Tue 2021-Jul-06

Vaccinations vs Votes

Tagged: COVID / MathInTheNews / PharmaAndBiotech / Politics / Statistics

Back in April, we did some statistical analysis showing Trumpy states tended to have dramatically lower vaccination rates. Is that still the case? Regrettably, yes: more than ever.

Where We Last Left Off in April

Let’s recap our analysis of last April. [1]

Omnibus dataset: state, voting data, vaccine uptake/usage data We looked at, among other things, vaccination data from the CDC and the state popular vote margins for Trump vs Biden. We assembled all that from a couple disparate sources into an omnibus dataset. [2]

Hypothesis test significances: % doses used, % population vaccinated vs Trump Margin Discrete: % doses used, % population vaccinated vs Trump Margin Continuous: % doses used, % population vaccinated vs Trump Margin Using that dataset, we did a discrete analysis by $t$-test to see if the mean vaccination level differed between Trump states and Biden states by popular vote, and a continuous regression analysis of vaccination percent on the Trump - Biden margin. Both analyses came up statistically significant: the Trump states are less vaccinated (1st or 2nd shot), and this effect is highly unlikely to be due to chance. The effect size was large enough to be bothersome, as confirmed by public health officials.

A July 4th Meditation on Willful Ignorance on the American Right

So why revisit this now?

Basically, Biden promised (actually more like “hoped” and promised to “work real hard”) that 70% of the US adult population would be vaccinated by the 4th of July, so we could all have barbecues or something. That was ambitious, but with the vaccination rates seen in March or April, it could have been achieved.

Could have.

Since then, vaccination rates have leveled off and then steeply declined. Everybody who was easy to vaccinate has been vaccinated. We have a couple of problems:

  • The JnJ vaccine was taken out of use for too long, and people never quite understood the thrombosis risk compared to anything else. We quantified those a bit [3] [4] [5] [6], but this is just a crummy little blog with about 6 readers worldwide (one of whom is my cat). So that didn’t help much, other than personally, to soothe my mathematical itch.
  • We’ve reached everybody who’s easy to reach, worldwide. [7] Now we have to do serious outreach: in other languages, using local community leaders, using local trusted physicians, … basically using all sorts of outreach to communities that are sometimes justifiably suspicious of the US medical establishment. We’re doing all that, but…
  • Now we’re down to hard-core right-wing vaccine resistance, which verges on outright irrational, rage-filled refusal.

So of all those potential things, is the last one really a driver? Our state-level regression analysis in April said yes. We could do that again, but I was very gratified to find that Charles Gaba at ACA Signups (odd name for a blog!) had done this [8], and at the county level to boot! (All 3100+ counties in the US, which is… special.)

Here’s what his regression of % of total population vaccinated vs Trump/Biden % popular vote margin looks like:

Gaba: Vaccination level vs Trump margin, county-level data

Each data point is a county. Those further to the right voted for Trump, while those further to the left voted for Biden. The higher points had more vaccination, while the lower ones had less. Note that each axis goes from 0% – 100%, so there’s no suspicious zooming in to make small effects look larger; this is the whole picture.

Note the negative sloping trend: Trump counties did terribly worse on vaccination levels! The regression line confirms this:

  • The negative slope of the line is -0.4224. That is, for every 1% increase in Trump voters, a bit under half of them will refuse vaccination. Also, this is a stronger effect than we found at the state level, where the regression coefficient was at most -0.165. So the county data, updated from April to July, is even more convincing.
  • The model explains $R^2 \sim 42\%$ of the variance in the data, which is about as good as models like this ever get. Trumpiness is a good explainer of vaccine resistance.
  • The circles around each point appear to be the county population size. See the large circles around the blue points in the upper left? See the tiny, almost invisible circles around the red points in the lower right? That’s the US urban/rural divide.
  • Note the horizontal line labelled “TARGET: 53.9%”. This was Biden’s goal: 70% of American adults works out to 53.9% of all Americans, once you fold in children. The current vaccination level is 50.8%… missing by 3.1%, which is oh-so-very-close!
  • Note the horizontal line at 85% vaccination: Gaba’s estimating this as the level of vaccination (or previous infection) required for herd immunity. From that, we infer that he thinks COVID-19 has $R_0 \sim 6.7$ or so. This is apparently based on some data from the Yale School of Medicine. [9] I’d have thought that a bit high, but the Delta (and Delta+) variants are a bit more virulent. For once I’m insufficiently pessimistic! (However: if there were sufficient “natural” immunity from people who’d already had COVID, that would be added to the immune fraction. But in order for us to be at herd immunity now, with about 50% vaccination, we’d need an additional 35% of natural immune people. We don’t have that. Get vaccinated. It’s better than getting COVID!)
  • Important point: Nobody is yet at herd immunity! This means the unvaccinated cannot walk around hoping to be protected by the vaccinated around them.

Since the population of unvaccinated is slowly shrinking, but the case rate is leveling off, that means the rate of infection of unvaccinated people is going up. If you’re unvaccinated, this should terrify you. (But if you’re vaccinated, relax for now: especially the mRNA vaccines work against all current variants.)

So basically: there are a lot of knuckleheads refusing to get vaccinated for incredibly stupid reasons. They risk their own lives, sure. But more importantly, they are volunteering their bodies as collaborators against humanity in the war against COVID by being factories for new variants.

Sooner or later, one of those variants will escape our current vaccines. And do you know what happens then? We start all over again, having to re-vaccinate everybody. And I mean everybody, planet-wide, not just in the US.

All over again.

It was hard the first time, even when we played nice and just asked people to accept a free vaccine. Next time, we probably won’t play nice, and will have to use Jacobson v Massachusetts to impose a vaccination mandate.

The Weekend Conclusion

What to do? Honestly, I just don’t know.

Ever since Reagan, the Republicans in the US have become increasingly irrational. Today, they’re more or less fact-proof.

How do you persuade a fact-proof knucklehead to do the right thing for the safety of everybody, when they’d rather believe — and I am not exaggerating here — that the vaccine is a plot by Bill Gates to implant a chip so the government can reprogram us by 5G?

Thanks, Republicans. If not for you, we could have been more or less over COVID-19 in the US by now. But you’re acting as though you want a variant that’ll preferentially kill you, and then make us do this all over again. Why would you possibly want that?!

Addendum 2021-Jul-07

Last April, when we did the analysis of vaccination vs Trump margin, we found the next day that the New York Times published a similar analysis (with an almost identical plot).

WaPo: Republicans prevent America from reaching vaccination goal Not to be outdone, yesterday the Washington Post published an article by Max Boot, very similar to this post [10] (emphasis added):

In December, I argued that the coronavirus vaccination campaign would be a test of who had the upper hand in the United States: the geniuses (not all of them American) who invented and produced these safe and effective vaccines in record time or the conspiratorial crackpots who believe Internet rumors that vaccines are unsafe and unnecessary. More than six months later, it’s evident that the forces of ignorance and irrationality are so strongly entrenched that no amount of scientific evidence and public exhortation will sway them.

While young people and African Americans exhibit vaccine hesitancy, the most problematic group by far is Republicans. According to a new Post-ABC News poll, 86 percent of Democrats have gotten at least one vaccine shot, compared with only 45 percent of Republicans. Forty-seven percent of Republicans say they likely won’t get vaccinated, compared with only 6 percent of Democrats.

The states that have the lowest vaccination rates — Mississippi, Alabama, Arkansas, Louisiana, Wyoming — are all Republican redoubts. It’s no coincidence that many red states are also seeing the worst outbreaks of the disease in recent weeks: The average number of daily cases in the past 14 days has surged 145 percent in South Carolina, 137 percent in Nebraska and 121 percent in Arkansas. The Centers for Disease Control and Prevention calculates that 99.5 percent of U.S. covid-19 deaths in the past six months were among unvaccinated people.

Note that crucial fact: almost all the deaths in the last 6 months were among unvaccinated people. So… get vaccinated!

Sadly, roughly one-third of the country is impervious to science and even self-interest. That’s a big problem for the rational rest. Widespread vaccine resistance ensures that covid-19 will continue circulating as more virulent strains emerge. The only way we are likely to achieve “herd immunity” now is the hard way — by having a lot more people fall ill and die from the delta variant. The anti-vaxxers have a lot to answer for.

At least some of the big-time journalists seem to grasp the problem correctly.

Addendum 2021-Jul-09

Economist: worldwide vaccine slowdown and Trumpiness relationship In another instance of mainstream journalists catching on, there was an apparently anonymous article yesterday in The Economist on this subject. [11] Now, since it’s behind another evil paywall, I can’t tell you about the whole thing. But from the portion I can see, it says political views and suspicion of vaccines appear to be the causes of vaccination rates slowing down.

Their graphic (after all, it’s called the Daily Chart for a reason) shows 2 interesting things:

  • On the left, we see slowdowns in vaccination more or less throughout the developed world, hitting a hard plateau somewhere around 70% – 80%. That’s evidence of a minority of hard-core vaccine resisters. Apparently the Lizardman Constant [12] is 20% - 30% nowadays.
  • On the right is a plot similar to the one above, except using the % of people having gotten at least the 1st dose of the vaccine on the vertical axis. It’s county-level data, and shows just as damning evidence of Trumpiness as the explanatory variable. (Though, as a grizzled old statistician, I can’t resist pointing out that The Economist won’t tell you the regression coefficients, but I did. Hmpf.)

Again, people are catching on. Whether we can do anything about it is the main question!

Notes & References

1: Weekend Editor, “Politics vs mask use & vaccine uptake in the US”, Some Weekend Reading blog, 2021-Apr-19.

2: Weekend Editor, Omnibus dataset of state popular vote and vaccine usage, from “Why did Republicans block a Trump impeachment guilty verdict?”, SomeWeekendReading blog, 2021-Feb-24.

3: Weekend Editor, “JnJ Revenant”, SomeWeekendReading blog, 2021-Apr-23.

4: Weekend Editor, “Another study of clotting and COVID vaccines”, SomeWeekendReading blog, 2021-Apr-16.

5: Weekend Editor, “US pauses JnJ vaccine on thromboses”, SomeWeekendReading blog, 2021-Apr-13.

6: Weekend Editor, “The AstraZeneca/Oxford Vaccine vs Blood Clots”, SomeWeekendReading blog, 2021-Mar-17.

7: Weekend Editor, “The Billionth Dose”, SomeWeekendReading blog, 2021-Apr-24.

8: C Gaba, Happy Independence Day. Here’s U.S. #COVID19 Vaccination Levels BY COUNTY, ACA Signups, 2021-Jul-04.

9: C MacMillan, “Herd Immunity: Will We Ever Get There?”, Yale School of Medicine News, 2021-May-21.

10: M Boot, “Opinion: Republicans are preventing America from reaching Biden’s vaccination goal”, Washington Post, 2021-Jul-06.

11: Daily Chart, “Why has America’s vaccination programme slowed so much?”, The Economist, 2021-Jul-08.

12: Scott Alexander Siskind, “LIZARDMAN’S CONSTANT IS 4%”, Slate Star Codex, 2013-Apr-12.

Of this effect, Scott says: “I have only done a little bit of social science research, but it was enough to make me hate people.” This from the analysis of Public Policy Polling’s work on conspiracy theories finding that 4% of Americans believe ‘lizardmen’ are running all nations on the Earth, while another 7% are for some reason ‘not sure’ about the answer to that question.

Scott, a psychiatrist, is possibly the best rationalist blogger ever. For complicated reasons involving a hit piece in the New York Times which threatened to doxx him (community reaction summarized here), he’s moved his blog to Astral Codex Ten. Astral Codex Ten’s motto:

“P(A|B) = P(A) * P(B|A) / P(B)”, all the rest is commentary.

I cannot but think that both Rabbi Hillel and Rabbi Shammai would agree in spirit.

It’s worth your time. It’s certainly been worth mine.

Published Tue 2021-Jul-06

Gestae Commentaria

Weekend Editor, Wed 2021-Jul-07 13:27

Wikipedia has surprisingly good coverage of the variants of SARS-CoV-2.

As of today, the WHO classifies Lambda (aka C.37) as a “variant of interest”. That’s somewhat below the level of “variant of concern” that includes Delta (aka B.1.617.2).

It’s not just terminological wrangling, it’s based on transmissibility, morbitity, mortality, diagnostic evasion, antibody evasion, natural immunity evasion, vaccine evasion, and sensitivity in particular subpopulations. It’s a pretty complicated decision tree, but based on real stuff that will kill people if we ignore it.

It appears that Lambda might have some vaccine resistance, and spread faster than Alpha or Gamma. But the vaccine evasion has not yet been proven. And to my (partial) knowledge, Delta spreads faster anyway, so it will out-compete Lambda.

Delta appears to be the most easily spread, and has hence taken over the UK. As of today, it is more than 50% of the new infections in the US. That’s worrisome.

However, the evidence seems to be that the vaccines (especially mRNA) are effectve against all current variants. If we continue to breed new variants by having large, continguous populations of unvaccinated people, this will change for the worse.

Almost all the new infections are among the unvaccinated. In Israel, they showed Pfizer was about 64% effective against Delta infection, but 93% effective against hospitalization and death. So if you’re vaccinated, you might get infected but you probably will shake it off at home. But if you’re not vaccinated, you may die.

That should be terrifying to the unvaccinated. Maybe terrified enough to get vaccinated?

Comments for this post are closed pending repair of the comment system, but the Email/Twitter/Mastodon icons at page-top always work.