Researcher Woes

Study: More Than Half of Psychological Results Can't Be Reproduced http://science.slashdot.org/story/15/08/28/0228224/study-more-than-half-of-psychological-results-cant-be-reproduced Found that headline interesting. As a researcher reading A-level publications, and working with A-level researchers to publish my own, I can understand why. Research involving people is tricky, and when you try to find "truth" using a sample, you are usually trying to prove some theory. If you work hard enough and long enough, you can make your sample line up with some theory, but I think many researchers are misleading themselves that they have definitely found truth in the process. I myself have tried to duplicate the research of others and gotten different answers. In my discipline, we hardly ever verify the work of others, and that is a problem. One problem is this, you can test a theory 100 times and have it fail each time and you can never publish it. If you get it to be more or less right just once, however, you can publish that paper. Thus, reality often gets buried behind the chance results. Also, the exceptional case result gets published, whereas the normal case likely does not because a non-finding isn't considered a finding worth publishing in most cases. The literature therefore gets distorted.