Some fascinating (to statisticians) numbers have surfaced. First of all, December was far and away America's worst Covid month to date. Sad, and true. The Thanksgiving Spike did in fact appear, and a few places, notably California, especially suffered. We'll report some of that in a future post. Oddly, a few places that were not doing well, e.g. South Dakota, did much better than anticipated.
South Dakota appears to have done particularly well by seeming to erase a number of their cases this week. Here, for example are the statistics for four of their 66 counties (Beadle, Hughes, Minnehaha, and Pennington) from January 22 through October 31
And here's the count from November 1 through December 29
Which gives a net overall year for those four counties as follows:
MUCH BETTER, DON'T YOU THINK?
WOW--they were able to have 16,713 cases in just these four counties decide not to be Covid-related after all, and 29 who died weren't from Covid upon further review, but from underlying causes. After suffering a heavy media attack in November, was Governor Kristi Noem able to persuade her Health Department to change the data in the last two months? True, she was abetted by locals--recall that South Dakota was the scene of the Sturgis motorcycle rally in August that eventually was traced to more than 300,000 Covid cases in the Midwest, an event that Noem defended vigorously at the time, and throughout the autumn.
These four counties are not the only place in South Dakota with such manipulation. Here's the national choropleth for November and December -- note how few places are white, e.g. pristine. Amazingly, virtually ALL OF THEM are in SOUTH DAKOTA
Compare that to the national choropleth for January 22 through October 31, and no anomalies leap out
On the national scene, it is worth noting perhaps that November and December (even with the South Dakota subtractions) had 53% of the cases for the entire year. Here's the past 59 days
And here's the national total for 344 days
Looking just at South Dakota choropleths, here is the Jan 22-Oct 31 graph next to Nov 1-Dec 29
So, just what the hell is happening in South Dakota? Perusal of South Dakota newspapers is not at all revealing, but review of several independent sources shows a different story. Here's the Harvard data https://covidactnow.org/us/south_dakota-sd/county/minnehaha_county?s=1482438
And here's the CDC data, through December 31
So, let's look at Minnehaha County on our Dashboard. Here's Jan 22-Dec 27
And here's Jan 22-Dec 28. MUCH BETTER, don't you think?
So, it appears that Harvard and CDC think it is still pretty high, and in fact has added 823 new confirmed cases and 13 deaths in the past three days. So, are we seeing truly manipulated data or is this just an aberration of distorted facts temporarily in USAFacts?
The above analysis "popped out" from just comparing the national choropleths last night. And it certainly gave me some interesting challenges for 'what happened'. But it apparently is NOT an aberrant governor, nor malfeasance. Instead, upon digging in, it is apparent that during the Christmas holidays, someone at USAFacts got some tables mixed up, and assigned the updated columns for all the new cases and deaths, displaced by one row. So, Minnehaha county's data went to Moody county; Lincoln's data went to Lyman county, and Pennington's data went to Perkins county. So, my guess is that next week, this will all be straightened around, but we'll hold the anomalous data in our database forever. Here's the tipoff--comparing recent changes by the 'worst' counties and the 'best'
So, a wild goose chase after all. But proof positive that the overall choropleths can reveal anomalous problems easily, something that I have asserted for years, but have never had a better demonstration
No comments:
Post a Comment