Interesting Twitter thread about the following question: If you plot the US state-by-state COVID infection rates since Sept 1, 2020–i.e. the recent and now receding surge–which is a better predictor:
The percent of the population infected before Sept 1, or
The margin of Biden’s victory in the state?
You might expect that states which had lots of COVID before Sept. 1 would have more of it after Sept. 1 for the same reasons they were getting it earlier. Or, I suppose, you might expect the reverse: states learn from their mistakes, and if infection rates were higher earlier then more people have immunity, so there’s a negative relationship between earlier infection rates and later infection rates.
According to Youyang Gu, both of those expectations are broadly wrong:
Instead, the single variable with the most predictive power is how strongly states voted for Biden.
As commenters in the thread note, at an R-squared of about 0.5, this is not a fully explanatory variable–there’s a lot going on, no doubt. Youyang Gu’s suggestive claim is only that as single-variable explanations go, this is the most powerful.
Florida is dismissing a recent warning from federal regulators about the accuracy of a popular COVID-19 test from one of the state’s largest testing providers — and continuing to use the test in a way the FDA has advised against.
Meanwhile, the state’s most populous county, Miami-Dade, is reconsidering how it uses the test.
The U.S. Food and Drug Administration issued a warning last month alerting the public to “the risk of false results, particularly false negative results,” with the test made by an enormous San Dimas, CA-based testing start-up, Curative, noting that false negatives can mean people unknowingly spreading the virus to others. The risk of false negatives is higher, the FDA noted, when the test isn’t used correctly.
Looks fairly dramatic to me, but of course I’m not a scientist.
The increase in cases has not been matched, yet, by as great an increase in deaths, although that number is also rising quickly.
I suspect this is because deaths lag cases by several weeks, and because treatment has improved. In addition, perhaps, a smaller fraction of the new cases are in the most vulnerable (e.g. elderly) population compared to the earlier peaks. But if the lag claim is correct, then expect the weekly deaths number to go up for some time before it goes down.