It seems to me that people put too much faith in statistics despite not having a fundamental understanding of the methods. I'm sure in plenty of cases these frivolous results get published due to academic fraud and the pressure of publish or perish. But in other cases, it seems like researchers really think "p<.05" translates to "your alternative hypothesis is irrefutably true" and will believe their flimsy conclusions because, as far as they're concerned, numbers don't lie. (If I remember correctly, when Brian Wansink was exposed for p-hacking, it was partly because he wrote a blog post openly describing his sketchy research methods, where he seemed totally unaware that he was churning out illegitimate results.)