This abstract is opened in a separate window. Your initial window may be hidden behind this one
Robert M. Brier and Walter V. Tyminski
J. of Parapsychology, 34-1 (1970), p.26-36.
Two questions were raised in the just preceding paper: (1) How successful was the majority vote method in raising the percentage of correct predictions over the run-by-run method of scoring the data? Analysis showed that the majority vote procedure raised the percentage of correct predictions in six out of seven runs that were suitable for this purpose. The number of runs was too small for evaluation of the significance. (2) Did clear cut majorities (those on which the vote was based on 5 out of 5 guesses) predict more accurately than slim majorities (those in which the vote was based on 3 or 4 out of 5 guesses)? Analysis showed that a majority of 3 out of 5 gave greater accuracy of prediction (for both psi-hitting and psi-missing) than 5 out of 5. The difference in favor of the former was great enough to yield a p < 0.02. The latter effect was consistent at the level of p < 0.01. A trend (p ~ 0.05) toward the same favoring of 3-out-of-5 majority was found also in that part of the calls which was used as a sample to indicate the magnitude and direction of scoring before checking the predictions.