Sunday morning you may have read (or heard) a story on NPR which focuses on something I’ve been writing about since last spring:
Pro-Trump counties now have far higher COVID death rates. Misinformation is to blame
Since May 2021, people living in counties that voted heavily for Donald Trump during the last presidential election have been nearly three times as likely to die from COVID-19 as those who live in areas that went for now-President Biden. That's according to a new analysis by NPR that examines how political polarization and misinformation are driving a significant share of the deaths in the pandemic.
NPR looked at deaths per 100,000 people in roughly 3,000 counties across the U.S. from May 2021, the point at which vaccinations widely became available. People living in counties that went 60% or higher for Trump in November 2020 had 2.7 times the death rates of those that went for Biden. Counties with an even higher share of the vote for Trump saw higher COVID-19 mortality rates.
If this seems awfully similar to the work I’ve been doing and posting about both at ACASignups.net as well as on Twitter and here at dKos, there’s a reason for that:
Now, it's true that all of the data itself is publicly available; I use many of the same sources myself. It's also true that different data analysts often work on similar projects at the same time; there's nothing nefarious about that. I presume they've been working on this story for some time...perhaps not as long as I have (since early this summer), but for awhile, anyway.
HOWEVER, when you contact a fellow data analyst specifically because they're working on the same data, then request a lengthy meeting with them in order to speak with them about their work and their methodology, and then follow up with them in order to schedule a second lengthy meeting to discuss the same work again, and then follow up with them a third time to clarify other details, it's generally considered professional to mention them at least once in the final piece you publish on your far higher-profile platform.
The good news is that this issue was rectified within a few hours of a backlash on Twitter:
Mr. Brumfiel called me shortly after I posted this to apologize and to say he plans on rectifying the situation. To his credit, he at least took full personal responsibility for the oversight. That is, he didn't try to pin the blame on an editor, intern or his colleague Mr. Wood, who he said did the data analysis itself but had nothing to do with the wording of the actual NPR story.
Sure enough, the NPR story has been updated with the following:
"In October, the reddest tenth of the country saw death rates that were six times higher than the bluest tenth, according to Charles Gaba, an independent health care analyst who's been tracking partisanship trends during the pandemic and helped to review NPR's methodology. Those numbers have dropped slightly in recent weeks, Gaba says: "It's back down to around 5.5 times higher."
The bad news is that after all that, the county-lookup database tool created to accompany the NPR story included two major errors...both of which I had specifically warned the authors about. One related to Florida’s data; the other related to Utah:
Yes, that's right: Florida's Health Department quietly stopped reporting county-level COVID death data on June 3rd, and while I've been using the Community Profile Report from the White House COVID-19 Team, Joint Coordination Cell, Data Strategy and Execution Workgroup (that's a mouthful!) instead of CDC data for Florida, Mr. Wood apparently decided to stick with the Johns Hopkins github data for Florida instead...even though the Johns Hopkins data is still reporting the same June 3rd data for Florida.
This is particularly problematic because I specifically warned Brumfiel and Wood about the Florida data issue and recommended that they use the Community Profile Report instead during one of our two hour-long interviews.
You see, for some inexplicable reason, Johns Hopkins University doesn't break out Utah's COVID data by county--they do it by region, at least partly. They group data from most of the states 29 counties together into 6 regions called "Bear River," "Central," "Southeast," "Southwest," "TriCounty" and "Weber-Morgan."
The problem is that this means that JHU shows 22 Utah counties as having no COVID deaths...even though those counties have over 1/3 of Utah's total COVID deaths. The 517 deaths from May - November are instead listed as being in those 6 regions, none of which show up in the NPR graph.
FORTUNATELY, after I called out each of these errors, they’ve both since been rectified:
Sheesh.
On the upside, at least the story itself is getting a wider audience, and the rest of the NPR article itself is well done. Also, I ended up getting some new followers on Twitter (and even a couple of new supporters) out of the whole mess, so there’s that.
In any event, now that that’s all finally settled, here’s updated looks at my own version of both the county-level vaccination rates as well as the county-level COVID case & death rates by both 2020 partisan lean and by vaccination rate.
At the top of this diary is, of course, my now famous (infamous?) Red/Blue Vaxx Scatterplot graph, up to date as of December 6th.
Here’s a bar graph look at COVID-19 case rates since the end of June (roughly the time that the Delta variant reared its ugly head here in the United States):
The entire population of the 50 U.S. states + DC have been broken into roughly equal brackets of ~33.1 million apiece. The bar on the left includes the bluest counties, where Trump received less than 26% of the vote last year; the bar on the right has the reddest ones, where he received 70.6% of the vote or higher.
From 6/30/21 — 12/06/21, the COVID case rate was ~2.7x higher in the reddest tenth of the U.S. than the bluest tenth, with a steady increase in between.
Next, here’s the most telling graph: The COVID death rate for the same time period: Over 108 people per 100K died in the Trumpiest tenth of the country vs. 18.7 per 100K in the least-Trumpy tenth, or nearly 5.8x as high:
Here’s what the death rate from the end of June through the beginning of December looks like at the county level in scatter-plot format:
If you’re trying to talk sense to a Republican and want nonpartisan visual aids to help make your case, here’s the county-level case rates and death rates broken out by vaccination rate:
From June 30th — December 6th, there were 2.1x as many COVID cases per 100K residents in the least-vaccinated tenth of the U.S. (less than 43% vaxxed) than in the most-vaccinated tenth (over 71% fully vaxxed).
As for the death rate, that contrast is, again, even higher: 107 people have died of COVID per 100K residents since June in the lowest-vaxxed bracket vs. 24 per 100K in the highest-vaxxed bracket, or nearly 4.5x as many:
(In fact, both the case and death ratios are even higher yet if you disregard Miami-Dade County, which is a special case due to what appears to be a massive issue with snowbirds and vaccine tourism.)
Finally, here’s a scatter-plot version of the June — December death rate by vaccination rate: