Data Visualization

Response to a COVID-19 School Opening Risk Analysis

I appreciate efforts to put risk and data into context, but I fear a recent analysis getting some coverage on local forums in NH as they each wrestle with local schooling decisions is misleading.


Wayne Goertel, a school board member in Hooksett, NH published a recent analysis of COVID-19 deaths and traffic fatalities in New Hampshire — New Hampshire COVID-19 Back to School Data — to provide this sort of context. The analysis compares annual traffic fatalities by age group to COVID-19 deaths by age group. The analysis concludes that:

“For many student and staff families without significant risk factors, it appears that the COVID-19 fatality risk of being present in school may be less than the fatality risk of driving there. In NH, there are no deaths reported among people under 60 without comorbidities.”



While I really appreciate analyses that help internalize risk and aid decisions that depend on one’s risk tolerance, it is also important to properly conduct the comparison so that the relative risk isn’t misleading. There are a couple of assumptions that I think could lead to poor conclusions about the relative risk of death from COVID-19 and traffic accidents. Accounting for those assumptions changes the conclusions. Rather than concluding the risk of COVID-19 is only greater for those 60+, the risk is likely greater starting at 40+, and rather than being twice as dangerous to those 60+ it is roughly 10 times greater based on the US wide death rate.

Annual traffic fatalities vs. COVID-19 deaths to date

The comparison that is being made is across two different time periods, yet there was no correction for the difference in time periods from the two data sources. Auto deaths are annual, while COVID-19 deaths have occurred over a 6 month period (or 4 months given deaths didn’t ramp up until April). To correct for the difference in time periods I doubled the deaths rates observed to date to provide expected annual COVID-19 deaths, although I think one could argue the total could nearly be tripled to estimate deaths over the course of a year.

New Hampshire COVID-19 death rate vs. US death rate by age group

The death rate in ages under 60 in NH has been quite low in comparison to the national death rate for those under 60. That the death rate in ages under 60 in NH will continue to be different from the US wide rates may not be a sound assumption. It’s possible that because NH was able to avoid exceeding hospital capacity NH was able to keep death rates lower than the nation wide rates, but it might only be due to luck. For comparison I determined the expected NH COVID-19 deaths had NH deaths occurred at the national rate.

Death rate increase from school opening

A third assumption that is that the death rates would stay the same, or nearly so, regardless of in-person schooling. I don’t know how much data has been accumulated on the effect, but here’s a result from my quick google search that cites a study on reduced deaths due to the school closures in the spring: School closures in spring linked to drastic decrease in Covid-19 cases and deaths

My Analysis

For those that wish to check my math here is a link to the calculations: Google Sheet


About jonathan.cummings

Leave a Reply