Read on Twitter
The COVID-19 data being used for public policy (and for the media to get clicks) is generally very poor in terms of presentation and transparency. - Thread
One large challenge is that the date that deaths are reported and the actual date on which these deaths occurred are routinely very different given reporting lag-time. This is not a conspiracy, but simply the way bureaucratic data collection and reporting tend to work.
Yesterday, media reported that Florida had its worst day of the epidemic yet, with 120 deaths. This shocked me, so I asked around and @GooglerDe helped me out. There is a trove of FL data here: http://ww11.doh.state.fl.us/comm/_partners/covid19_report_archive/
Within this trove are files labeled "state_linelist" and they are lovely excel files containing granular info on cases and deaths. Here is today's: http://ww11.doh.state.fl.us/comm/_partners/covid19_report_archive/state_linelist_20200710.xlsx
@justin_hart has been doing this for a while, and yesterday showed that this "new peak" number was incorrect. https://twitter.com/justin_hart/status/1281274249097797633
Today FL *reported* 93 deaths, also a very high figure. The FL spreadsheet labels these "Newly identified death(s)." But, of course, there were only 9 deaths (so far reported) on July 9. Most of the deaths are lagged reporting going back as far as April 24, but mostly for June.
Using the data on date of death, and some basic excel work, I present to you with this chart of the COVID-19 deaths in Florida by date of death, with the caveat that more old deaths, when finally reported, are likely to make June look bigger than it does already.
Florida deaths by date of death
And if the data reporting-lag problems can be this big for one state, imagine what they do to US and global reporting in aggregate? https://twitter.com/justin_hart/status/1281682833417760768
So the lesson here is to ignore media hysteria. Unfortunately, we can no longer trust journalists to be diligent. Reporting deaths in the way they are is misleading at best. They should not, at this point, be able to plead ignorance. Do you own thinking and question everything.
UPDATE: The chart I produced in this story is inaccurate. Please see my correction/clarification and updated info here: https://twitter.com/JoshFriedlander/status/1282099281495830532
If you don't want to read the entire above thread, the bottom line is that the accurate Florida deaths figures are here: https://fdoh.maps.arcgis.com/home/item.html?id=230270972343459a812a1ae8c28574a6&view=list&sortOrder=true&sortField=defaultFSOrder#overview
