Comment by mycodesucks

My bet would be that a surprisingly large number of psychology studies are completely irreproducible. I'm even skeptical of a lot of fMRI experiments. Meta-analyses certainly can give a bit better insight, but have their own problems.