The Null Hypothesis and School Closures, 2/14

Bryan Caplan writes,

Given how little students actually learn in school, its only clear-cut benefit is daycare.  Yet during Covid, many schools across the country were closed for over a year.  Extraordinarily inconvenient for parents, especially moms.  Yet virtually no one publicly stated the ugly-yet-obvious fact that great convenience for tens of millions of parents vastly outweighs a few extra Covid deaths.

Last year, Scott Alexander wrote,

So my prediction is that an average student could miss a year or two of school without major long-term effects. Their standardized test score would be lower at the end of the two years they missed than some other student who had been in school the whole time. But after a short period they would equalize again. I don’t think you need to burn yourself out working overtime to send your kid to a private school, I don’t think you need to risk your immunocompromised kid’s health to send her to the classroom, I think you can just chill.

The standard argument against school closures is that kids will fall behind in terms of learning. But this is difficult to reconcile with the Null Hypothesis, which is that attempts to measure the effect of educational interventions almost never find a significant, replicable, long-term effect.

Bryan’s point, with which I agree, is that school closures disrupt parents especially.

I do think that the overall adverse effects on young people of social isolation during COVID may be large. But it’s the social isolation in general, with school closures only a part of it. And if you keep schools open but don’t let kids be kids, that could be just as harmful as closing schools.