Political ideology, vaccine hesitancy, and death
John Clark
GPT: "Yes. In the United States, there was a substantial relationship among political affiliation, COVID-vaccine hesitancy/uptake, and mortality—but the timing is crucial.
The clearest evidence comes from studies that looked specifically at what happened before versus after COVID vaccines became widely available.
What happened
Early in the pandemic, there wasn't much of a Republican–Democrat mortality difference once researchers accounted for age, location, and other factors. In fact, Democratic-leaning urban areas were hit extremely hard during the first waves.
But after vaccines became available to essentially all adults in spring 2021, a striking divergence appeared.
A particularly good study, published in JAMA Internal Medicine, linked voter-registration records to death records for 538,159 people in Florida and Ohio. It found:
- Before May 2021: there was essentially no statistically significant Republican–Democrat difference in excess mortality.
- After May 2021: Republicans had substantially higher excess mortality than Democrats.
- In the researchers' adjusted analysis, the Republican excess-death rate was 43% higher than the Democratic rate after vaccines became broadly available.
- The disparity was largest in counties with lower vaccination rates.
That timing is extremely informative because it corresponds almost exactly to the point at which vaccination ceased to be primarily a question of availability and became increasingly a question of whether people choose to get vaccinated.
And there really was a political difference in vaccination
Numerous studies found that Republican identification was associated with greater COVID vaccine hesitancy and lower vaccination uptake. One county-level analysis found political affiliation to be among the most important predictors of vaccine hesitancy, with Republican-leaning counties generally exhibiting greater hesitancy.
So there is a fairly coherent chain:
political identity → attitudes toward COVID vaccination → vaccination behavior → different risk of severe COVID/death
The mortality study is particularly persuasive because the partisan mortality gap wasn't prominent before vaccines were available, but emerged afterward and was greatest where vaccination rates were lowest. That pattern is difficult to explain as merely "Republicans happened to be less healthy."
The JAMA study had an important limitation: the researchers didn't have individual vaccination records, so they couldn't say, for example, "this particular Republican died because he was unvaccinated." They inferred that vaccination differences were an important contributor from the population-level patterns.
Nevertheless, the evidence becomes considerably stronger when you put several observations together:
- Republicans were, on average, more vaccine hesitant.
- Republican-leaning areas generally had lower vaccination rates.
- The partisan mortality difference was small or absent before vaccination was widespread.
- It became large after vaccination became widely available.
- The difference was largest in areas with low vaccination rates.
- Other studies found higher COVID mortality in Republican-leaning counties even after accounting for demographic differences.
That's a remarkably consistent pattern.
An important distinction from your previous question
This is also why your friend's argument that "COVID became less deadly because the virus evolved, not because vaccination helped" doesn't fit the data very well.
The virus certainly did evolve, and later variants had different characteristics. But if declining lethality were the primary explanation for the dramatic decline in deaths, you wouldn't expect mortality to diverge so sharply according to political affiliation and vaccination prevalence after vaccines became available.
The partisan mortality pattern is actually one of the more striking pieces of real-world evidence that vaccination was making a substantial difference in who died, even though it wasn't the only reason mortality eventually fell.
There is an especially interesting quantitative result here: the researchers found that the Republican/Democratic mortality gap essentially appeared at the point when vaccination became widely available. That makes the political/vaccine connection much more compelling than simply observing that "red counties had more deaths. I looked up the actual study's weekly data/figure, and the pattern is quite striking. . It reports weekly estimates graphically, plus aggregated periods.
The study by Wallace, Goldsmith-Pinkham, and Schwartz in JAMA Internal Medicine examined 538,159 deaths of registered voters aged 25+ in Florida and Ohio, using 2017 voter-registration records linked to mortality records.
What the weekly pattern looked like
The important part is the timing:
| Period | Republican excess mortality | Democratic excess mortality | Adjusted Republican–Democrat gap |
|---|---|---|---|
| Apr. 2020–Mar. 2021, before widespread adult vaccine eligibility | 19.4% | 18.8% | −0.9 percentage points |
| Apr.–Dec. 2021, after widespread eligibility | 25.8% | 18.1% | +7.7 percentage points |
The researchers' definition of excess mortality is essentially observed deaths divided by the number of deaths expected from pre-pandemic patterns. So, for example, 25.8% means roughly 26% more deaths than would have been expected absent the pandemic.
And the really interesting part is what happened week by week:
- Winter 2020–21: both Republicans and Democrats experienced a large surge in excess deaths, and the two curves were fairly close.
- Spring 2021: the curves came back down and remained relatively close.
- Summer 2021: the Republican curve began rising noticeably above the Democratic curve.
- Fall 2021: the separation became substantially larger.
The 43% figure I mentioned previously comes from comparing the adjusted excess-death rates after vaccines became widely available: Republicans' excess mortality was estimated to be 43% higher than Democrats'.
Why I think the timing is more important than the 43% number
Imagine that Republicans had simply had substantially higher mortality throughout the entire pandemic. That wouldn't tell us much about vaccination.
But that's not what happened.
Before vaccines were broadly available:
Republican excess mortality: 19.4%
Democratic excess mortality: 18.8%
That's essentially the same.
After vaccines became broadly available:
Republican excess mortality: 25.8%
Democratic excess mortality: 18.1%
That's a very large divergence. And the researchers found that the divergence was largest in counties with low vaccination rates and became much smaller in highly vaccinated counties.
That is exactly the pattern you would expect if differences in vaccination were contributing substantially to the mortality difference.
There is an important caveat
This particular study cannot prove that an individual Republican died because he or she was unvaccinated. The researchers didn't have individual vaccination records. They had political affiliation for individuals but vaccination rates only at the county level.
Also, the effect was much stronger in Ohio than Florida; in Florida, the overall Republican excess-mortality difference wasn't statistically significant.
So I would phrase the conclusion carefully:
The evidence strongly supports an association between political affiliation, vaccination behavior, and COVID mortality, and the timing of the association is unusually suggestive of a causal role for vaccination. But this particular study cannot quantify exactly how many Republican deaths were caused by vaccine refusal.
And this is actually a very good answer to the argument you raised in your previous question. The virus becoming less lethal cannot by itself explain why the mortality curves of Republicans and Democrats, which were nearly identical before widespread vaccination, suddenly separated after vaccines became available."
John K Clark