Daniel Kahneman
Princeton University, USA
The Psychology of Scientific Overconfidence: The Case of Psychology
There is evidence of massive miscalibration in the probability that investigators attach to the success of their experiments — especially in between-subjects designs. Three contributing causes are likely involved: (1) the investigators inevitably have a within-subject perspective; (2) their hypotheses are almost certainly directionally true, though most likely to be weak; (3) intuitions about effect sizes are generally defective — a phenmenon we label „noise-neglect.“ As a consequence researchers are likely to overestimate their chances ex ante and to resist negative evidence ex post.