Abstract

From August 2000 to December 2018, two online psi experiments based on a five-target, forced-choice protocol together collected 114 million trials from an estimated 200,000 people around the world. The hit rate combined across both experiments was consistent with a null effect. However, a planned secondary analysis, designed to detect a predicted sequential pattern in the data, resulted in a small magnitude but statistically unambiguous outcome. With a chance expected rate of po = 0.32, the combined observed p1 = 0.320502 ± 0.000044, z = 11.28, p = 1.7 × 10-29. Control tests found no evidence that this small deviation, found independently in two separate experiments, was due to human behavior such as optional stopping or response biases, nor to computational or randomization errors that might have provided participants with subtle cues. If analysis of other forced-choice psi tests replicates this effect, it would suggest the existence of an unconscious process that tends to obscure accuracy in psychic perception. Gaining a better understanding of that process might lead to more robust results in psi experiments.