| From the ąrticle [1] references [2]: > fluoride intake levels (0.93 [0.43] vs 0.30 [0.26] mg of fluoride per day; P = .001). > Children had mean (SD) Full Scale IQ scores of 107.16 (13.26), range 52-143, with girls showing significantly higher mean (SD) scores than boys: 109.56 (11.96) vs 104.61 (14.09); P = .001. The average IQ of people in the UK is 99.12 and in Canada 99.52 [3]. Let us suppose that the hypothesis is true; it is expected then that in a country with higher average consumption of fluoride, we would, consistently, see lower average IQ. > Above-average consumption of tea, as recorded in Great Britain, could result in fluoride intakes as high as 8.9 mg per day [4]; We observe that the average consumption of Fluoride is about an order of magnitude larger in the UK, yet, IQs are almost identical. Which hints that if such a relationship exists, it is not as pronounced (5 IQ points) as the article claims. Maybe somehow women in the UK consume 20x less Fluoride when they are pregnant (what's the prob of this?), maybe people in the UK are just built different and their superior genes compensate and without the Fluoride they'd be at 104-105ish. Given history and similarity between societies, this looks unlikely. So until a causal mechanism is found, I am inclined to disagree over the "proven fact" statement. [1] https://www.nature.com/articles/s41390-020-0973-8 [2] https://jamanetwork.com/journals/jamapediatrics/fullarticle/... [3] https://wisevoter.com/country-rankings/average-iq-by-country... [4] https://onlinelibrary.wiley.com/doi/abs/10.1002/jsfa.2740340... |
Ecological comparisons are towards the bottom of the evidence hierarchy because there are any number of cofounders at play.
For example you could accidentally conclude that smoking is healthy because smoking tracks with affluence in poorer countries.
So I would reject the notion that we should or shouldn’t see IQ differences in very different populations which is why you want to do controlled research on different cohorts in the same population.
Finally, a causal mechanism is only a cherry on top but it’s not necessary for strong causal inference. Our mechanistic explanations are repeatedly wrong and/or inexhaustive. Fortunately, we can perform good quality studies instead like mendelian randomization.