As COVID-19 continued to rage through the United States in late September 2021, a tweet (later shared to Imgur) claimed that the virus killed 47 out of 100,000 people in counties where former U.S. President Donald Trump received 70 percent or more of the vote versus 10 out of 100,000 in counties where he received 32 percent or less of the vote — location metrics which were oddly specific:
That tweet cited and linked to a New York Times article of the same date, “Red Covid,” which was unfortunately paywalled. The Imgur user shared screenshots from the article in question:
A copy of the Times piece was archived; the screenshots matched the article’s first few paragraphs and graphs. As for the statistic mentioned in the tweet, it appeared in a second section (“When Covid was bluer”) alongside several contextual details:
It’s worth remembering that Covid followed a different pattern for more than a year after its arrival in the U.S. Despite widespread differences in mask wearing — and scientific research suggesting that masks reduce the virus’s spread — the pandemic was if anything worse in blue regions. Masks evidently were not powerful enough to overcome other regional differences, like the amount of international travel that flows through major metro areas, which tend to be politically liberal.
Vaccination has changed the situation. The vaccines are powerful enough to overwhelm other differences between blue and red areas.
Some left-leaning communities — like many suburbs of New York, San Francisco and Washington, as well as much of New England — have such high vaccination rates that even the unvaccinated are partly protected by the low number of cases. Conservative communities, on the other hand, have been walloped by the highly contagious Delta variant. (You can find data for hundreds of counties here.)
Since Delta began circulating widely in the U.S., Covid has exacted a horrific death toll on red America: In counties where Donald Trump received at least 70 percent of the vote, the virus has killed about 47 out of every 100,000 people since the end of June , according to Charles Gaba, a health care analyst. In counties where Trump won less than 32 percent of the vote, the number is about 10 out of 100,000.
The New York Times story referenced a date range (June 2021 through September 2021), an exacerbating factor (the Delta variant), and a source for the data, health data expert Charles Gaba.
On September 26 2021, Gaba tweeted about the Times citation:
In the tweet and the article, verbiage stipulated in part that in “counties where Trump won less than 32 percent of the vote, the number is about 10 out of 100,000” (versus 47 out of 100,000). The Times cited a September 17 2021 blog post by Gaba in which he explained that unusual line of demarcation.
First, Gaba explained exactly what he was calculating and why:
Two weeks ago [on or around September 3 2021] I was inspired by a fellow wonk named Christopher Ingraham of The Why Axis to plot out COVID vaccinations, cases and deaths via bar graphs instead of my usual scatter plot graphs.
Gaba created bar graphs in which one axis of the chart showed a COVID-related metric, vaccinations or deaths. The chart’s other axis represented percentages of general election votes that went to Donald Trump.
Gaba’s analysis produced what he described as “brackets,” as seen in the version below:
Gaba’s graph is titled “New COVID Deaths/100K since 6/30/21 by Trump 2020 County-Level Vote %.” Its horizontal axis was labeled “2020 County-Level Trump Vote %,” and the vertical axis was “COVID Deaths per 100K Residents Since 6/30/21.”
Gaba not only presented graphs, he discussed the figures and his findings at length — indicating limitations in the underlying numbers and resulting charts:
Note that this [graph] doesn’t include Nebraska and is still missing over 1,000 deaths in Florida due to their reporting methodology change and the fact that even the CDC’s report is still several days out of date.
Again, while the highest bracket is 8.6x higher than the lowest one, it’s more representative to use the 2nd & 8th brackets: The death rate in the reddest counties are around 5.6x higher than in the bluest counties since June 30th.
UPDATE: I’ve found a way to resolve the issue of some of the brackets above having wildly larger or smaller total populations than others (the 40-50% bracket has 84.5 million people, the 90-100% bracket only has 116,000…meaning the former has over 700x more people, potentially skewing the results dramatically).
I’ve therefore broken all 3,114 counties nationally from lowest Trump vote percent to highest as evenly as possible, in groups of as close to 33,145,000 people apiece as I could get them. I couldn’t do this exactly since most brackets cut off a counties with several hundred thousand or even over a million residents, meaning the bracket will either be a few hundred thousand lower or higher than the target, but it’s the best I can do.
As you can see, while both the case and death rate graphs show a somewhat less dramatic (and less even) trend line doing it this way, they’re both still clearly telling the same story …
As indicated in the tweet embedded above, Gaba followed up with a blog post on September 27 2021, “Time To Check In On The COVID Red/Blue Divide Again.” It began:
A month or so ago I was inspired by a fellow wonk named Christopher Ingraham of The Why Axis to plot out COVID vaccinations, cases and deaths via bar graphs instead of my usual scatter plot graphs.
This morning my most recent update of this analysis was linked to by David Leonhardt of the New York Times.
Today I’m posting updated versions of all three, with some important data/methodology updates …
Gaba indicated that he switched to a “decile view” (sectioned in tens) in order to better streamline the data parameters. The update post concluded with updated estimates of the statistic cited by the Times:
The case rate since 6/30/21 is 3.1x higher in the reddest counties (Trump >70%) than in the bluest counties (Trump < 26%).
The death rate in the reddest counties are around 5.5 higher than in the bluest counties since June 30th .
At the end of the post, three caveats were listed:
- These graphs only include the 50 U.S. states + DC (well, 49 in the case of the death rate graph). U.S. territories aren’t included since they don’t vote for President in the general election.
- The case & death rate graphs only start as of 7/01/21. As I’ve noted before, both graphs looked very different in spring/summer of 2020, fall/winter 2020, and spring of 2021…and that could change again one way or the other going forward.
- Most importantly: These graphs only lay out where the higher/lower vaccination, case and death rates are, not necessarily WHO. Remember, the reddest counties still usually have blue patches and vice-versa.
A viral Imgur post of a Twitter screenshot asserted that “in counties where Trump received at least 70% of the vote, COVID has killed 47 out of every 100,000 people since the end of June [2021, whereas in] counties where Trump won less than 32% of the vote, the number is 10 out of 100,000.” Those figures originated with health data analyst Charles Gaba, who began graphing rates of vaccination and death against “Trump counties” in early September 2021. Although there was no reason to suggest the calculations were inaccurate or misleading, Gaba’s original posts included significant context about the limitations of the data and charts; as such, we rated the claims Decontextualized.