Post #1234: Redoing my baseline for analysis of COVID-19 spread in Virginia K-12 schools.

Posted on September 5, 2021

That headline above showed up on Google News this morning.  Their conclusion is consistent with what I thought I saw late last week, described in Post #1229.  Cases ticked up, but hospitalizations ticked down.  That’s what you’d get if the latest influx of cases was school-aged kids, as hospitalization for COVID is extremely rare in that age group.

Redoing the baseline for analysis of Virginia K-12 public schools

I took that Wall Street Journal article as a sign to continue to focus on Virginia K-12 schools.  My goal is to do, within Virginia, what the Wall Street Journal did, across the states:  Compare early-opening to late-opening school districts in terms of the pediatric COVID-19 cases.

In my last post, I showed how to look up Virginia school to see the status of any active COVID-19 outbreaks.  Just prior to that, I tested out the programming for setting up an analysis of Virginia school districts based on school start dates.

I was rewarded by finally finding the consolidated 2021-2022 public school calendars for Virginia school districts (on the directories page of the Virginia Department of Education website).  So I figured I’d redo my baseline using the actual start dates, not the start dates from last year.

The first thing I learned as that a lot of Virginia school districts delayed opening last year due to COVID-19.  So using last year’s calendar as a stand-in for this year would not work.

It appears that no Virginia school districts are delaying their start dates this year.  Which, depending on how you look at it, either is or isn’t a somewhat odd thing.

Data source for this and other graphs of new case counts:  Calculated from The New York Times. (2021). Coronavirus (Covid-19) Data in the United States. Retrieved 9/4/2021, from”  The NY Times U.S. tracking page may be found at

On the one hand, the likelihood of spreading COVID-19 in Virginia schools is far higher this year than it was last year.  The level of cases in the community is four times higher, and the circulating variant (Delta) is far more infectious.  We have vaccines now, but that doesn’t do much for kids under 12.

On the other hand, given the situation, there may have been little point to any delay in Virginia.  Judging from the shape of the Delta wave for Virginia (and all the other states in the South Atlantic region except Florida), we should expect some weeks of rising cases yet.  Any reasonable school start date will have the children in school at (what we hope is) the peak of the Delta wave in Virginia.

Early versus late school start comparison, redone.

The table above shows the distribution of 2021 K-12 start dates across Virginia public school districts.  About a third of the population lives in districts that start on 8/19/2021 or earlier.  About a quarter live in districts that start after Labor Day.

Ultimately, though, I have to crosswalk those individual school districts into the Virginia Department of Health health districts.  That’s because the cases x date x age group x location data file published by Virginia uses health districts as the geographic location.

If I take population-weighted school start dates by health district, I come up with the following 2021 school start date distribution across health districts.   It’s not hugely different from the distribution by school district.   But the cutoffs shown above will define the early-middle-late classification below.

Assuming I did all the programming right, the graph above is my true baseline for this analysis.  Using my two-weeks-to-get-into-the-data rule, the early group (in session on 8/18/2021) would just barely start to have cases showing up in the data this past week.  By contrast, the late-start group hasn’t even gone back to school yet.

I wish there were less variation in those baselines, but the data are what they are.  I’ll now start looking for an increase in the early-start group, relative to the late-start group.  Unfortunately, given the natural variation of the data, if the impact of going back to school is small, I’m not going to be able to see it on this chart.  In statisticians’ terms, this lacks statistical power. But it is what it is.