Weakliem writes:
I [Weakliem] had a post on state differences in Covid rates around the beginning of September, another one around the beginning of October, and another around the beginning of November. I wasn’t planning on having another one, but then I read Bret Stephens’s column this morning. He said “Over the summer, as Covid cases started rising from their midyear lows, it became popular to blame Republicans for fueling the pandemic. The argument was that Covid had become a red-state scourge because of lower rates of mask-wearing and vaccination — along with high doses of vaccine misinformation — in places that went heavily for Donald Trump. . . . It even seemed true for a while . . . But the virus has had a way of making fools of us all.” He didn’t say the relation to partisan leanings had disappeared, but gave a few examples that seemed to point in that direction and said “let’s end the partisan blame games. They’re pointless, divisive, and dumb.”
The correlations between Biden’s share of the 2020 vote and Covid hospitalization rates at the end of the month:
Aug -.42
Sept -.63
Oct -.51
Nov -.25
The latest data:
As I’ve mentioned before, state rankings change quickly. Current hospitalization rates are positively correlated with hospitalization rates one month ago (.56) but have no correlation with rates in late September and a negative correlation with rates in late August (-.39). What if we take the sum of rates at the four times?
That’s a correlation of -.71, stronger than the correlations at any one time. Of course, it’s not directly due to partisanship, but to differences in vaccination rates. But state levels of vaccination have consistently been related to state differences in partisanship. Stephens suggests that the relationship between partisanship and vaccination is a matter of individual differences: “people have needs and ideas that differ from ours,” but they are at least partly (I would say “mostly”) the result of the leadership from party elites. And blaming political elites for policy failures is an important part of democracy.
At this point I’d say, “read the whole thing”—but I just posted the whole thing! It’s an interesting example of the statistics of aggregation. Also I agree with Weakliem that blaming political elites for policy failures is an important part of democracy. When Ted Cruz etc. are going around dissing vaccines, and then millions of people don’t get vaccinated, and thousands of people die, then, yeah, blame is appropriate. And I’m still accepting blame for coauthoring a paper called “Enhancing democracy through legislative redistricting.” I think Cruz is worse because he’s gotta know better, whereas we suffered from more of a lack of historical imagination. But still.
It’s easy to blame Cruz (who btw has been open about how he’s been vaccinated), but the anti-vaccine tendency in the GOP is more base driven than elite driven. I’ve seen multiple videos of Trump and Lindsey Graham get booed by rally goers when they promote vaccination.
Tucker Carlson has been super anti-vaxx, true, but I still think the bigger factor is that the GOP electorate is increasingly made up of non-college whites. Just see how red Appalachia (an area that’s been rocked by Big Pharma’s opioids and has been distrustful of authority even before that) has become.
There’s also one more thing that must be mentioned when discussing low vaccine takeup in America. The Biden-backed J&J pause, which permanently cratered vaccination rates. The daily vaccination rate peaked exactly before the pause , and then permanently collapsed
DZK writes, “It’s easy to blame Cruz . . . the anti-vaccine tendency in the GOP is more base driven than elite driven.”
The African American base of the Democratic Party is anti-vax too. Biden still pushed the vaccine. Political leaders can stand up to their base. The anti-vax movement use to be a lefty movement. The Republican party used to be anti-authoritarian and pro-democracy and pro-free trade. Political parties are coalitions that chose to emphasis which parts of their coalitions’ preferences they will promote. You have to blame leaders when they side with (not just tolerate) the worst part of their coalition.
While plenty of Republicans have dissed vaccines, I’m not sure Ted Cruz has. Best I could find from a cursory search was him arguing against vaccine and mask mandates and complaining about Sesame Street having Big Bird get vaccinated… Man, who has the energy to get worked up over Big Bird?
D.C. as an outlier is interesting. Not sure what it means (if anything), but it’s interesting.
One thing it means is that correlation is an inappropriate statistic. D.C. should be set aside and the correlations recomputed
I honestly don’t think anyone knows anything about COVID. I don’t think we have any good data. Literally none. I think the pandemic has been politicized so badly that data has been tampered with and metrics have had their definitions changed constantly. So while I think the plots in this post really ought to be deaths and not hospitalizations, it doesn’t make a bit of difference: nothing can be said at all.
You can definately say these waves are following a seasonal cycle.
While the tests are not very accurate for identifying an individual case, they do seem to measure the aggregate level of certain mRNA/protein in the community. Presumably that represents the amount of SARS-2 virus circulating.
Thanks for the thoughtful comment. I guess this means we can just rely on our gut instincts or political preferences to guide us. Or we can recognize that all data is imperfect (especially so with COVID), that the world is complex (statistics especially so), and do the best we can and be humble about it. What are you advocating we do?
Yep! Humility is good. I vote that the government stay out of the pandemic altogether, and let people and their own chosen personal medical providers decide for themselves what’s in their own best interests.
Except that this is a public health issue. Each individual deciding for themselves is likely to lead to a poor outcome for the community.
Should the government have stayed out of setting policy that led to quick vaccine development? Should it stay out of encouraging vaccination? Should it stop requiring vaccination for school enrollment (policy well before covid)? Should it stay out of public health entirely?
Adrian,
My gut feeling is that you’re kind of a nutcase, but my gut feeling isn’t always right so I’d like to find out a bit more. Perhaps your extreme statements are hyperbole.
You say “I don’t think we have any good data. Literally none.” Unless you are using “good” as a synonym for “perfect” this seems like an unsupportable claim. Even if you don’t trust any covid-specific data whatsoever, we still have data on all-cause deaths, by county, in each week. We have data on hospitalizations, by county, in each week. We have data on the number of people who have received a vaccination, by location…actually I’m not sure about the timescales for those data, maybe it’s by month and not by week…but anyway we have it. Even if you don’t trust any of the data on causes of death etc., we still have the data I’ve listed above, and others besides.
And of course we have had rather accurate COVID tests for quite a while now, and…well, are you actually denying that we can determine when someone died with COVID? I understand complaints about ‘died with’ versus ‘died of’, but when you say “nothing can be said at all”, that makes it sound like you don’t agree that we even know how many people have died -with- COVID.
Finally, your implicit claim that patterns like the one shown in the plots on this post are due to cooking the books seems…well, pretty incredible. I presume (perhaps falsely?) that you aren’t disputing the statewide Biden vote numbers, so you seem to be proposing that hospitalizations in the low-Biden-vote areas are overstated by a factor of 2.5, or that hospitalizations in the high-Biden-vote areas are understated by a factor of 2.5, or that there is some combination of overstating and understating that leads to that effect. Do you have any evidence for this whatsoever, or even a proposed mechanism for how the numbers in so many states would be manipulated?
Bertrand Russell was quoted as saying that when he met his Maker, he would say, “Sir, you did not give us enough information.” Kurt Vonnegut’s gloss on that quote was: “All the same, Sir, I’m not persuaded that we did enough with the information that we had.”
We have plenty of information to be actionable on COVID. Of course it is imperfect. In various ways, we have simply chosen not to make practical use of that which we can, but rather to beat each other with half-baked ideas ginned up to fit private prejudices. The willful ignorance and self affirmation has been dominated by—but not exclusive to—the political right. Whatever your politics, we could’ve done—and still could do—a lot better with the substantive information we DO have , and a lot better than we are almost sure to do with it going forward.
A shame.
Of course not. It is due hotter summer temperatures in the south driving people indoors, leading to a larger summer wave where there are fewer Biden voters.
What does the chart look like if you include Dec-Mar? The winter is when the Biden voters in the northeast got sent indoors.
Here you go. The third and fourth plots on this page are germane to your claim…which turns out to be completely false. The death rate in high-Trump counties was far higher in winter (of 2021 and 2022) than in high-Biden counties. Indeed, the difference was much larger in winter 2021-2022 than it had been a few months earlier.
Link is broke.
https://www.pewresearch.org/politics/2022/03/03/the-changing-political-geography-of-covid-19-over-the-last-two-years/
Thanks.
Obviously these numbers are far different from those in the OP (county instead of state, deaths instead of hospitalizations, etc). But whatever, the main problem is counting total deaths rather than the rate per population like people usually do. They should do a control plot of all cause mortality.
Those plots do show rates. And the quintiles are quintiles of population. The death _rate_ is much higher in Trump counties.
And yes, they’re deaths instead of hospitalizations…I think that’s better, don’t you? And yes it’s counties.
It’s the same basic story as in the states data on hospitalizations: areas with more Biden voters have a lower death rate, on average, and that’s especially true if you just look at the time period after vaccines became available.
Anon,
Those correlations with temperature are very weak. In our modern society there is very little difference b/w time spent indoors vs. outdoors, regardless of the season. Many self-proclaimed ‘outdoorsy’ types don’t realize it’s just hours at a time they spend in nature (then they get back to check their social media). People all go back home to sleep with the rest of the family, sleep eight hours, eat, go to work etc. Our work shifts don’t shorten in summer.
Same with vitamin D and such. Manaus, Brazil, is on the equator, with plenty of sunshine and plenty of covid mass graves. OTOH, African countries at the same latitude didn’t suffer as much (or nobody bothered to count). Go figure.
The biggest problem I have with graphs above is the population dynamic. Heck, all of those states left of 50% of Biden support would fit in grater LA area with a room to spare. Mixing of folks, and consequently an infection opportunity are not the same in Wyoming and downtown San Francisco. There are other variables the graphs would need to be adjusted for too.
Temperature is just a proxy for time spent breathing recycled air. And the correlation is amazingly strong, you have obviously never looked into this.
Both Weakliem and Andrew should consider ‘no linear correlation’ as an option. The DC outlier is driving the relationship. Anselm’s Quarter provides an old and, evidently, forgotten lesson about the fallacies of correlation.
https://en.wikipedia.org/wiki/Anscombe%27s_quartet
That’s just nonsense. On the second graph, if you cover up everything past 70 on the x-axis, there’s still clearly a straight line. DC actually dampens the trend line, not strengthens it.
There’s of course still questions about data quality and causal interpretation.
We responded to different scatterplots. The first scatterplot, once the DC outlier is removed, clearly shows no relationship, imo that is.
Reporting Pearson correlations in this case is inaccurate, if not misleading, not to mention that the Pearson is a measure of linear dependence only.
Everyone seems to be forgetting about the amazing natural experiment we’ve been running for the last two years on the interaction effect between political party and COVID spread: Congress! The sample is fairly balanced between Republicans and Democrats, and there’s no question who’s who. They skew older, but that’s true for both groups. Dems are more female, but that’s true in the population. They’ve been in-and-out of DC, but mostly back to their respective partisan communities, which maybe lends greater external validity. The sample’s far more partisan, and politically engaged, compared to the population, but we can think of that as giving an extra-large dose of the treatment so that effects are unambiguous.
What’s the result? Well, for the last year of Trump’s presidency, Republicans fell like flies. There were several months there when you’d see a new R reported as sick nearly every day. Dems, not so much. Then we had a within-subjects manipulation, on Jan. 6, when everybody was locked in a room together for several hours. Immediately after that, Dems, and particularly older, more vulnerable ones, started coming down with it. Things settled down, as I recall, after that–Dems cocooned again, Republicans got their own version of herd immunity, excepting the occasional tragic death.
Interestingly, the policies in both the House and Senate have been quite similar, and consistent over time, so we can arguably say we’ve controlled for the factor of masking/social distancing policies. And now that those policies are starting to get rolled back? I’ve noticed Dems popping up in the news with COVID again, including Pelosi. Just as you’d expect if enacting liberal health policies, which are actually followed, protects people.
I feel very comfortable concluding that political leaning has a strong, positive interaction effect with getting/spreading COVID, with the standard caveats about external validity.
“It even seemed true for a while . . . But the virus has had a way of making fools of us all.” –from the Stephens article.
In other words, people were getting sick more often in more conservative places, time passed, and then people were getting sick more often in more liberal places. We call this an interaction with time, or time trend, and it’s consistent with the hypothesis that COVID (and communicable disease generally) incubates in the most vulnerable places before spreading to the least vulnerable places.
Fools, indeed.
Michael, you say “In other words, people were getting sick more often in more conservative places, time passed, and then people were getting sick more often in more liberal places.” That’s backwards. The initial spikes were in New York City and environs, and some other major cities that are also left-leaning.
Check out the fascinating plots in the Pew Research link I posted above in response to Anoneuoid.
Yes, sure, communicable diseases spread more easily in some areas than others and this is one of the effects that we see. Another effect is that vaccination greatly decreases the risk of hospitalization and of death, in people of a given age and initial health status. There are also seasonal effects, and ‘herd immunity’ effects, and effects of school and business closures leading to less interaction between people, and so on so forth. There’s not just one effect.
Some of the effects are correlated with political beliefs. Vaccination is more common among Biden voters than among Trump voters. I suspect other effects are as well, such as mask-wearing and crowd avoidance and other factors…although I’m not so sure about crowd avoidance, since urbanites have a harder time avoiding crowded places and also tend to trend liberal.
The net effect is that, especially since the distribution of the vaccines, the death rate in heavily Republican areas has been substantially higher than in heavily Democratic areas, which is the opposite of the way it was for the first few months of the pandemic.
Phil, ya got me–that’s definitely what the data show. I should’ve added a caveat that major cities are an exception because they are both more liberal and more vulnerable to pandemics, particularly in the first wave, being the initial point of entry to the country for most international travelers, squeezing the population closer together, and having far greater volumes of people moving in and out generally. But, those are structural factors, only indirectly related to being liberal, if at all. I suspect if you redid those analyses controlling for such factors, especially once liberal policies for mask wearing and social distancing were put in place, the effect would disappear or reverse for urban areas. In other words, it’s an exception that proves the rule.
I also agree there are many other factors that can’t be partialled out, and I’d generally prefer not to ignore them except when they can be adequately measured or manipulated. But I’m not making an empirical argument. I’m making a rational argument, in which, for rhetorical purposes, I’ve acceded to Stephens’s apparent parameters for drawing his conclusions. His quote implies that he’s using geography of voting trends to predict geography of COVID cases. And that he’s concluded on the basis of that comparison that there used to be a pattern consistent with an effect of political ideology, but that pattern later disappeared.
I’m not endorsing his model, I’m giving him the benefit of the doubt. My point is that, even on his own terms, the change in those patterns *as a function of time* is actually consistent with an effect of political ideology on COVID rates. Stephens appears to be looking at the prevalence of cases by area aggregated across time, and I’m pointing out that you can’t do that.
What if I didn’t get vaccinated despite being in a state that went for Biden, because I’d rather die than have some doctor decide my risk tolerance for me? Should you just give ppl like me morphine so we can stop messing up your neat theories?
Not sure where this is coming from. It’s an empirical fact that, especially since vaccines became available, Biden supporters have been much less likely to die of Covid than Trump supporters, and that this is reflected in statistics of the statewide death rates. Why do you feel the need to be so defensive about this?
As long as I’m here I’ll share some of my own opinions.
Prior to the vaccines, I, and many other people, took rather extreme precautions against COVID…not because of our personal risk tolerance but because we didn’t want to endanger other people. I’m at very low personal risk of a really bad COVID reaction, but I interact with other people who are at much higher risk. In that context it seems extremely selfish to insist on making my own decision about risk tolerance, sort of like saying “if I want to drive drunk, that’s my decision, I should be able to decide how big a risk of killing someone I am willing to tolerate.”
Now that we have vaccines I think there’s a good argument for letting people decide for themselves a lot more. I think precautions such as masks should still be required in grocery stores and public transit — that is, circumstances that vulnerable people would have great difficulty avoiding — but they should be voluntary for more optional locations like restaurants, movie theaters, etc.
I completely agree with you pre-vaccine. Post-vaccine, I just don’t get it. Yes, we all have some obligation to avoid infecting the immunocompromised. But the immunocompromised are always in danger from everything! Is there any evidence that the immunocompromised have done any worse in the Covid era than in the eras before? Death and hospitalization rates among the immunocompromised are always much higher than the general population. If we are wearing masks on planes not to protect ourselves, or others in general, and if we agree that those who choose not to get vaccinated, at this point, are willingly accepting a higher risk of death, then why aren’t we always required to wear masks on planes in perpetuity to avoid accidentally infecting the immunocompromised with colds, flu or ten other communicble diseases? The continuation of masks mandates in planes (for example) now seems explictly to protect the immunocompromised (and to some extent, and to my way of thinking, unfairly, the unvaccinated).
It takes an unusually broad definition of “immunocompromised” if you want to encompass everyone with, say, greater than 1/10,000 chance of death if infected. I think that’s why covid is still different, even with the vaccines, in the eyes of many (including me). It’s one thing to say 1% of the population has to watch out for themselves, to the extent that they are taking a big risk if they fly in a plane or go to a concert or movie; quite another thing, quantitatively obviously, if that applies to 10 or 20% of the population.
I think most of society has had enough of doing things like wearing masks to protect the elderly and others who are at rather high risk even if vaccinated, even if it’s just doing only mildly unpleasant things like wearing a mask in the grocery store and the subway. I think that’s a great pity but I understand it.
“then why aren’t we always required to wear masks on planes in perpetuity to avoid accidentally infecting the immunocompromised with colds, flu or ten other communicble diseases?”
Here in Japan, everyone wears a mask in anything vaguely public. I wear a mask to take out the trash. This started in March 2020. Since then, life expectancy (in the country that already had the highest life expectancy) went up.
So, count me in for requiring everyone to wear masks all the time. It works.
By the way, it’s not just the immunocompromized. Spring 2019, I had a vicious flu. The highest fever I’d ever had for 4 days, and you really don’t want to know what the next week was like. Probably something a tourist brought, so your flu shot didn’t work, said my doc. It’d be real nice if we didn’t have to put up with BS like that, and for the nonce, here in Japan, we don’t.
Different people have different amounts of empathy and different tolerance for sacrifice. You and Phil both seem to have weighed these against one another, and he’s drawn the line a little farther down the line than you have. Granted, you can each put up a very strong argument for why your line is objectively better. But the reality is, for most people, disposition is far more determinative of rationale than the other way around.
One could argue that rsm just has a lower tolerance for sacrifice, or less empathy. Certainly he doesn’t seem to empathize with “some doctor”! I suspect, though, the difference between you two and rsm is categorical, not dimensional. Maybe the spectrum for rsm, for example, balances fear of losing self-determination against the fear of himself dying. Losing self-determination to “some doctor” is concrete, while dying, expressed as a (very small) probability, is abstract. Fear tends to prioritize the concrete over the abstract.
As an aside, I really do get the dark attraction some feel toward defiance and self-determination at all costs. Those who embrace it live in a much more exciting world than those of us who, sheep-like, line up behind the other cars, accept our drive-through injection, and sit in the parking lot until we’re told we can go. Truly, they are the heroes of their own stories. That’s why I’ve long thought the Dems should’ve “leaked” fake emails with Hillary and the DNC plotting to engineer COVID in a conspiracy to decimate red state populations. Then Biden could initially sign an executive order that no vaccines can go to states where he lost, only to “reluctantly” withdraw it. Then conservatives would rise up and defiantly demand the vaccine. That way, all their incentives would be aligned! It’s like nudging, but with conspiracy theories.
I agree that our definitions of immunocompromised are quite different. I’m not sure where your 1 in 10,000 definition comes from. The death probability of an *average* 66 year old pre covid is about 1 percent (and happens to be my cohort) and an average 72 year old is around 2 percent. https://www.ssa.gov/oact/STATS/table4c6.html The death rate for the immunocompromised (of the sort I mean… not anyone with a risk of dying from Covid) is way higher than that.
Jonathon (ao), I had in my head that the infection fatality rate is of the order of 1/1000 and that that is (empirically) high enough that, if a lot of people get infected at the same time, the hospitals get overwhelmed, healthcare quality drops, most people agree that something oughta be done, and so on. I divided by ten to get a number at which I figured the situation, though less dire, would still warrant action in most people’s eyes.
But OK, make it 1/1000 if you wish, or even 1/100. For any of those numbers, if you define someone as “immunocompromised” if they have that chance of dying given they become infected then you have a very large immunocompromised population.
People who have chosen not to get vaccinated, but go to crowded restaurants or nightclubs or whatever, have decided to assume whatever personal risk they’re subject to, and I’m OK with that. I think it’s not great from a societal standpoint — it’s as if everyone decided to take up motorcycling or some other high-risk activity, in the sense that it places a big burden on the healthcare system that we all pay for, and it contributes to burnout of health providers — but I’m on board with people being allowed to take that sort of risk.
But we we should (in my opinion) take simple steps to reduce the risks for vaccinated 80-year-olds who go to the grocery store. They can easily avoid crowded restaurants or movie theaters if they want, but it’s a lot harder to avoid buying groceries, or getting on the bus or subway.
Michael Nelson, I agree that the differences between me and rsm are probably matters of degree rather than either/or. We both presumably weight the importance of self-determination, danger to self, danger to others, inconvenience of masking, inconvenience of covid testing, etc., and our differing weights lead us to different behaviors. I also agree that there’s no objective way of determining the “right answer”: if I think he’s a selfish jerk because he doesn’t weight the risk to other people nearly as heavily as I do, and he thinks I’m a subservient sheep because I don’t weight the importance of self-determination nearly as heavily as he does, well, what can I say? Different people have different values. You heard it here first!
David Leonhardt and Michael Osterholm take my side — one-way masking is the way to go for the immunocomprised of any stripe: https://messaging-custom-newsletters.nytimes.com/template/oakv2?campaign_id=9&emc=edit_nn_20220422&instance_id=59157&nl=the-morning&productCode=NN®i_id=142695995&segment_id=89932&te=1&uri=nyt%3A%2F%2Fnewsletter%2Fb3b74c5d-8ded-5c02-be22-c1e3f2d267c1&user_id=c20bea7284649cf0dac919367927a03a
Lousy take from Leonhardt, imo.
Why is it so hard for (some) people to accept that (with fairly significant uncertainty) there’s a potential for marginal reduction of risk at the individual level to compound into a meaningful impact at the population level?
Then we get nonsense like…
> The most glaring example in the U.S. is that liberal communities, where masks are a cherished symbol of solidarity, have experienced nearly as much Covid spread as conservative communities, where masks are a hated symbol of oppression.
As if he can actually draw viable conclusions about the efficacy of mask-wearing by a confirmation bias hack at a correlation = causation hanging curveball. There are a lot of relevant casual variables there and he mentions nary a one (along those lines, I have a lot of sympathy to Brett Stephens’ take).
And then he goes on to largely of ignore all the uncertainty related to one-way masking and whether masks are as effective for protection to the wearer as opposed to source control.
Sad.
I’m in New Mexico at the moment. When I flew out here last week a mandate was in force, and I expected it to be in force when I fly back East next week.
Of course when I flew out, there were some people in the airports who weren’t wearing masks, a high percentage of the time. There were a non-trivial amount of people wearing masks under their noses. Yes, people in the plane (including myself), periodically pulled down their masks to eat and drink.
No kidding – the mask-wearing didn’t eliminate all the risks of flying 100%. But now, on my way back there will be a significant number of people in the airport and sitting in seats near mine in the plane who will not be wearing a mask at any point.
I have no reasonable way to calculate the actual amount that my risk will be changed because of the mandate being lifted. Maybe the risks will go from very small to just a tiny, tiny bit larger. Or maybe it will go from reasonably large to just a tiny, tiny bit larger. Maybe the change in risk will be so small that it’s only a tiny % of the overall, absolute risk.
I don’t know. But maybe the change in risk will be meaningful. And as such, the chances that I will get infected or pass the virus on to my 92-year old mother-in-law (who lives in an attached apartment) or my 5 year-old granddaughter (we’re her primary caregivers) will also be be meaningfully increased.
And sorry, given all the uncertainties it pissed me off when people use bad logic combined with outsized sense of privilege (as if wearing a mask for a few hours is some huge inconvenience or loss of freedom) to rationalize why the mandate should have been lifted.
Phil –
Reading this article:
https://www.nytimes.com/2022/04/23/health/mask-mandate-transportation-response.html
I thought of a discussion you and I had, where I said I think that selfishness wasn’t what really underlies the reasoning of people who don’t want to wear masks, rather the issue goes back to the perception of whether not masking is selfish.
I dunno. It’s a mistake to generalize from andecodtal sampling but maybe I need to shift a bit more in your direction after reading what Patrick McDonnell had to say.
I like the figures in the link posted at 3:56 PM today.
I wonder about two issues.
First, what physical and economic factors that affect the spread and severity of covid correlate with voting patterns? I feel that rural areas (1) tend to have more Trump supporters and (2) have poorer access to medical care. That correlation could lead to a relative undercount of covid cases and fatalities in areas with Trump supporters. Similarly, I can easily believe that there are regional and state-to-state variations in medical practices that may affect both diagnosis and outcome.
Second, to what extent are we seeing a form of forking paths? Could there be a different way of selecting modeled observations that would lead to different results?
For example, I found data on total covid deaths by state and Biden vote share by state. I used Mathematica to fit a linear model. This approach also showed a relationship between covid deaths/100K pop and Biden vote share—the higher the Biden vote share the lower the death rate/100K. It was a weak association—adjusted R^2 of 0.2.*
Bob76
*I am not a statistician. I took a graduate course in statistics about a half-century ago. I have never done statistics for a living. I have never used the Mathematica LinearModel tool before.
See the Pew Research link farther upthread. Super interesting.
I wouldn’t read too much into the data presented on the two graphs above. The range of data on both variables is quite small. It’s all relatively close.
Just take IL, CA and NY as examples of densely populated states. Well, 40-45 % are Trump supporters there (judging by the graph). Yet, they are ‘enjoying’ the benefits of overall lower hosp. rates.
Other, more important variables, are at play here for sure. The article is using political affiliation as a proxy for vaccination rates, negating the fact that age distribution among vaccinated is more of a predictor for hospitalizations.
Speaking of dense states. The thing I love about Florida is DeSantis railing against vaccines and then making sure that distribution sites were located disproportionately in affluent areas. It’s like the old joke, “This restaurant has terrible food! And such small portions!” :)
Using hospitalization rates, you cannot disentangle these patterns without state-wide data on the five important preconditions for getting a severe case of Covid. As I am observing from Europe, I guess it is reasonable to assume red states also have higher ratios of obese people. Not sure about the rates of smoking and diabetes, age and existing resp. symtoms (esp. asthma) should not be big factors. Just food for thought.