If at First You Don’t Succeed, Revise, Revise Again

I’ve posted a few times about OC COVID-19 results the last few weeks. One of the issues is the data has HUGE (YUGE perhaps?) revisions on the number of tests well after the fact.

Here is a chart showing the number of tests reported by he day I downloaded the data.

Now, I get that if I look at data on May 25th, the last couple of days may need revisions. But on May 29th, the data for May 15th (2 weeks!!!) was revised up by 40%.

The same thing happened on May 19th. That data was revised on June 4th, two weeks after the fact.

I have no idea what is going on with these major revisions, but it makes some analyses hard.