Abstract

Focuses on the roles of testing and estimation in examining a database of ganzfeld research. In particular, power is computed for the 24 direct-hit studies with p = .25, for 2 different alternative hit rates. Also, confidence intervals are computed for the probability of a direct hit in each of those studies. The issue of power is discussed as it relates to the concept of replicability. Specific comments are offered on how meta -analysis is viewed by some statisticians. The overreliance on significance tests is discussed, and it is suggested that there should be more reliance on estimation. Distributional assumptions are also evaluated.