3.5 weeks in to Alberta’s mask mandate, cases just keep right on rising, proving once again that if you keep trying something that’s already failed everywhere, unsurprisingly it will continue to fail pic.twitter.com/CJirP42wKP
— IM (@ianmSC) September 28, 2021
How would you check if mask mandates worked in schools?
I can think of only one way. Contrast kids in schools with mandates and in those without. Get them all, too; the kids, I mean. Measure something useful, like serious illness or death. Not infection, because every kid is likely to get this bug. Or has already had it.
You’d have to do this carefully, because the kids would almost surely differ class by class, or at least school by school, by all kinds of things causative of infection and illness. You’d have to count very delicately.
What you wouldn’t do is come up with some weird, derived measure, something that takes you away from the decisions you need to make. Like, “If we impose a mask mandate, are fewer kids going to get sick?” And, “Is it worth the costs, which only zealots say are non-zero?”
The new CDC report is “Pediatric COVID-19 Cases in Counties With and Without School Mask Requirements — United States, July 1–September 4, 2021” by Budzyn and a slew of others.
Before the details, look at their proof pic:
Caption is “Mean county-level change in daily number of COVID-19 cases per 100,000 children and adolescents aged <18 years in counties (N = 520) with and without school mask requirements before and after the start of the 2021–22 school year — United States, July 1–September 4, 2021”.
Why not all counties? “Among the 3,142 U.S. counties included in the initial sample, 16.5% (520) were included in the final analysis after applying the selection criteria.” And that selection criteria? “Counties with conflicting school mask requirements were excluded from this analysis; only those counties with the same known mask requirements for all schools were included.”
This is the start of the problems you give yourself by counting the wrong thing. They measured counties, not kids. And, as we’ll see, they measured something not terribly interesting: “cases”.
If you, for instance, started on the east coast early in the morning and measured kids exposed to sunrise and compared these kids with kids in California, it looks bad for living on the east cost, if being exposed to sunrise worries you. But if you pondered it for a moment, you’d realized every damned kid is eventually going to get exposed. So why not measure something useful, like skin cancer?
Now, what’s this before and after the start of the school year business? They should have been looking inside schools during the school year. Not before. That’s the only valid comparison. Who in the world knows what these kids were doing in areas of more and less freedom before the school year.
They did not look at the kids wearing masks or not, just “cases” in counties with out without mandates. Cause is already suspect.
And what is a “case”, dear reader? Regular readers will—should—know the answer to this.
A “case” is a function of test quantity, test sensitivity and accuracy, and disease prevalence. A “case” does not measure previous infections. A “case” does not measure disease severity. Test quantity—here normalized by population—is a function of madness, panic, and “concern” level.
Did those counties without mandates, thinking they were necessary, have more testing, and thus possibly higher “case” numbers? Did the masked kids have higher previous infection rates. I have no idea. The CDC isn’t saying about either.
And what about prior infections? All the panic that goes into creating mandates may, and likely did, have seen differences in behavior that are causally related to the outcome. Which kids already had the bug, but which the testing doesn’t reveal?
This means, again, individual kids must be checked, both for prior infection and thus excluded, and for something real, like hospitalization for Covid, or serious illness defined in some defensible clinical way. Or death.
Ah, no deaths reported. Kids aren’t really dying of this disease—only about 500 in the entire USA in two years—so comparing that most useful number wouldn’t have been possible.
All this change in rates of change is also impossible to keep straight. What’s wanted is direct comparisons of rates between kids wearing and not wearing masks. Cumulatively, too. That is, “This many infected day 0, this many day 7, this many day 14”, and so on. Again, properly excluding those previously infected. We can’t get any solid idea of effect from these weird numbers.
CDC says “Statistical significance was defined as p<0.05 for all analyses.” P-values? Uh oh. Get ready to be flashed.
Then came the regressions—I know, I know. “To further assess the association between pediatric COVID-19 cases and school mask requirements, a multiple linear regression was constructed that adjusted for age, race and ethnicity, pediatric COVID-19 vaccination rate, COVID-19 community transmission, population density, social vulnerability index score, COVID-19 community vulnerability index score, percentage uninsured, and percentage living in poverty.”
If you can’t get a wee P out of all that, you’re just not trying.
Any study in which a researcher with pride waves his–or her!—wee p-value at you should not be trusted.
Subscribe or donate to support this site and its wholly independent host using credit card or PayPal click here