Read on Twitter
New this morning: an ambitious data collection effort by @ProfEmilyOster and @AASAHQ is underway and early results aren't showing widespread outbreaks at in-person schools. @anya1anya and I report their findings here: https://www.npr.org/2020/09/22/915738935/new-dashboard-tracks-covid-cases-in-schools-across-49-states?live=1
Infections among students and staff are modest but not zero...~230 students in ~100k over 2 weeks, a daily rate of about 16 per 100k, which Harvard calls an orange risk level, advising rigorous testing. (cf https://globalepidemics.org/key-metrics-for-covid-suppression/)
~130 staff infected or suspected in ~38k, a MUCH higher rate of 35 daily infections per 100k, which Harvard classifies as its RED alert level. Curiously, this is elevated even among staff in remote ONLY schools. (lots of caps used here ¯\\_(ツ)_/¯)
Notably, these rates might lend themselves to widespread mitigation efforts by schools, per recommendations by Harvard et al. But the data shows traditional public schools -- where most students attend -- using fewer safety strategies than private schools and charters.
Important caveats abound. The variable infection rates among learning types is NOT meant to imply causality. There are many, many things that lead to these rates, including things like COVID prevalence in surrounding communities, nearby testing availability, etc.
And at this point the data set is still growing. The sample sizes are still modest. I'm looking forward to more schools, granular geo info, longitudinal data, etc. Stay tuned for these analyses from @anya1anya and....prob @shmcminn/ @connjie, bc I'm going on parental leave soon 
See the data in it's full glory at the link in this tweet, including the ability to DO YOUR OWN REGRESSIONS 
https://twitter.com/ProfEmilyOster/status/1308760305267355649
https://twitter.com/ProfEmilyOster/status/1308760305267355649
