[efavdb]

I read one guess that 2/3 of the published results in social science are wrong. Suppose you tried to develop a deeper theory of these things and derived consequences from these “results” as one does in math and physics. If your corollary depends on 4 prior results, each with a 1/3 chance of actually being true — assuming no logical errors on your part — then the chance your result really is correct is (1/3)^4 <0.01. With results like this it’s not going to be easy to get much depth that holds water.