The Planet Money podcast did a terrific episode about science’s file drawer effect, and other ways that seemingly good science can be surprisingly wrong. They tell the story of Brian Nosek’s Reproducibility Project: do-overs for 100 well-known psychology experiments, which found that a majority of could not be replicated.
Not all science is so uncertain and unsettled, but psychology is particularly messy and difficult, and medical science is also probably more prone to this problem than physics, and even worse with musculoskeletal medicine. This is why I’m really not kidding around when I say that “one study means nothing” … especially if it’s one study produced by people who have something to prove or an ax to grind. Here’s a laughably perfect example: “proof” that a fascia-focused treatment works, reported without a trace of critical thinking.