## Anti-vaxxers and statistics

22 Nov, 2021 at 17:26 | Posted in Statistics & Econometrics | 17 Comments

If we make a Pearson’s correlation analysis on the variables in the scatterplot above — the corona vaccination rate and the (7 day) case rate per 100 000 people (data from last week) — we get an r = -0,15. And still in the youtube video below Sahra Wagenknecht says there is “keine Zusammenhang” between vaccination rate and case rate for the countries in the plot. Maybe someone ought to teach high-profile anti-vaxxers some basic statistics …

.

## 17 Comments »

1. May I post a (long) link to a graph showing a strong positive correlation between share of people fully vaccinated vs. Covid deaths per million? Why do African countries have such low vaccination, and death, rates? What expert predicted that? Is correlation causation?
.
https://cms.zerohedge.com/s3/files/inline-images/2021-11-24_16-47-53.jpg?itok=SFDruswb

• 《Source: Our World in Data (Confirmed Deaths per Million, November 19, 2021; Share of People Vaccinated against Covid-19, November 19, 2021).》

2. May I agree with comments by Crash Carson and Kingsley Lewis on this one?
.
Wikipedia says:
.
《Like many commonly used statistics, the sample statistic r is not robust, so its value can be misleading if outliers are present. Specifically, the PMCC is neither distributionally robust, nor outlier resistant. Inspection of the scatterplot between X and Y will typically reveal a situation where lack of robustness might be an issue, and in such cases it may be advisable to use a robust measure of association. 》
.
Is Lars doing the same kind of logical analysis backed by weak statistics that he so often decries in economics?
.
What if you included Africa? Can I cherry-pick sample countries, and get correlations to support opposite claims? Why ignore the huge US population relative to the other countries?
.
Is Lars’ analysis frightfully ergodic?
.
Don’t you owe us commons where we used to be able to escape your state prescriptions? What if some of us want to self-isolate outside, thus posing no threat to anyone?
.
Shouldn’t you declare full vaccination an asset on your central bank’s balance sheet, against which you can pay an inflation-protected basic income?
.
Why dictate the social contract based on shaky statistics? Why not negotiate with anti-vaxxers? How much of a basic income will you allow your central bank to create the money for?

3. If you only drop Austria, correlation will be positive.
I also think that causation is another issue!

• Interesting observation.
.
And it almost looks like there a two populations.
.
Perhaps the difference between the populations is the lockdown rules applying and how rigorously they were obeyed/policed?

4. Hi Lars, Sarah Wagenknecht is not an anti-Vaxer rather she is one of the few German political figures to try to sensibly discuss the issues. Maybe you should try listening to the whole talk first? Steven Jennings

Von meinem iPhone gesendet

>

• I have listened to the whole talk, and I have listened to Sahra appearing on German talk shows repeatedly the last year, and I do consider her — somewhat roundabout and veiled argumentation — anti-vaxx. She is an intellectual that has lost a lot of her reputation among radicals in Germany the last year precisely because she has not stood up for the necessity of people vaccinating if we are going to have a fair chance of dealing successfully with this epidemic. Her stance on this question has been deeply disappointing, not the least for those of us who actually used to admire her often sharp critique of things that were wrong in Germany when it came to social and economic issues.

• Ich kannst nicht Deutsch and I’m not familiar with Wagenknecht or her argument. It seems to me like the difference between “keine Zusammenhang” and a small, spurious looking correlation is not very meaningful.

Why such disparity between Italy and the Netherlands, or Austria and the US? What would happen to your correlation if you added or subtracted a couple of countries? Or picked a different time?

It seems pretty clear to me that the current vaccines aren’t going to do it. They help individual outcomes but they have little effect on infection, reinfection and spread. As such general mandates are unwarranted.

I am fully vaxxed and think most people would be better off if they were too, but forcing them is not helpful or desirable. Call me back when we have sterilizing vaccines.

5. I haven’t done any correlating in a long time but isn’t 0.15 actually a ver poor correlation? Making Wagenknecht right?

• Pearson’s r = 0 would mean there is no (linear) correlation (“keine Zusammenhang”)
Since r = -0,15 there is a correlation. Conclusion: Wagenknecht is wrong.

• Please give the full results of the regression. Is the Standard Errror of the slope significant? To my eye this is unlikely – “keine significant Zusammenhang” .

• There must be some kind of misunderstanding here. Wagenknecht does not pretend to make any inferences to a presumed population of countries based on her non-random sample of 15 countries. So why on earth should we be doing any significance testing at all?

• I agree with your comment.
.
But now I can’t see any substantive reason for your being interested in Pearson’s r = -0.15.
Is the only purpose of this post to merely make the narrow pedantic point that in statistics “keine Zusammenhang” implies r = 0?
Perhaps Wagenknecht’s “keine Zusammenhang” has a looser and more interesting common sense translation.

• What happens when a virus, pre-dating humanity goes novel [zip data] and incentivized private sector and captured political, see aforementioned are blinded by balance sheets and not science.

• It looks to me there as if are two populations:
.
Pop. 1. Austria, Nederlands, Belgium, Ireland, Portugal, Spain.
.
Pop. 2. US, Switzerland, Germany, UK, Norway, France, Sweden, Italy.
.
The two populations might be explained by differences in lockdown policies, how they were obeyed, how they were policed.
.
If the idea of two populations is accepted, then they each appear to have significant negative correlations factors.

• Environmental conditions will always proceed human reactions.

• Are you cherry-picking a cherry-picked sample?