Why Most Published Research Findings Are False
John P. A. Ioannidis · PLoS Medicine 2005 · doi:10.1371/journal.pmed.0020124
One paper, 84 independent model reviews: 3 text versions x 7 rubric configs x 4 judges. Flip one dimension at a time and watch what moves. 84 of 84 cells have a run; missing cells say "not run" and nothing is computed when you load this page.
Every judge was handed these 6 figure/table images alongside the text (identical across text versions):





