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

Patterns over time. The forced-choice studies were reported between 1945 and 1982. There is a significant decline in the magnitude of the ESP/extraversion relationship over this period (r =-.40, r= -2.89, 43 df, p = .006, two-tailed). Moreover, methodological quality, as assessed in terms of threats to the validity of the ESP measure, has not improved over the survey period (r = .01, t = 0.03, p = .976, two-tailed). These results are contrary to the patterns found in meta-analyses of three other parapsychological domains, which exhibit constant effect sizes and significant methodological improvement over time (Honorton, 1985; Honorton & Ferrari, 1989; Radin & Nelson, 1989). There has been substantial improvement with regard to threats to the validity of the ESP/extraversion relationship; more recently reported studies have generally involved administration of the extraversion measure before the ESP task (r = .78, t = 7.02, 32 df, p [less than] [10.sup.-6]). These findings are consistent with the conclusion that the FC ESP/extraversion relationship is artifactual.

FC Outcomes by Investigator

The FC study outcomes by investigator are shown in Table 2. The order of ESP and extraversion testing is indicated following the investigator's name for investigators with single studies or multiple studies involving uniform testing order. Separate breakdowns are given for investigators with studies involving different testing orders. Significant outcomes were obtained by four of the 13 FC investigators (31%); using the investigator as the unit of analysis, the overall results are significant (z = 3.49, p = .00048, two-tailed) but nonhomogeneous ([[[Chi].sup.2].sub.12] = 41.20, p [less than] .05). The effect of testing order accounts for the overall significance and nonhomogeneity. The five investigators who measured extraversion before the ESP task have outcomes that are nonsignificant and homogeneous (z = -0.71, p = .761, two-tailed, [[[Chi].sup.2].sub.4] = 3.97, p [greater than] .05), while the outcomes of the eight investigators who measured extraversion after the ESP task are significant and nonhomogeneous (z = 3.51, p = .00045, two-tailed, [[[Chi].sub.2].sub.7] = 17.29, p [less than] .05). Thus, the impact of ESP/extraversion testing order is consistent across investigators and is not attributable to idiosyncratic research styles or other characteristics of a single prolific investigator.

TABLE 2
FORCED-CHOICE OUTCOMES BY INVESTIGATOR

                      N           N
Investigator       Studies     Subjects       r       .z       p

Astrom - ?            1            48        .24      1.63   .103
Casper - A            1            20        .53      2.46   .014
Green                 2           148        .00      0.00   .500
?                     1           108        .00      0.00   .500
A                     1            40        .00      0.00   .500
Humphrey              6           138        .26      2.86   .0042
?                     3            55        .27      1.84   .0658
A                     3            83        .22      1.95   .051
Kanthamani            7           301        .21      3.59   .00033
?                     1            60        .00      0.00   .500
A                     3           108        .38      4.03   .000056
B                     3           133        .02      0.25   .400
McElroy - A           1            31        .00      0.00   .500
Nash                  8           207        .14      1.92   .054
?                     2            60        .29      2.22   .026
A                     6           147        .08      0.86   .390
Nielsen               3            60       -.04     -0.29   .771
A                     2            48        .00      0.00   .500
B                     1            12       -.23     -0.69   .755
Sargent               3            85        .17      1.46   .144
B                     2            40       -.02     -0.12   .548
M                     1            45        .31      2.07   .038
Shields - ?           2            99        .30      2.98   .0029
Shrager - B           2            76        .18      1.48   .139
Szczygielski - A      1            17       -.36     -1.42   .922
Thalbourne - B        8           939       -.04     -1.28   .90

Overall forced-choice z by investigators = 3.49, p = .00048,
two-tailed, [[[Chi].sub.2].sub.12]= 41.20 p[less than] .05.

Extraversion measured before ESP test (5 investigators) z by
investigators = -0.71, p= .761, two-tailed, [[[Chi].sup.2].sub.4]
= 3.97, p [greater than] .05.

Extraversion measured after ESP test (8 investigators): z by
investigators = 3.51, p = 00045, two-tailed, [[[Chi].sub.2].sup.7]
17.29, p[less than] .05.

Notes. r is the weighted average correlation coefficient (Hedges &
Olkin, 1985). The letters B and A following the investigator's name
indicate whether the extraversion/introversion measure was
administered before (B) or after (A) the ESP task. Studies where
this information is not available are indicated by a question mark
(?). Testing order was mixed (M) in one study. [[Chi].sup.2] is the
within group homogeneity statistic (Rosenthal, 1984).