measure it, and how big the denominator is. The fact that they saw the same effect size in groups that didn't get the vaccine is a reason to doubt the results, regardless of effect size. It was weird/credulous that they called it out as some kind of mysterious woo-woo advantage ("maybe it helps their immune system somehow!"). When you see stuff like that in a paper, it makes you scrutinize the results. When you hear it in a conference talk, it makes you reserve judgment until you see the paper.

roywiggins 3y ago
> The fact that they saw the same effect size in groups that didn't get the vaccine is a reason to doubt the result Where does it say that in the article? The closest it gets is "The mortality benefit was documented even in the areas with the lowest RTS,S coverage" but the lowest coverage was 62% (the highest was 75%).
- timr 3y ago
  Yep, but then they go on to make the credulous woo-woo argument. None of us know what the actual results are here (since this is a conference talk, translated by a reporter), so I can only work with what they say. In any well-done study with an effective treatment, you'd expect, a priori, that a reduction in intervention produces a reduction in effect. In other words, you don't benefit if you don't get the drug.
  - roywiggins 3y ago
    The article is actually silent on whether the mortality effect was reduced, or whether the study was powerful enough to reliably detect that the mortality effect was reduced in lower-vaccinated communities. It just says the effect "was documented" in those communities, not whether it was exactly as strong.