I am a physicist. I do believe that a large percentage (less say 50%) of scientific publications do not meet basic quality standards, let alone the scientific method. It depends also on the level of the journal. But even in the best journals you can find articles and results that can not be reproduced. The pressure to publish is too strong. Anyway, 70% or more of the articles are only academic exercises, or do not have robust statistics or do not receive more than one or two citations (including a couple of my own). Only the best of the articles (should) prevail with time (Darwinism).