Thread by @TheLawyerCraig

Did you reply similarly to JHU as you did me here, or to the many (very prominent) accounts who linked to JHU’s data, when JHU put out its daily % positive for every single state by using CTP’s “totalTestsResults” while including antigen positives in case counts?

https://twitter.com/kissane/status/1370579655615336451

I'm sure you cannot have missed the ubiquitous use of the two columns in CTP's CSV/API (which CTP used for its topline daily data updates) to calculate and blast state-by-state positivity—including by media and policymakers—despite admonitions from CTP.

https://twitter.com/kissane/status/1370579552330649600?s=20

In GA, and FL, and AR, and TN, and TX, and LA, and MS, and others…this results in a massive error. Just saying, “We warned you against comparing states” while venerated institutions are doing just that and putting it out into the world like a shockwave is willful blindness.

Take MS, for example. It conducts a MASSIVE amount of antigen tests. For the week 2/22 – 3/1 (CTP data), if you use just confirmed cases and molecular tests, MS would have shown a 7.87% positive testing rate. But add in probable cases without the tests, and it became 16.37%.

When JHU (among others) is blasting the data, 16.37% simply becomes *the* number. Knowing the data in the two columns was being used how it was (and fully realizing we’re choosing between imperfect data) I would have ensured those columns produced the smallest error.

Short answer: no, I didn’t miss it (see tweet below). But the speed with which state-by-state data traveled despite CTP’s pronouncements that it’s “irresponsible” might have counseled better aligning certain data in certain states in the most-used columns.

https://twitter.com/TheLawyerCraig/status/1369663456748703750?s=20