This reminds me of a similar point I first heard by Jordan Peterson, regarding how stereotypical judgements often have a true basis given the extremes of the di

This reminds me of a similar point I first heard by Jordan Peterson, regarding how stereotypical judgements often have a true basis given the extremes of the distribution with regard to a trait, but are effectively false with regard to the vast majority of the population under the curve, explained in a purely statistical manner. E.g. Why is it always X (e.g. men) that do bad thing Y (e.g. violence)? Well, it's because, if the mean of the distribution of X vs ¬X is shifted by even just a 0.00001, then at the super-rare-extremes of the distribution, X will still have a much higher probability than ¬X for that event at the tail. However, for the most part, at anything but the extremes, there's virtually no difference. So in many ways, stereotypes are simply the basic fallacy of mistaking a likelihood for a posterior.