Since two papers have been published that found school closures in 2020 had no effect on the spread of Covid in Japan (https://linkinghub.elsevier.com/retrieve/pii/S1201971220305981 and https://www.nature.com/articles/s41591-021-01571-8], you’d think Japanese politicians wouldn’t bother closing schools again. But on Jan 26th 2021, 27.2% of schools in Shimane Prefecture were closed due to Covid compared with 3.1% nationwide.
(https://www.asahi.com/articles/ASQ245QZXQ24UTIL025.html https://www.yomiuri.co.jp/kyoiku/kyoiku/news/20220204-OYT1T50264/)
Apparently, after closing schools in various Shimane towns from Jan 20 to 31, the number of reported cases declined to 56% that of the peak the previous week. The best improvement in prevelance in the country!
Wow, what an effect! But what happens when you correct for the lag between infections and reported cases?
[Graph by Hiroshi Nishiura for the MHLW (https://www.mhlw.go.jp/content/10900000/000895931.pdf)
And that is how you can magically make ineffective NPIs look effective to people who can’t count backwards. Class dismissed!