If you’re thinking about digging into the scientific literature with the hope of finding some remarkable discoveries, but you’re worried that novel findings move at a glacial pace, fear not. According to John Ioannidis and his colleagues in JAMA in 2016, 96% of the biomedical literature report statistically significant results.6Among 1.6 million abstracts and 385,393 full-text articles with p-values in MEDLINE (1990-2015), 96% reported at least one statistically significant result. “Significant discovery has become a boring nuisance,” Ioannidis says, in a Keynote address at the Lown Institute, in the same year. It’s therefore not hyperbolic to say that almost all scientific papers claim that they have found “significant” results.
So what does significant really mean in the scientific literature? Let’s backup slightly to answer this question.
The results of scientific papers are generally based on experiments. These experiments are statistical tests of hypotheses. A hypothesis is ...
It looks like you’ve tried to access a page only available to Nerd Safari members.
Not a member yet? Sign up for free.
Already a member? Login here.