Frank Bednarz wrote back to me on twitter to describe the calculations used to represent the Covidtracking data, and he corrected the units to accurately represent deaths/week.

You can read what I originally posted to Frank below.

I was wrong and Frank was right about the data in this graph.

Also, Frank was using the graph to make a point about the CDC reporting delay, which doesn’t relay on the Covidtracking data. I seized on a part of the graph that jumped out at me, but didn’t relate to Frank’s point.

Darnit Frank Bednarz, that’s not how the Covidtracking database works.

You plotted COVID deaths for 8 April as 6,268 as if that were deaths per week, like the CDC data.

It’s not.

The Covidtracking website is cumulative. That’s all the COVID deaths, covering a period of four weeks.