Direct and indirect mental interaction on random binary event generators: an updated meta-analysis

Abstract

Abstract

This updated meta-analysis evaluated evidence for direct and indirect mental interactions with random binary event generators (RNGs) in studies published after the meta-analysis by Bösch, Steinkamp, and Boller (2006). The aim was to estimate the overall magnitude and precision of reported effects, assess between-study heterogeneity and potential publication bias, and compare findings across direct intention and indirect field-based protocols.Studies were identified through database searches, specialist journals and conference proceedings, an AI-assisted literature search, and contact with researchers in the field. Forty-nine eligible studies were included: 18 direct-interaction studies, in which participants intentionally attempted to influence RNG outputs, and 28 indirect-interaction studies, in which RNG activity was examined in relation to collective human attention.Direct interaction studies showed a small positive overall per trial effect size d = 0.0017, 95% CI [0.0001, 0.0033], p = .041; I² = 0%. Indirect interaction studies showed a larger but statistically non-significant and less precise effect size estimate, d = 0.0006, 95% CI [-0.00001-0.0012], p = .064; I² = 25.2%).Moderator analyses of direct studies suggested larger effects among selected participants and in bidirectional-intention tasks, although these estimates were imprecise and based on few studies. Selection-model, excess-significance, and proportion-of-significant-results analyses indicated possible publication bias in both research paradigms.Overall, the findings suggest small deviations from chance in direct RNG-interaction studies, whereas evidence for indirect effects remains uncertain because of lower precision. Potential selective reporting and the limited number of studies constrain firm conclusions, highlighting the need for preregistered, adequately powered, and independently replicated investigations.