An Agenda for Purely Confirmatory Research

This paper helped launch the preregistration movement in psychology — and it used ESP research as Exhibit A for why it was needed. The core argument: scientists too often blur the line between exploring data (poking around for interesting patterns) and confirming hypotheses (testing a specific prediction). When you explore first and then present your findings as if they were planned all along, your statistical tests become meaningless — they're rigged in your favor without you even realizing it. The fix? Preregistration: before collecting any data, researchers publicly post exactly what they plan to test and how they'll analyze it. Only those pre-planned analyses count as real confirmatory evidence. To walk the walk, the authors ran their own preregistered replication of Daryl Bem's famous 2011 precognition study, testing 100 women across two sessions of 60 trials each. Using Bayesian statistics (a method that directly measures how much the evidence supports one theory over another), they found strong evidence favoring the "no ESP" explanation — roughly 17 times more likely than precognition under default assumptions. This paper is a landmark because it simultaneously made the case for better scientific methods and delivered a concrete null result against psychic abilities.