In my view, the take-home message is slightly different from the BWAS paper. The latter argued you need large N.

This paper shows that if your target is unreliable, even very large numbers can’t save you

Also, association vs prediction

Comments