Extraversion and ESP performance: a meta-analysis and a new confirmation - extrasensory perception

Journal of Parapsychology, The, Sept, 1998 by Charles Honorton, Diane C. Ferrari, Daryl J. Bem

FC studies involving individual testing. Twenty-one of the FC studies involved individual testing (Table 1, row 4). These studies were performed by 11 independent investigators and included 920 subjects. The mean weighted r is .15 (z = 4.54, p = .000006, 95% CI from .09 to .22). The correlations are nonhomogeneous ([[Chi].sup.2].sub.20] = 42.99, p [less than] .05).

FC studies involving group testing. The remaining 24 FC studies involved group testing. These studies were conducted by eight independent investigators and included 1249 subjects. They yield a mean weighted r of -0.00 (z = -0.12, p = .904, 95% CI from -.06 to .05). The group FC studies are homogeneous ([[Chi].sup.2].sub.23] = 37.35, p [greater than] .05). Thus, the FC studies involving group testing yielded uniformly null correlations between ESP performance and extraversion.

Individual versus group testing. Evidence for a relationship between extraversion and forced-choice ESP performance is limited to FC studies involving individual testing. The difference between the ESP/extraversion correlations for individual and group testing is significant (Cohen's q = .14, z = 3.47, p = .00052, two-tailed).

Free-Response Studies

The results for the 14 free-response (FR) studies are shown in row six of Table 1. The FR studies were contributed by four independent investigators and included 612 subjects. The mean weighted r is .20 (z = 4.82, p = .0000015, two-tailed). The 95% CI is an r from .12 to .28. The overall FR outcomes, while highly significant, are significantly nonhomogeneous ([[Chi].sup.2].sub.13] = 23.40, p [less than] .05). This nonhomogeneity is due to a moderating variable, test setting.

Free-response studies involving individual testing. Twelve FR studies employed individual testing Table 1, row 7). These studies were contributed by three independent investigators and include 512 subjects. The results are both significant and homogeneous. The mean weighted r is .20 (z = 4.46, p = .0000083, 95% CI from .11 to .29, [[Chi].sup.2].sub.11] = 15.85, p [greater than] .05).

Free-response studies involving group testing. Only two FR studies involved group testing (Table 1, row 8). Both studies were contributed by the same investigator. The mean weighted r is .19 (z = 1.83, p = .067, 95% CI from -.01 to .37). The results are significantly nonhomogeneous ([[Chi].sup.2].sub.1] = 7.53, p [less than] .05).

Free-Response versus Forced-Choice Studies

The mean correlation between ESP performance and extraversion is significantly larger in studies with free-response ESP tests than in those using forced-choice tests (Cohen's q = .14, z = 3.11, p = .0019, two-tailed).

Quality Analysis of Forced-Choice Studies

Sensory leakage. Thirty-one of the FC studies describe the use of methods that satisfy our criteria for adequate control against sensory leakage in the ESP task. The ESP/extraversion relationship is not significant for these studies; the mean weighted r is .04 (z = 1.47, p = .142, two-tailed). The ESP/extraversion relationship is significant in the 14 FC studies that are amenable to sensory leakage (r= .14, z = 3.20, p = .0014, two-tailed). The difference between the two correlations is significant (Cohen's q = -.10, z = -2.03, p = .042, two-tailed).