Exploring the limits of science and beyond: research strategy and status

Journal of Parapsychology, The, March, 1994 by J.E. Kennedy

Majority-Vote and Goal-Oriented Psi

Studies of repeated-calling methods also provide support for the goal-oriented psi hypothesis and offer methods for further testing. Majority-vote or repeated-calling techniques will enhance the accuracy of psi effects if psi operates on the individual events comprising the majority vote. On the other hand, if goal-oriented psi operates directly on the experimental outcome or on each majority-vote outcome as one complex event, majority-vote procedures will not truly enhance psi accuracy. Here, too, the differences between models are clear and testable. The rationale and initial findings have been reviewed previously (Kennedy, 1978, 1979).

The goal-oriented experimenter-effects hypothesis is supported by evidence that majority-vote results are determined by the experimenter's expectations rather than by the majority-vote process. In a study of goal-oriented psi that directly compared majority-vote and single-event trials, Schmidt (1974) expected and found approximately equal (and very significant) scoring rates in both conditions, not increased accuracy in the majority-vote condition. However, in other studies, the experimenters expected and found that majority-vote procedures enhanced psi scoring rates (reviewed in Kennedy, 1978, 1979).

The studies showing enhanced scoring rates with majority-vote procedures obtained this result in a way that was more consistent with the goal-oriented psi model than with the traditional majority-vote model (see Kennedy, 1978, 1979). In six studies from three research groups, the psi effects were focused on the majority votes with narrower majorities.(5) These results correspond to low scoring rates (minimal psi effects) on the raw trials. For example, in the Brier and Tyminski (1970) study, the entire psi effect was due to majority votes of 3 out of 5, while votes of 4 and 5 out of 5 were not significant and were significantly different from votes of 3 out of 5 (Kennedy, 1979). Under the usual assumptions for majority-vote signal enhancement, votes of 4 and 5 should have produced higher scoring rates. This consistent pattern of results for the six studies is anomalous under the usual assumptions for majority-vote procedures and may be evidence for efficient goal-oriented psi.

Two recent majority-vote studies found a similar pattern that also supports the goal-oriented psi hypothesis. Puthoff, May, and Thompson (1986) and Radin (1990-1991) obtained significant psi results using sequential sampling majority-vote procedures. These procedures result in varying numbers of events or sequence lengths in the majority-votes. Both studies found that the psi scoring rate was significantly higher on the majority votes for longer sequence lengths than for shorter sequence lengths. As noted by Radin, this pattern is not expected under the usual assumptions for sequential sampling which result in approximately equal scoring rates for all sequence lengths. Radin suggested that the higher scoring on longer sequence lengths may result from some type of efficient psi operation. In fact, these data may be consistent with the pattern in the six studies discussed in the previous paragraph because, with sequential sampling, the longer sequence lengths have lower scoring rates on the raw trials (e.g., 14/20 = 70% compared with 4/4 = 100%). However, the interpretation of these data is complicated by the varying sequence lengths as well as by the sequential sampling procedure of discarding many trials.