Problems with, and alternatives to, the NMR R factor