macaroni

@mccoy, #2 was just supposed to be an illustration of #1, nothing more. The relevance is that excess mortality cannot be ascribed to the concurrent pandemic based on the meer temporal association of the two, as you seemed to be saying.

YOU THINK I WILL AGREE THAT EASILY! Basically, all I am saying is this. You can use sample data to make statistical inferences about the original population from which you sampled the data, but you cannot make such inferences about a different population. In your example of smoking and lung cancer, you could say that someone who smokes likely has lung cancer. But if you consider only those individuals whose astrological sign is Aries or only those whose blood type is A then your previous observations alone dont give you any reason to say how likely it is that one of those individuals who smokes also has lung cancer. To do that you also need judgement.

I am afraid that this is completely wrong. The strength of the correlation cannot diminish the importance of your causal assumptions.

^No lol.

McCoy, causal inference depends on assumptions. These are necessarily judgmental. The data cannot provide nor test them.

Allow me to point out that there is nothing logical about deducing causation from correlation.