Two cognitive DMILS studies in Bali

Journal of Parapsychology, The, Fall, 2004 by Hoyt Edge, Luh Ketut Suryani, Niko Tiliopoulos, Robert Morris

Two cases (5% of total cases) were identified as univariate outliers because of their extremely high scores in either the Control or the Help periods ([z.sub.control] = 4.42, [z.sub.help] = 3.79, p < .001, two-tailed). By using the Mahalanobis distance (Tabachnick & Fidell, 2001) as an unbiased measure of the degree a value of a case differed from the average of all cases, we found that the response of one of the above cases in the Control period was also characterized as a multivariate outlier, [chi square] (1, N = 37) = 19.5, p < .001. Both cases were removed from the dataset, leaving 35 valid sessions.

For the analyses that required the use of the t-test, the effect size is reported through Cohen's r (Cohen, 1988), while for the ANOVA, effect size was measured through partial [eta.sup.2] (Tabachnick & Fidell, 2001). Power is always reported at alpha = .05 and beta = .20.

Planned Analyses

1. The primary hypothesis predicted that there would be a direct mental interaction between the Helper on the Helpee, shown through fewer button presses during the Help period as opposed to the Control period. Indeed, the mean of the total button presses in the Control period was 2.80 (SD = 2.13, range = 8), and in the Help period it was 2.06 button presses (SD = 1.61, range = 6). These means were significantly different from each other, indicating that on average there were fewer distractions in the Help period t(34) = 2.16, p < .02, one-tailed, Cohen's r = .35, power = .55.

2. The second hypothesis was two-fold and concerned those classified as more or less needy based on results from the two questionnaires.

2a. In the first part, it was hypothesized that high-need Helpees would experience a greater psi interaction score than the low-need ones, as they would be open to be helped by the Helper in focusing their meditation. Due to low observed frequencies, the Fisher-Irwin exact test was used on the cross-tabulated data of evidence of psi (yes/no) vs. needy Helpee (high/ low). The result seems to support the hypothesis (exact p = .013, one-tailed). Further analyses revealed that high-need Helpees tended to have significantly higher scores on the PIS index (M = .68, SD = .13) than the less needy ones (M= .33, SD = .37), t(15) = 2.86, p < .01, one-tailed, Cohen's r = .59, power = .92. This result was almost entirely attributable to the responses in the Control period, t(15) = 3.12, p < .005, one-tailed, Cohen's r = .63, power = .99, in which Helpees who were classified as more needy indeed appeared to have experienced a greater psi interaction score (M= 4.45, SD = 1.97) than those classified as less needy (M = 1.50, SD = 1.64). (5)

2b. The second part of this hypothesis predicted that Helpers who were classified as more needy would contribute a lesser psi interaction score than those classified as less needy. Following the same logic as above, the Fisher-Irwin exact test was applied on the cross-tabulated data of evidence of psi vs. needy Helper, but the results did not seem to support this hypothesis (exact p = .53, one-tailed).