There are times when correlation really seems as though it is obviously representative of causation. Some set of data lines up so neatly that it’s hard to escape reaching an obvious conclusion. This can be dangerous, certainly, but usually it’s simply because the data are causally related. A plus B seems to equal C because it does.
An example of this is the overlap of covid-19 deaths since 2020 and partisanship. Last month, I reviewed the evolution of the pandemic, assessing where fatality from the virus was sharpest. Even in the era of omicron — a moment when we’re dealing with a less-deadly virus and have better vaccines and treatment to address it — there is a noticeable gap between Democratic- and Republican-voting places, just as there has been for months.
The word “places” is important there. What we could produce was analyses like the one below, showing how Florida and Ohio counties that voted for President Donald Trump in 2020 had higher rates of death, month after month, than ones that voted for Joe Biden. That didn’t necessarily mean that more Trump voters were dying, just that places where more of them lived saw higher fatality rates.
(In mid-2021, Florida briefly stopped reporting county-level death tolls.)
Why might it be the case that places with more Trump voters saw more deaths?
Well, we know that many of those who died of the virus last year were unvaccinated. The Kaiser Family Foundation estimates that about 234,000 deaths from June 2021 through March 2022 could have been prevented had the decedents been vaccinated against the virus. That protection, too, held into the omicron era.
We also know that Republicans were less likely to get vaccinated than Democrats. Republican officials often downplayed the utility of vaccination, responding to framing of the shots as an unnecessary intrusion from the government. Trump-voting counties were also more likely to seek alternative treatments for covid, such as the drug ivermectin — treatments that were shown repeatedly not to be effective.
All of this, though, was correlation. We couldn’t say that more Republicans were dying, specifically, so the link to partisanship was indirect, however clearly rational an assumption it might have been.
Last month, though, the National Bureau of Economic Research published an important study from researchers affiliated with Yale University. They took 577,659 death records from Ohio and Florida between January 2018 and December 2021 and matched the decedents to a 2017 voter file. In other words, they were able to identify the partisanship not only of the places those people lived but of the people themselves.
What they found is that the rate of excess death — that is, deaths above the expected toll relative to the pre-pandemic baseline — was higher for Republicans, particularly after vaccines were rolled out.
“Registered Republicans in Florida and Ohio had higher excess death rates than registered Democrats, driven by a large mortality gap in the period after all adults were eligible for vaccines,” the researchers write. “These results adjust for county-by-age differences in excess deaths during the pandemic, suggesting that there were within-age-by-county differences in excess death associated with political party affiliation.”
You can see that on charts included in the report. Democrats and Republicans in Ohio and Florida died at higher rates than would have been expected based on 2018 patterns. But Republicans were much more likely to do so, particularly after the second dark, vertical line on the graph — the point at which all adults became eligible for vaccines.
In another chart, the point is made explicitly. Before the vaccines, the pattern of deaths during the pandemic looked the same for Democrats and Republicans, even in counties that would later have lower vaccination rates. Then, after vaccinations became available, the divergence emerged — and was much wider in counties with lower vaccination rates.
This is something else we knew, too: Counties that backed Trump more heavily in 2020 had lower vaccination rates. In other words, there was a separate correlation between vaccination rates and Trump support. If we separate out the counties in Ohio and Florida that backed Trump by a margin of at least 40 points and those that backed him by a narrower margin, we see another separation.
More-Trump-supportive places had higher death rates than less-Trump-supportive ones during peaks in the pandemic, and both had higher death rates than Biden-supportive ones.
“The results suggest that the well-documented differences in vaccination attitudes and reported uptake between Republicans and Democrats have already had serious consequences for the severity and trajectory of the pandemic in the United States,” the researchers conclude. “If these differences in vaccination by political party affiliation persist, then the higher excess death rate among Republicans is likely to continue through the subsequent stages of the COVID-19 pandemic” — exactly as we’ve seen in county voting data.
Sometimes correlation is causation.