my data does not care about your virtue signaling.
at the helpful suggestion of my longstanding amigo phil @kerpen i took a cut at the EMILY OSTER DATA on US schools to see whether mask mandates there are having any effect. the answer is, plainly, no.
for this cut, i looked only at schools that were 30%+ in person (top 3 of 4 cohorts) as the only other category was “remote only” and that one seemed a poor proxy and lacked student case rates.
i then plotted the case rates and aggregated them over the full dataset for students, staff, and community.
this is the aggregate data.
students in masked schools had a 21% higher case rate than students in schools with no masks. masks do not look to protect your kids.
staff case rate was 23% higher in masked schools.
areas imposing masks had a 19% higher overall community case rate.
there is no evidence here that masking kids in school stopped spread and masking is associated with higher, not lower case rates. this might lead some to claim that masks are upping spread of covid, but i would caution against drawing that conclusion from this data.
the higher community rate is almost exactly the same as the increase in rate in kids and staff in masked schools. this suggests that masks were just useless and the rate variance is just reflective of baseline community variance. that could easily be driven just by a higher testing level (sample rate).
this is easy to see here. variance drops to 2% higher (kids) and 3% higher (staff) in masked schools. we’re almost certainly getting within the margin of error here.
of course, it’s also possible that districts that masked kids also masked community and that masking overall drove up infection rates. this is pure conjecture and i have no data to support such a claim. i raise this merely to round out the scenarios that would be consistent with this data. i very much doubt that it’s the case.
“masking children had no material effect on covid cases among students or staff in schools” looks like the strongest hypothesis here.
it’s tempting to claim that there may be some temporal shifting here but i am finding this notion to be pretty iffy despite the graphics.
while i cannot rule it out, i suspect this is unlikely to be a function of earlier infection and earlier herd immunity generation.
if this were so and masks delayed infection, you’d expect to see them have an effect on overall infection rates, yet they do not once one adjusts for baseline community infection rates.
additionally, i’m struggling to see a plausible pathway whereby masking delays but does not stop covid and am always leery of trusting correlations where you have no reason to suspect causality. that’s how you wind up thinking that the last redskins game of the season predicts electoral winners (which it did, flawlessly, for decades. datamining if weird stuff.)
i suspect the more likely culprit here is sample rate and higher testing levels in schools having cross correlation with mask adoption. it seems plausible that schools that are adopting aggressive policies of one sort will also adopt others.
this temporal variance does not appear to anything like the same degree in community data. combined with the mismatch in temporal shifting between kids and teachers, this makes me suspect a sampling artifact in schools.
masked schools have higher covid rates for kids
masked schools have higher covid rates for staff
adjusted for the community case rate baseline, masked an unmasked case rates are basically indistinguishable.
you can have correlation without causality, but material causality without correlation is extraordinarily implausible. the lack of correlation in outcomes to masking looks like pretty conclusive evidence that masking children in schools does nothing to affect covid outcomes.
it’s a policy with significant COSTS and RISKS that exhibits no visible benefits.
it’s a complete fail rooted in pseudoscience and talismanic superstition.
stop masking your kids. it’s not science. it’s just wrong.