Statistical Problems in ESP Research

Back in 1978, Stanford statistician Persi Diaconis -- who also happened to be a former professional magician -- took a hard look at decades of ESP research and found it riddled with problems. His unique combo of statistical chops and stage-magic know-how let him spot both number-crunching tricks and physical ways information might leak to subjects. He nailed four big culprits: researchers stopping data collection the moment results looked good; testing tons of conditions but only reporting the hits; using "random" card sequences with detectable patterns clever guessers could exploit; and plain old sensory leakage through subtle visual, sound, or smell cues. The kicker? When experiments actually plugged these holes, the impressive ESP effects shrank or vanished. This hugely influential paper directly shaped the famous Hyman-Honorton debate and pushed the field toward pre-registration, computer randomization, and double-blind designs -- reforms echoed in today's broader reproducibility crisis across all of science.