[ZeroGravitas]

It seems to me that if it is the case that testosterone is responsible, then it would be much better science to identify that and have "high testosterone" vs "low testosterone" groups. You'd be able to predict more strongly if the evidence isn't binary but rather a variable.

This is similar to the claims that "race" predicts health issues, when actually it's usually some specific gene which is only correlated with race.

I don't see how it helps to introduce that extra layer of indirection.