[benbenolson]

The author of that blog post sees that states like Tennessee and Texas had a high number of school closures, but that plot is simply plotting raw numbers of school closures. Not only are those numbers affected by a plethora of variables (for example, population density, population number, vaccination rates, number of schools, closure policies...), but how could you possibly compare raw numbers of school closures in individual states? Then, they use this incredibly skewed plot to draw the conclusion that "Public health mitigation measures in schools, like mask requirements, work." Talk about jumping to conclusions.

For the first two studies, these kinds of probabilities are extremely dangerous and prone to "playing with the data." Given the exact same dataset, one could draw nearly any conclusion with such disparate and varied data, especially when attempting to control for other factors. Could these increased "change in daily number of cases per 100,000 children" be due to other factors; for example, a correlation between a mask policy and other policies, such as closure and a decrease in class size? I'm not saying that I know the answer, either, but plots like this should be taken with a pinch of salt.