We don’t usually talk much about what we’re up to at work, but in the last few weeks I’ve had the opportunity to share some research work from earlier this year. Back in June, I teamed with a group to produce an analysis of how (or whether) U.S. government policy had affected COVID-19 infection rates. We’ve written up the work in a post on the Elder Research blog, and I presented the same at the Data Science Conference on COVID-19 (DSCC19).
I’ll defer most details to the linked post, but we analyzed months of data at the U.S. county level to test for policy impacts on the growth of COVID-19 cases. We also adjusted for many other potential explanatory inputs including testing, population size and density, and key demographic factors including income and minority representation. Even after these adjustments, stay-at-home orders were associated with a slight decrease in cases on average and were linked to a continuing reduction in cases over time (i.e., these orders seemed to decelerate the disease progression).
We’ve published the source code for both the original project and our article. Having code available made the project a good fit for DSCC19. The article source, particularly, contains data and an R Markdown document that, when compiled, should produce the same model fits we show in the article.