At this point, I suspect that the majority of published results in any field cannot be replicated. A certain number of the ones that get a great deal of publicity[1] don't seem to pass a sniff test.

[1] https://news.ycombinator.com/item?id=7143630 [2]

[2] Before you go all "grr, journanists, grr", I first heard about that one on both the Science and Nature podcasts.