Read on Twitter
This is shocking irresponsible. The study Oster is promoting here piles unsupported assumption on unsupported assumption to produce a wildly speculative number ("estimate" is too generous) for purported deaths from online schooling, to compare with actual deaths from covid. https://twitter.com/ProfEmilyOster/status/1327579152690667522
Let's walk through what they do here. First, they assume that "primary school–aged children received minimal meaningful instruction beyond what is being delivered by their parents or other caregivers at home." 1/n
This assumption is key to the whole project. The estimate is based on first assuming that online is the equivalent of no school, then looking for studies of the effects of interruption o schooling on eventual educational attainment, and then linking that to life expectancy. 2/n
But what is the basis for the assumption that online is the equivalent of no school? The first cite (and one of only two to published studies) is to a 2010 meta-analysis that in fact finds exact opposite - that to outcomes in online learning were just as good as in person. 3/n
They quote from this "effect size [for online learning] is not significant for ... K-12 students." Which of course means no difference between online and in person! But the authors are stupid or dishonest enough to read it as meaning no benefits from online learning at all. 4/n
The only other evidence they offer for their assumption - key to the entire paper - that online students are effectively not in school at all, is a couple of quotes from various people saying ion person schooling is better. That's it! 5/n
Once they've assumed that online students are not in school at all, they take two studies of teacher strikes in Belgium and Argentina to come up with a number of the total loss of educational attainment from schools being closed for that long. 6/n
Multiply that by the length of typical school closures and they get 0.12 to 0.15 years of schooling eventually lost as a result of school closures. 7/n
Then they assume that more schooling leads to longer life expectancy in a predictable way, so that we can say that kids whose schools were closed will live about a 0.25 years less as a result. Then throw in life expectancy, and voila! you've got total years of lost life. 8/n
The layers of speculation piled on unfounded assumption piled on unwarranted inference here are staggering. Tweak any assumption a little bit and you could get a totally different estimate. To compare this made-up number to actual deaths from covid is utterly irrresponsible. 9/n
