# Articles to 2022-02-13

First the link to this week’s complete list as HTML and as PDF.

***

I’ve no idea what Dean is all about. The power, possibilities, and necessity of random sampling are taught to first semester introductory classes in all the social sciences. Random sampling for Covid prevalence has been suggested and advocated at least from March 2020, probably much earlier. If politicians and governments steadfastly refused to follow that recommendation, it must be that they did not want to know the truth. To tout random sampling now as a new and novel suggestion is beyond ridiculous.

***

Gibson et al. had already published a quite similar study on the same set of data that I critically commented on in the lists of 2020-07-26 and 2020-08-02. Here they look at a subset of some 13 thousand subjects they could track long time. Among those a little over 400 have become delinquent to a higher or lesser degree. The obvious thing would have been to look at those subjects’ blood lead levels taken years earlier and ask whether there might be a correlation. This is what Gibson et al. have not done. Instead they construct a complex multiple regression to predict blood lead levels. Nowhere, neither in the former nor the current article, do they tell us, how good their prediction is. As they do find a strong correlation with delinquency, the predictive value of some of the variables they actually do use seems to be quite good. We can’t easily tell which, but it is certainly not the binary choice between community water service and private wells – the only one mentioned in their title. While drinking water in childhood can plausibly be expected to be one of the major lead sources, the measured averages of blood level for the two kinds of source are 23.6 and 25.2 µg/l respectively, while the standard deviations inside the two groups are 17.8 and 16.5 µg/l. (The slightly larger variation incidentally and surprisingly, is the one for communal water.) The only prediction you can make from the kind of source is thus one of two values, both in the order of 25 µg/l. Instead they construct four groups spanning the range from less than 10 to more than 100 µg/l – and they completely fail to tell us the size of those four groups. From the resolution of their stated delinquency rate there seem to be at least about 100 in each leaving the possibility that 12 of the 13 thousand cluster in the one group of 20–40 µg/l. We’re not told and just don’t know. In any case the compelling results demonstrated in their figure 2 may be highly influenced by skilfully choosing the optimal group boundaries. One result comes out as absolutely certain, though. Whatever it is that influences and allows to predict delinquency, it is not the kind of water source. How this study passed any kind of peer review is a mystery to me.

***

Muik et al. is purely a laboratory result. As such it is unclear if and by how much it might be relevant epidemiologically.

***

I’m not sure what to make of Slimak et al. On the one hand it seems a sound and convincing result, I can see no reason to find fault with. On the other hand it looks totally implausible unless those moderns were transported there by Däniken’s space ships. The German yellow press quotes Jean-Jacques Hublin as rather skeptical about it. As so often I’ll wait for confirmation by at least a second similar find.

***

At first glance protecting the vulnerable against flu is a good thing and who can find fault with that? Looking closer that is not what Milkman et al. is about. Instead it is one more of the steeply rising number of studies in mind control. Centrally influencing the behaviour of the masses is a transferable task and, once mastered, easily adaptable to other causes. The authors themselves call this a “megastudy”, i.e. hugely expensive. It is financed by a few of the richest individuals on earth, those who already got rich by skilfully aligning spending preferences into their own pockets, and who now want to refine their capabilities even more. Every single one of those studies yields a modest effect size and does not seem that dangerous, but taken all together their sheer number amounts to something.

As an interesting aside the authors remark that lay respondents turned out much better at predicting the outcome than scientists in the field. Of course it’s easy to be clever in hindsight and I might be fooling myself, but for me the result seems utterly obvious. The wording of the winning reminder implies, that a vaccine dose is personally earmarked for the recipient and might be thrown out, if he failed to retrieve it. That eventuality tends to make us feel uncomfortable and spur us to prevent it. Of course no one could believe that, if he were to think about it. But the very point of these nudges is, that you don’t spend much thought on them and they work at the pre-conscious level.