Statistical Problems in ESP Research

Foundational skeptical paper by a Harvard statistician who was also a former professional magician — dual expertise that gave the critique unusual specificity regarding both statistical artifacts and physical mechanisms for sensory leakage. One of the most influential criticisms of Rhine-era ESP research. Directly cited in debates that produced the Hyman-Honorton Joint Communiqué (1986). Pairs with kennedy_1981_skepticism_negative_results and the modern reproducibility-crisis literature (Simmons 2011, Ioannidis 2005, Button 2013).

Landmark statistical critique of ESP research by Harvard statistician Persi Diaconis, published in Science (Vol. 201, No. 4351, pp. 131–136). Analyzes four classes of methodological problems that generate spurious positive results: (1) optional stopping — analyzing data repeatedly and halting collection when significance is reached, without a pre-specified stopping rule; (2) multiple testing — examining many subjects, conditions, and measures then reporting only significant outcomes; (3) inadequate randomization — pseudo-random target sequences in Rhine-era experiments had detectable statistical regularities that subjects' guessing strategies exploited; and (4) sensory leakage — insufficient physical isolation in card-guessing paradigms provided olfactory, visual, and auditory cues. Concludes that better-controlled experiments consistently yield weaker or null results, and that statistical analysis alone cannot validate ESP claims without rigorous experimental design. Directly motivated methodological reforms including pre-registration, automated randomization, and double-blind protocols.