Thread by @COVID19Tracking

Several dozen of our volunteers spent the last 5 days evaluating the new @CDCgov data. This is the white paper we produced.

covidtracking.com/documents/CDC%…

We generated as complete a report as could comparing the CDC data with the numbers we compile from the states themselves.

There is a known difference between what states report and what the CDC says it is reporting. Some states report "people tested" while the CDC reports "specimens tested."

4 states report both—so we used them to calculate a ratio to adjust the "people tested" numbers.

Unfortunately, while this method makes the numbers closer in *some* cases, it doesn't in many others.

Here's a bonus chart (not in the whitepaper) that we generated during this analysis. It shows the range of possible specimen values compared to the CDC's reported number.

All the data that we used for the analysis in the whitepaper is available on this Github repo:
github.com/COVID19Trackin…

The upshot: we will continue compiling and publishing data as we have been.

We're also ingesting the CDC data.

And we'll be working hard to understand *and* close the gaps we see.

Ultimately, one dataset should exist, and it should be managed by @CDCgov. That's the goal.

And just speaking as me, @alexismadrigal, I cannot believe the effort that our contributors put in on this report.

We've seen some incredible runs of work on the CTP, but this one was awe-inspiring.

Just Appendix B, gathering all the state orders on testing was a huge chunk of work. Same goes for Appendix D, an evaluation of the accessibility of the CDC's Data Tracker.

So many people worked through the night, worked through the weekend, worked through every meal to get every detail right for all of you who have come to depend on this project.

We're extremely lucky to have these contributors—and we extend our thanks.