The Wall Street Journal's Saturday essay today discusses the "reproducibility crisis" in biomedical research. One study found that initial positive results which were reported in the media were found to be false half the time. A study at Bayer found that they could replicate only 25% of the findings in various studies, and the head of cancer research at Amgen found they could reproduce only 6 of 53 promising studies, even with the help of the original researchers. (Links to the papers are in the essay).
This reminded me of what I remember as one of Jeff Novick's rules: your response to the results of a single study, no matter how exciting, should always be "hmmm, interesting."