1/ The dates are ALL wrong. We ALL know it & yet ALL of the reporting & models are STILL using them.

The devil is in the details & so I'm shocked at how many of my pollster/political friends have failed to dive deep into the COUNTY-level #COVID19 data.

Let's take a trip to PA
2/ In particular to Chester County, PA
- population: 524K
- 977 positive cases
- 65 deaths

According to the dashboards over at @COVID19Tracking, PA posted ~300 new deaths on Tues. Thx to the NYTimes we have county-level data and we can see that 21 of those belong to Chester
3/ So according to the dashboards a third of ALL deaths in Chester county happened on April 21st or (you could infer) the 20th.

But previously, the highest 1-day total Chester had posted was 7 on April 11th. Looking over the details on the county dashboard we see municipalities
5/ "The coronavirus death toll at the nursing home climbed from one to nine between April 9 and Wednesday, the report shows. Chester County’s coroner said Friday that a 10th resident who had tested positive died in the hospital"

Article dated 11th but the deaths go back further!
6/ A quick news search on "Southeastern Veterans Center" yields the following stories. The rest of the tale will play out in dramatic fashion. National Guard called in, fired staff needs to be replaced, auditor sent to sort our scandal.
7/ Bottom line. That was ONE dive into ONE county and ONE municipality and ONE medical provider showing that a ONE DAY increase of 21 is spread over the entire month and into march more likely.

All. The. Data. You. See. Is. Nearly. Worthless.
You can follow @justin_hart.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: