Saying the effects of masking on younger age groups and with cloth masks in this study were statistically insignificant is also a misrepresentation of what the study found. The actual paper itself can be downloaded here:

https://www.poverty-action.org/publication/impact-community-...

From the paper:

    We found clear evidence that surgical masks are effective in reducing symptomatic seropreva-
    lence of SARS-CoV-2. While cloth masks clearly reduce symptoms, we find less clear evidence of
    their impact on symptomatic SARS-CoV-2 infections, with the statistical significance depending
    on whether we impute missing values for non-consenting adults. The number of cloth mask vil-
    lages (100) was half that for surgical masks (200), meaning that our results tend to be less precise.

They're talking about the analysis of the sub-group that consented to blood tests for Covid antibodies through the trial, and the fact that a smaller group of people (which could also be a confounding factor) consented to these, but that if they assume the non-consenting participants had similar rates of infection as the consenting group (instead of throwing them out), that the advantage becomes more pronounced even for cloth masks.

Furthermore, they said that even without that consideration, the cloth masks still reduced symptoms of Covid, which is still a great win for masks, as it seems they reduce the viral load received by an individual enough to reduce the severity of their infection. So saying the overall advantage of masks was statistically insignificant isn't really true.

What the trial really found was indeed that masks seem to work, and surgical masks work better than cloth masks. This aligns with what many other studies have found as well. However, even though that's probably a valid inference to draw from the data collected in the trial, it's not really what the trial was designed to find, which was more specifically that if you actively distribute masks to people and encourage them to be worn, then more people will indeed wear them, and this in turn seems to have a positive effect on reducing both the transmission and severity of Covid.

The question is how effective masks are w.r.t. Covid. For this the test really needs to be seroprevalence, not just symptoms. Symptoms can be caused by many non-Covid causes like allergies or the common cold.

The researchers state:

> We selected the WHO case definition of COVID-19 for its sensitivity, though its limited specificity may imply that the impact of masks on symptoms comes partly from non-SARS-CoV-2 respiratory infections

Teachers are the "50+ population" in schools, right?

I never know what to make of posts like this. It seems almost insulting to have to explicitly point out that masks are about reducing transmission, and that transmission could occur between students (not 50+) and teachers (many 50+).

Right?

So your conclusion should be that the Bangladesh study is quite relevant to schools, yet you seem to be suggesting the opposite?

Though I knew about it, I don't think I really appreciated how people could cherry-pick information to fit their preconceived ideas until covid.

> That study showed that the effect of surgical masks was statistically significant when used by the elderly 50+ population. For other age groups and for cloth masks, the advantage was statistically insignificant.

The study could prove that the masks worked in 50+. And this was with a maximum of 40% of people complying.

This does NOT mean that the masks didn't work in 50-. It means that the statistical power was insufficient to prove that they worked.

It could also mean that 50- didn't comply with masks as well. It could also mean that 50- engaged in other risky behavior. etc.

Why everybody seems to think that "Provably works in 50+" and "Provably doesn't work in 50-" are completely compatible ideas is completely beyond me. "Provably works in 50+" significantly increases the Bayesian prior on "Works in 50- even if we can't prove such."

The advantage of this study is that the villages were randomized so it's unlikely that the control group engaged in more risky behavior than the treatment group.

This result for 50+ was right on the edge of the significance interval so proves is probably too strong of a word. It means you would expect this result to happen by chance if the experiment were repeated 20 times. It is definitely evidence for some efficacy but it is about as weak as it gets to be considered evidence from a scientific POV.

My point was just that the headlines painted a significantly different picture than was contained within the paper itself.