Document Type: 
Document Number (FOIA) /ESDN (CREST): 
Release Decision: 
Original Classification: 
Document Page Count: 
Document Creation Date: 
November 4, 2016
Document Release Date: 
April 10, 2000
Sequence Number: 
Case Number: 
Publication Date: 
June 1, 1991
Content Type: 
PDF icon CIA-RDP96-00792R000400070001-6.pdf1.02 MB
Approved For Release 2000/08/11 : CIA-RDP96-00792R000400070001-6 Journal of Parapsychology, Vol. 55, June 1991 EXAMINATION OF SIX QUESTIONNAIRES AS PREDICTORS OF PSYCHOKINESIS PERFORMANCE BY LOFTUR R. GISSURARSON AND ROBERT L. MORRIS ABSTRACT: Data from five studies were examined for a possible connection be- tween scores on a computer PK test and scores on six questionnaires: Vividness of Visual Imagery Questionnaire, Auditory Imagery Scale, Gordon's Test of Visual Imagery Control, Greene Luck Questionnaire, Locus of Control Scale, and PK At- titude and Perceived Experience Questionnaire (PAPEQ). In none of the five stud- ies was significant psi-scoring encountered. The most promising finding was that the more subjects reported having "had a psychokinetic experience" on the PAPEQ the higher their PK scores tended to be in all five studies (weighted com- posite z = 3.03, p = .001, one-tailed). One study produced a strong positive cor- relation between PK scores and the sheep-goat factor of PAPEQ, z = 2.90, p < .005. This finding, however, was not consistent throughout the other studies. One way of exploring the PK hypothesis is to see whether scores on PK tests correlate systematically with some measurement of in- dividual differences, such as that obtained with paper-and-pencil tests. If a relationship is found, it can then be used to predict PK test performance, to select subjects, and to construct theories re- garding the possible processes involved in PK. The data reported in this article come from five studies carried out at the University of Edinburgh. In all studies, the subjects com- pleted a set of psychometric tests and then took a PK computer test called "Synthia" (Gissurarson & Morris, 1990). Overall, 170 sessions were conducted in the five studies, and the PK data were compared with the following six scalar instruments: (1) Vividness of Visual Im- agery Questionnaire (VVIQ) (Marks, 1973), (2) Auditory Imagery Scale (AIS) (Gissurarson, 1991a), (3) Gordon's Test of Visual Im- agery Control (GTVIC) (Gordon, 1949), (4) Greene Luck Question- naire (Greene, 1960), (5) Locus of Control Scale (I-E Scale) (Nowicki & Duke, 1974), and (6) PK Attitude and Perceived Experience Questionnaire (PAPEQ) (Gissurarson, 1989). The authors wish to acknowledge the assistance of John Beloff, Eric Dylan Darley, and Cynthia Milligan during the preparation of this paper. Approved For Release 2000/08/11 : CIA-RDP96-00792R000400070001-6- 0 CD 0- 0 0 0 0 0 0 03 0 CD 6 0 CD 0 0 0 0 0 0 0 0 0 6 120 The Journal of Parapsychology Imagery and PK When asked to describe their mental activity (mentations) during attempts to influence PK targets, subjects often mention engaging in specific target-related imagery (Gissurarson & Morris, 1991). A few PK studies have suggested that imagery may be connected with PK performance in controlled tests. Subjects who attempt to visualize the feedback they will receive for PK hits tend to obtain more hits than do subjects attempting other types of imagery or no imagery (Forwald, 1969; Levi, 1979; Morris & Reilly, 1980; Morris, Nanko, & Phillips, 1982; Nanko, 1981; Stanford, 1969, 1981; Steilberg, 1975; see also review in Gissurarson, 199 lb). Three studies have at- tempted to correlate imagery scale scores and PK performance (Stanford, 1969, 1981; Steilberg, 1975). All three used a free-asso- ciation test to measure the tendency to organize one's ordinary thinking around sensory imagery. Only Stanford (1969) found a sig- nificant relationship suggesting that the more subjects tend to use sensory imagery in their thoughts the higher their PK scores will be as long as they use a visual-imagery strategy in attempting to influ- ence their target. In an attempted replication, Stanford (1981) ob- tained a correlation of only r = + .03, although it was in the ex- pected direction. Steilberg reported nonsignificant results but did not provide any further information. For present purposes, the Vividness of Visual Imagery Question- naire (VVIQ) was selected to measure visual imagery rather than.. the more frequently used Betts Questionnaire upon Mental Imagery (QMI) (Betts, 1909). Honorton (1975) wrote: While the Betts QMI appears to be satisfactorily reliable, the failure of the test to relate significantly to a variety of verbal and visual recall tasks calls into question its construct validity as a measure of individual dif- ferences in vividness of mental imagery. (p. 330) Honorton concluded that a better measure of imagery for parapsy- chology studies was needed. George (1981) urged that future re- searchers in parapsychology should use "strongly validated mea- sures such as the Vividness of Visual Imagery Questionnaire" to evaluate vividness of imagery. Marks (1973) introduced the VVIQ, which was simply an expan- sion of the visual section of the abridged QMI (Sheehan, 1967). The VVIQ seems to be fairly reliable (Marks, 1973; McKelvie, 1986; McKelvie & Gingras, 1974; Rossi, 1977) and reasonably valid (Gur & Hilgard, 1975; Marks, 1973; McKelvie, 1979, 1986; McKelvie & Questionnaires as Predictors of PK Performance 121 Demers, 1979; McKelvie & Rohrberg, 1978; Rossi & Fingeret, 1977). Additionally, control of visual imagery was measured by Gor- don's Test of Visual Imagery Control (GTVIC). The GTVIC was designed primarily to differentiate autonomous from controlled vis- ual imagery (Gordon, 1949). It could be argued that the abilit.j. to control one's imagery is every bit as important as the ability topro- duce vivid imagery when attempting to engage in a specific P re- lated mentation strategy. The GTVIC seems to have adequate geli- ability (Juhasz, 1972; McKelvie & Gingras, 1974) and defines a single dimension in factor analytic studies (Di Vesta, IngersoT & Sunshine, 19'71; Forisha, 1975). n We know of no studies in the parapsychology literature that Rave explored the possible link between auditory imagery and PK sailing in those studies where there is an auditory component to thOtker- formance feedback. To assess the vividness of auditory image', a short questionnaire was designed, the Auditory Imagery Scale (IS) (Gissurarson, 1989; 1991a). Modeled after the Betts' QMI foaiat, the AIS has seven questions about the ability to imagine vaitous sounds, each of which requires a rating on a four-point scale (elow rating indicating high clarity and vividness). 0 Self-Perceived Luckiness 0 1:1 The Greene scale. Four studies have examined a possible relSon- ship between PK performance and self-perception of one's lucki- ness" as measured by the Greene Luck Questionnaire (Brouiton, 1979; Greene, 1960; Ratte, 1960; Ratte & Greene, 1960). The uues- dons on the Greene scale are, for instance, whether respondents' ex- pect to win or lose when it comes to games of chance, or whoth? er they have ever had the feeling that they cannot lose when plating a game of chance. The Greene (1960) study found a negatilm but nonsignificant relationship between self-perceived luck an PK scores. The Ratte (1960) and Ratte and Greene (1960) studieopro- duced a significant difference in PK scores in favor of self-perggived lucky subjects over self-perceived unlucky ones. Stanford (1977) however, argued that the statistical analyses in these two studieF were inappropriate. Broughton (1979) found a significant positiv( correlation between Greene scale scores and PK scores in his pilo data for subjects tested in a group, a relationship that turned ou nonsignificant but in the expected direction in the confirmator data. 9-1?000/00017000t1Z6/00-96dCIU-VI3 : 1.1./80/000z aseeletliOd peAoiddv 122 The Journal of Parapsychology Locus of Control. To further explore the possible connection be- tween PK ability and some sort of self-perceived luck, the present studies use a locus of control scale (Internal-External scale, or just I-E scale). The I-E scale is a forced-choice self-report inventory, which first came into prominence with the publication of a mono- graph by Rotter (1966) (see also Jackson & Paunonen, 1980, pp. 535-537; Lefcourt, 1976; Phares, 1976). Low scores on the I-E scale are thought to indicate that respondents perceive environmental events in general as if they are contingent on their own behavior (internal control). High scores are thought to indicate that the in- dividual perceives a general environmental event as not being con- tingent on his own actions but rather being the result of chance, fate, or luck (external control). Because external control implies self-perceived dependence on chance, fate, or luck, it would seem that this dimension could mea- sure some sort of self-perceived luckiness. The more internally con- trolled a person was, however, the more he would feel directly re- sponsible for, and the physical cause of, external events. Thus, there would be less and less opportunity for hitherto unrecognized means of interacting with the environment, resulting in no place for PK with high internally controlled people. We have found only one study in the literature that has explored whether high-PK-scoring individuals differed on an I-E scale from low-scoring individuals (Schmeidler, Gambale, & Mitchell, 1976). Using Rotter's I-E scale, Schmeidler et al. did not find any significant difference; however, they did not provide any information about the direction of the re- lationship, and it may be worth a further investigation. Nowicki and Duke (1974) attempted to improve the original Rot- ter scale, which had been criticized for being influenced too much by social desirability, for confounding different types of locus of control, and for difficult reading level. They published the Adult Nowicki-Strickland Internal-External control scale (ANS-IE). We have selected the ANS-IE for the present experimentation and refer to it hereafter as, the I-E scale. The expected relationship is that in- dividuals scoring high on the I-E scale (those who perceive them- selves as being dependent on luck, fate, and so on) will have more chance of obtaining a high score on a PK task. General Attitudes Regarding PK In an attempt to measure various characteristics of one's under- lying, general attitude toward PK, we designed a questionnaire, the Questionnaires as Predictors of PK Priforrnance 123 PK Attitude and Perceived Experience Questionnaire (PAPEQ; see Appendix A). The following factors were selected and incorporated into the questionnaire during its construction. I. Belief in the existence of PK. The first two questions of the PAPEQ are intended as sheep-goat items. The sheep-goat varOle, well known in parapsychology, was first introduced by Schmealler (Schmeidler, 1943; see also Schmeidler & McConnell, 1958)13.3She and others who followed found that subjects who accepted anyrs- sibility of ESP under the conditions of the experiment (thesetaub- jects were called sheep) tended to score above chance in ESP asts whereas subjects who completely rejected all possibility of"ESP (termed as goats) tended to score below chance (see review al- mer, 1978). We have found seven studies that have tested a aela- tionship between "belief' in PK and performance on a Pit test (Dale, 1946; Mischo & Weis, 1973; Nash, 1946; Van de Cfutle, 1958; Weiner, 1979, 1982a, 1982b). Only Weiner (1982a) delion- S strated a significant effect related to belief. Interestingly, two st1 ies have reported a positive relation between PK success and su cts' answers to questions about their belief in ESP (Rubin & Honoton, 1972; Watkins, Watkins, & Wells, 1973), but no mention is matte of subjects' belief in PK. The sheep-goat classification does not sei to have been adequately and systematically tested for PK. The r55ults so far are ambiguous, the reports are sketchy, and the numbar of subjects participating in these studies is low with the exceptia of Dale (1946). cr) O 2. Belief in one's own PK ability. The second intended fact z of the PAPEQ is the degree of certainty about one's own PK abkh.t) ies. There are two questions about whether the subject thinks he cpa she personally can demonstrate PK, in general and in this partgular o test. .P. 3. Luckiness. Three questions are on self-perceived luclgness (e.g., whether the subject has experienced his hopes or wishes 5j3out the future coming true). Here we attempt to get at a more g.-: eral self-perceived luckiness, as opposed to Greene (1960) who2on1y asked about luckiness in terms of betting and playing casino gianes. 4. Fear of PK. Four questions ask subjects about their fears of PK (e.g., whether the subject will be afraid of possessing PK abili- ties). This is an attempt to get at fear factors that may possibly block PK functioning, as suggested by Tart (1986a, 1986b), Batcheldor (1984), and others. 5. Prior experience of PK. One question asks how often, if at all, the subject has had a PK experience. 0 (D 0- 11 0 80 0 0 The present five studies were conducted as exploratory and ? ? O screening experiments at the University of Edinburgh and have not reported elsewhere. As screening studies, their aim was to se- ? lect subjects for further PK experimentation (see Gissurarson & co Morris, 1990). As exploratory studies, they were conducted to look for individual-difference correlates of first-session PK performance 6 and pretest intended experimental conditions, such as the use of co different random number generators (RNGs). 124 The Journal of Parapsychology 6. Previous involvement in PK-related activities. Four questions are concerned with activities indicative of a general interest in PK (e.g., if the subject reads books about psychic phenomena). Haraldsson (1981), for instance, used one such question in his ESP sheep-goat scale. 7. Willpower and success. Two questions ask subjects to evaluate their own willpower and success in life. No study seems to have gone on record to state whether it explored a connection between these variables and PK performance. We hoped that by asking a range of questions related to attitudes toward PK and its involvement in daily life, we could begin to ex- amine more thoroughly how performance in controlled laboratory situations might relate to perceived real-world PK functioning and associated mental life. METHOD 0 0 0 0 0 0 0 0 0 6? Study I Subjects. A predetermined number of 10 subjects participated. They were the research staff at the parapsychology laboratory and friends of the experimenter. PK apparatus. A PK computer test called "Synthia," written in BASIC for an IBM XT 286 16-bit machine, was used to measure PK. In "Synthia," four green rectangles (windows) appear in a row in the upper half of the computer screen (CRT). A pseudorandom number generator (PRNG), designed by Wichmann and Hill (1982, 1984; see also Gissurarson & Morgan, 1988, and Jacobs, 1987) and embedded in the computer program, produced a random designa- tion of one of the four windows. The Wichmann-Hill PRNG has a very large cycle length (6.95 x 1012) and produces numbers rectan- gularly distributed between 0 and 1. An arrow appeared beneath the designated window showing that it was the target. The PRNG Questionnaires as Predictors of PK Performance 125 selected a new target window for every 10 trials of a 30-trial run. On each trial, subjects were asked to "make the computer" select the designated window when they pressed the space bar. If the trial- selected random number matched the target window number, as tal- lied by the "Synthia" program, the trial was counted as a hit. > A blue star appeared on the computer screen and abeep sounded each time a hit was made during the feedback moae. Ir the nonfeedback mode, no such feedback was provided. All Infor- mation regarding each run was stored in an outfile (date4ime whether feedback or nonfeedback mode was being played, till des- ignated target window numbers, and the numbers generatzl or each trial).' Gissurarson (1989) provides a detailed discussiongif tin security measures that the program and laboratory offered ains possible human fraud or electrical bias. Only one fresh seed was selected for the PRNG for a 30-trgl rui at the moment when the test was initiated and before the fotg-win dow display came on the screen. After the selection of the8ingt fresh seed, which was based on the computer clock, the WichSann Hill PRNG algorithm automatically generated the 30 random-num bers needed, one by one at each press of the space bar. neve, . whole run of 30 trials was predetermined once the test was inritated The experimenter always initiated the program from the keAjoard Theoretical justification of this PRNG setup can be fold ii Schmidt's quantum collapse (QC) model (1982, 1984, 1987)rhic: is a refined version of his earlier model (1975a, 1975b). Cei6ra1 t the QC model is the assumption that it should be possible fir ht, man observers to influence the output of a RNG by affectlag th "collapse of the state vector" of binary probabilities. The QOnodt can be used to argue that the mechanism behind any RNGSesult in Study 1 was PK triggered at the moment of observation,lissun ing that the computer clock was an adequate randomizing stem. 'Two types of control randomness tests were run usually before, dug, ar after the studies. First, these included tests of the RNG for large series of Elmabex using the same algorithm (p = .25) as the "Synthia" program. For Studies 2-5, total of 4, 21/2, 3, and 9 million trials were made, respectively, the overall Gfifferent between duplets being insignificant for all four studies; x2 = 1.25 (p = .74), x2 6.25 (p = .10), )(2 = 2.75 (p = .43), x2 = 5.00 = .17). The exact results ha been misplaced for Study 1. According to the log-book, however, the one milli( numbers run before and after Study 1 also yielded insignificant differences betwee duplets. Second, Studies 2-5 were simulated via programs, which included a rando time interval between trials. For Studies 2-5, a total of 5, 17, 21, and 34 studies we run, respectively, with two significant studies found at the p = .05 level (two-taile for Study 4, one above chance and one below chance, which is about what one mig expect by chance. No simulated experiments were run for Study 1. 'The literature on PK research and PRNGs is growing (Braud, 1980; Gissurars, 126 The Journal of Parapsychology Psychometric material. Three scalar instruments were used: the VVIQ, the PAPEQ, and the Greene Luck Questionnaire. The Greene scale was edited for a U.K. sample (e.g., dollars were > changed into pounds), but all items were retained. -o Experimental rooms. Two rooms in the parapsychology laboratory -o O were used. One room was for filling out questionnaires. An adjacent partially sound-attenuated room was for doing the computer test. o. Procedure. The experimenter started by chatting with the sub- -n j o ect, then described the experimental session, its purpose, and set- ? up, followed by a description of the questionnaires and a demon- 2. of the PK test. Then the subject was left alone in the "ques- tionnaire room" to answer the three questionnaires in the following to ? order: the VVIQ, the PAPEQ, and the Greene scale. After corn- pleting the three scales, the subject and the experimenter went to g the sound-attenuated room where the experimenter initiated the -a computer test. The subject completed 60 trials on the computer test: Za 30 trials in the feedback mode, and 30 trials in the nonfeedback " mode. The subject was asked to take a break after the first run of O 30 trials and call the experimenter. After the break, the experimen- ter initiated the other mode of the test, producing a "fresh" seed *I for the next run of 30 trials. Half the subjects started with the feed- ? back mode first, and the other half started with the nonfeedback -0 cig mode. A flip of a coin by the experimenter decided for the first 6 subject which mode of the computer test he or she would start with. The second subject got the reverse sequence to that of the first sub- ject. This alternation continued throughout the series. g Study 2 A predetermined number of 40 volunteers participated in this ? study: (a) those responding to advertisements put up around the & Morris, 1990; Jacobs, 1985; Jahn & Dunne, 1987; Katz, 1983; Lowry, 1981; Radin, 1982a, 1982b; Schmidt, 1981; Shafer, 1983; see also Radin, 1985, on the practical use of pseudo-RNGs in parapsychology). Researchers in parapsychology have re- peatedly failed to find a significant difference between scores with random and pseu- dorandom targets (Schmeidler, 1987), suggesting that there may be a similar mech- anism responsible for the observed effects. Observed bias in PRNGs has been hypothesized to be the result of one of two functions: (a) PK affecting the system (computer dock or live RNG) that is used to generate fresh seed numbers that initiate the PRNG (e.g., Jacobs, 1985; Schmidt, 1981; see also theoretical arguments on this point in Vassy, 1985; and Walker, 1984), or (b) precognition of favorable moments for selecting these seed numbers (Radin, 1982a; see also May et al., 1985). At our current level of understanding, however, the actual cause of those biases we observe remains unknown. Questionnaires as Predictors of PK Performance 127 University of Edinburgh; (b) those who had indicated an interest in parapsychology research to someone at the parapsychology lab; and (c) those who came via participants already tested (each participant was given a copy of the advertisement to give to an interested friend). The procedure and the experimental environment we the same as those in Study 1. The AIS was brought in at this sa e in addition to the VVIQ and PAPEQ, and the Greene scale wt re- placed by the I-E scale (see Discussion). The four questionaaireE were administered in the following order: VVIQ, AIS, PAPE(gi and I-E scale. Again the computer test "Synthia" was used, with the-same set-up of the PRNG as in Study 1. The trials, however, we in- creased in this and all remaining studies from 30 to 40 per irTun tc a) provide more data and to make the statistical analysis more attrac- m tive. The experimenter initiated the PRNG as before. Study 3 n.) o o o o co A predetermined number of 10 subjects participated, wher'werc staff and visitors at the parapsychology laboratory and friends -of th( experimenter. Eight of the 10 subjects had taken part in Stiidy 1 five months earlier. The procedure and experimental envirokmen was the same as before. Three questionnaires were administead it the following order: VVIQ, AIS, and PAPEQ. The selection y th( initial "fresh" seeds was changed such that for every trial r nev fresh seed was automatically generated by the "Synthia" praran based on the computer clock. These seeds were then procesad 13,, 1 the Wichmann-Hill PRNG algorithm to produce the trial de ions one per trial. New initial fresh seeds via the computer doc wen also selected (before each 10-trial block) to determine which ithdov was to be the target for each of the four 10-trial sequences Thi meant that the subject's exact timing when pressing the spqe-ba, for the next trial was the key event in what random numbgr wa: generated. To distinguish between the two different set-ups y th( PRNG, we called the former version (used in Studies 1 *Acl 2 PRNG1 and the one used in Study 3 PRNG2. Study 4 Twenty volunteers participated, mainly people responding to ad vertisements about the study around the University of Edinburgl campus. The procedure and the environment were the same a those in previous studies. Four scalar instruments were adminis 130 The Journal of Parapsychology TABLE 2 PK SCORES FOR STUDY 5 BROKEN DOWN INTO THEIR VARIOUS R_NG COMPONENTS Live PRNG PRNG1 PRNG2 PRNG1-E PRNGI-S Feedback Hits 430 443 241 202 118 123 Trials 1840 1760 880 880 440 440 MCE 460 440 220 220 110 110 46 44 44 44 22 22 -1.62 0.17 1.63 1.40 0.88 1.43 Nonfeedback Hits 455 470 242 228 115 127 Trials 1760 1840 920 920 440 480 MCE 440 460 230 230 110 120 44 46 46 46 22 24 0.83 0.54 0.91 - 0.15 0.55 0.74 Combined Hits 885 913 483 430 233 250 Trials 3600 3600 1800 1800 880 920 MCE 900 900 450 450 220 230 90 90 90 90 44 46 -0.58 0.50 1.80 - 1.09 1.01 1.52 Note: PRNG1-E denotes PRNG1 when the experimenter initiated it; PRNG1-S de- notes PRNG1 when the subject initiated it. the studies yielded overall feedback z -0.27, overall nonfeedback z = 1.34, and combined total score z = 0.76.4 No significant PK scoring was found on any of the various RNGs used in Study 5 (see Table 2). Among the RNGs for all studies combined, scoring on PRNG1 when the experimenter initiated was the highest. Combin- ing z scores from PRNG1 (when the experimenter initiated it) for all four studies where it was used yielded an overall feedback z = 1.68, an overall nonfeedback z = 1.02, and a combined total score z = 1.88. No other RNG condition approached overall significance. 'The method is that of combining z scores weighted by some reasonable criterion related to the studies in question. Following the method of Mosteller and Bush as described in Rosenthal (1984), we weighted z = (Wizi + Wszs) / V W1' + W,, using the z scores associated with a given result. Each z was weighted by sample size. Questionnaires as Predictors of PK Performance 131 TABLE 3 SPEARMAN RHO (7''S) CORRELATIONS BETWEEN PK HITS AND SCORES ON SCALAR INSTRUMENTS Study VVIQ-PK, AIS-PK, Greene-PK I/E-PK GTVIC-W4 1 (n = 10) rs -.003 -.18 0 0.009 0.50 CD 0- 2 (n = 40) 0 rs -.15 -.06 .20 0.96 0.36 1.23 3 (n = 10) TS - .41 - .41 to 1.24 1.22 4 (n = 20) rs - .21 - .03 .35 0.93 0.14 1.53 5 (n = 90) rs - .02 - .07 - .06 .15 0 0.18 0.63 0.60 1.37 r," Note: For VVIQ, AIS, and GTVIC, Spearman r is correlated with feedback SPK scores only. Greene refers to the Greene Luck Questionnaire. IIE refers to thZ3I-E scale. Imagery, Luck, and PK 6 c.o The Greene Luck Questionnaire was used only in Study 1gee Table 3). Greene scale scores produced a nonsignificant neggive correlation, rs(8) = - .18 (z = 0.50), with the total PK score (raore self-perceived luck relating to lower PK scoring instead of the *er way around as was the case with Ratte, 1960, and Ratte and Greene, 1960). Some subjects voiced reservations about it, for example,Eitat people did not tend to patronize casinos in Edinburgh and there- fore the questions were irrelevant. Thus, it was decided to elimPlate this questionnaire from later studies. External locus of control was suggestively but not significantly correlated with PK scores in Stud- ies 2 and 4, rs(39) = .20 and rs(19) = .35, respectively, which was the expected direction. In Study 5, one question (Question 22) was eliminated from the I-E scale. It was the only question that showed a negative correlation with PK scores in Studies 2 and 4. Unfortu- nately, the correlation for Study 5 was slightly in the negative direc- 128 The Journal of Parapsychology tered in the following order: VVIQ, AIS, PAPEQ, and I-E scale. The computer test "Synthia" was used with the same arrangement of the PRNG as that used in Studies 1 and 2 (i.e., only PRNG1), initiated by the experimenter. Study 5 Ninety subjects participated, selected from the same three sources used in Study 2. One question was deleted from the I-E scale (Question 22; see Discussion) prior to its administration. The Gordon's Test of Visual Imagery Control (GTVIC) was brought in at this stage. The following scales were administered in the follow- ing order: the VVIQ, the AIS, the GTVIC, the PAPEQ, and the I-E scale. The following revision of the procedure was made: The description of "Synthia" and the demonstration game were provided not at the beginning of the session as were done in Studies 1-4 but after the subject had answered the questionnaires. Thus, any effect on questionnaire responses related to the subject's attitude toward the computer test was minimized. While preparing Study 5, we obtained a live-source RNG called RBG 04CA-S, which is based on an analog noise generator and pro- duces wide-band noise (reversed biased PN-junction noise, recom- bination noise, often called Zener noise). The RBG 04CA-S is made by the Synchronicity Research Unit in the Netherlands (for details, see Gissurarson & Morris, 1990, and User's Guide Random Bit Gener- ator RBG 04CA-S, 1988).3 Two versions were made of the "Synthia" program, a live RNG version and a PRNG version. Each version had 40 trials, and the trials could be run in either the feedback mode or the nonfeedback mode. The PRNG version included both PRNG1 and PRNG2. In the PRNG version, if the subject started doing the 40 trials with PRNG1, the program automatically changed over to PRNG2 after 20 trials, and if the subject started doing the 40 trials with PRNG2 the program automatically changed over to PRNG1 after 20 trials. Before the test was run, the program prompted for whether the experimenter or the subject would initiate the test. When the return key was pressed after this prompt, the test was run and initial "fresh" seeds were selected for PRNG1. For half of the runs, the experimenter initiated the test; for the other half, the sub- Because the random bit output is slightly biased, p(i) = p (0) = .5 ? .02, the User's Guide recommends performing a debiasing in software. After adding the rec- ommended debiasing procedure to the RBG, we tested it for over a million trials and no significant deviations from chance were found. Questionnaires as Predictors of PK Performance TABLE 1 OVERALL PK SCORES OF THE FIVE STUDIES 12! Study Feedback mode Nonfeedback mode Both mode> 1 (n = 10) Hits Trials 2 (n = 40) Hits Trials 3 (n = 10) Hits Trials 4 (n = 20) Hits Trials 5 (n = 90) Hits Trials 77 0.27 300 426 1.50 1,600 100 0 400 202 0.16 800 873 1.04 3,600 77 0.27 300 409 0.52 1,600 108 0.92 400 210 0.82 800 925 0.96 3,600 -o -on 154 2 0.38 a 600,, 835 a) 1.43 (sr: 3,200 02. 60 58 sas 800 oa 412 0 0.69 1,600 -0 1,798 ce, 0.056 7,200.9 ject initiated the test. The initial seed numbers for PRNGIgver? stored in an outfile. After the study, these were checked fk8 di( highest scoring subjects to make sure that the seeds recordetwert consistent with the random numbers generated. The necessary counterbalancing of the four conditions (feegback vs. nonfeedback; PRNG vs. live RNG; PRNG1 vs. PRNG2; tperi- menter-initiated vs. subject-initiated) was obtained by four cothflipt using procedures analogous to the one described earlier. RESULTS AND DISCUSSION Overall, 170 sessions were conducted in the five studies. In nom of the five studies was significant psi-hitting encountered (see Tabk 1). In general, scoring tended to be slightly higher in the nonfeed- back mode than in the feedback mode. Combining the z scores frorr 9-1?000/00017000t1Z6/00-96dati-VIO : l? 1?/90/000Z aseeieu JOd 130A0iddV 132 The Journal of Parapsychology don, rs(89) = - .06 (see Table 3). Weighted composite z score for the I-E/PK correlations yielded z = 0.26. The VVIQ score (lower score indicating better vividness) was negatively but nonsignificantly correlated with PK performance in the feedback mode in all five studies (see Table 3) as expected. An analysis combining z scores yielded z = 0.84. The GTVIC was only used in Study 5 where it also correlated in the expected direction with feedback PK scores, albeit nonsignificantly (higher score on the GTVIC indicating better imagery control). Finally, the AIS score (lower score indicating better vividness) was negatively but nonsig- nificantly correlated, as expected, with feedback PK scores for all four studies where it was used. Combining the z scores yielded z = 0.85. Of the 14 analyses in Table 3, none were close to significance, but 12 were in the expected direction, including the 10 related to imagery. PK Attitude and Perceived Experience Questionnaire Prima facie, the PAPEQ was intended to measure seven com- ponents that might be involved in an overall attitude germane to PK. An exploratory factor analysis was conducted, based on the pooled data, to verify that logic. (For those who filled PAPEQ out twice, which was done in Studies 1 and 3, only the second occasion was included in the analysis because on the first occasion the PAPEQ was still in its pilot form.) A principal components factor analysis of the PAPEQ showed that all the questions loaded positively on a single dimension (see factor loadings, means, and standard deviations in Appendix B). The variance in response to individual PAPEQ items overall was low, the standard deviation for only two questions exceeding 1.00. If we look at individual studies, the total PAPEQ score correlated nonsignificantly and inconsistently with the total PK score through- out Studies 1-5 (see Table 4). The total PAPEQ score without Ques- dons 5, 7, 16, and 17, which had the poorest loadings on the single component, yielded no improvement in correlations with the total PK score. It may be noted, however, that a homogeneous sample, such as the present one and as indicated by the low response vari- ance, produces a restriction of range for correlation coefficients, thereby reducing their power. Future researchers may want to use an unselected pool of subjects to ensure more variability in re- sponses. Questionnaires as Predictors of PK Performance 133 TABLE 4 SPEARMAN RHO (rs) CORRELATIONS BETWEEN PK HITS AND THE SEVEN FACTORS AND QUESTION 15 ON PAPEQ Study 1 Study 2 (n = 10) (n = 40) Study 3 (n = 10) Study 4 (n = 20) Stud5 (n =10) F1-PK (Belief in PK)' rs .16 .46 0.46 2.90** F2-PK (Fear of PK) rs .65 - .23 1.84 1.44 F3-PK (PK interest activities) rs .34 - .24 1.03 1.48 F4-PK (Luckiness) rs .14 - .02 0.43 0.11 F5-PK (Mindpower training) rs .57 - .17 1.50 1.05 F6-PK (Success on tasks) rs .12 .24 0.33 1.50 F7-PK (Wishing/willing) rs .15 - .09 0.46 0.55 Q15-PK (PK experience)" rs .37 .36 0.83 2.24* PAPEQ total` rs .25 -.04 0.74 0.27 PAPEQ without Questions 5, 7, 16, rs .33 - .07 0.92 0.42 .47 1.42 .09 0.27 - .03 0.09 .49 1.46 - .09 0.26 .34 1.01 .25 0.74 .41 1.24 .12 0.35 17 - .09 0.27 - .04 0.18 - .15 0.67 - .17 0.73 - .09 0.37 .14- 0.60 .03 0.12 .08 0.34 .32 1.40 -.19 0.77 - .04 0.17 0 < m - .030. 0.32m0 n X - .01M 0.04 U) a) .11 0.998 0 oa - 0.777% 0 .01: 0.05, 1:1 to co - .016 0.130 -4 CD n.) .0g 0.8? 0 0 .200 1.887j 0 0 -.0 0.67 - .05 0.48 'For Fl, Questions 1, 2, and 3 make up the sheep-goat scale. 'Question 15 had five possible answers (the range being 0-4). PAPEQ total refers to the connection between total scores on PAPEQ and PK. * p < .05, two-tailed. ** p < .005, two-tailed. 134 The Journal of Parapsychology The 18 PAPEQ items (see Appendix A) were rotated, using a simple structure orthogonal rotation, with factor loadings greater than or equal to .60, and the Kaiser criterion. Seven factors were extracted: Belief in PK (F1 = Questions 1-3), fear of PK (F2 = Questions 8, 12-14), PK interest-related activities (F3 = Questions 9, 10), luckiness (F4 = Questions 5, 7), experience of "mind power" training (F5 = Questions 6, 11), success on tasks (F6 = Questions 16, 18), and wishing-willing (F7 = Questions 4, 17). By and large, these factors are similar to those we had in mind when making the PAPEQ. One question, Question 15 (Have you had a psychokinetic experience?), did not relate to any of the separate factors, although it did load reasonably well (.58) on the single dimension. Looking at individual studies (see Table 4), F2, F3, F4, F5, and F7 correlated somewhat inconsistently with PK scores. However, Fl, F6, and Question 15 would appear to merit further discussion. Fl: Belief in PK. Three questions (1, 2, and 3) loaded greater than or equal to .60 on Fl. Question 3 had a factor loading of .606, which is marginally above the criterion level that was chosen, whereas Questions 1 and 2 loaded .87 and .86, respectively. Typical sheep-goat questions have been about overall belief in the existence of ESP/PK. Too specific questions (such as ones about personal abil- ity to demonstrate psi, as in Question 3) may perhaps be demanding different responses than do questions about overall belief in psi. Al- though it is debatable whether to include Question 3 in the sheep- goat scale, we decided to do so. (For comparison, Fl with only the first two questions included yielded: for Study 1, rs = .14, z = 0.40; for Study 2, rs = .57, z = 3.58; for Study 3, rs = .37, z = 1.11; for Study 4, rs = .01, z = 0.06; for Study 5, rs = ? .01, z = 0.07.). Only in Study 2 was there a significant relationship between the sheep-goat scale and total PK scores, rs(39) = .46, z = 2.90, p = .0037, two-tailed (see Table 4), although Study 3 is also encourag- ing. Combining z scores for the F1-PK correlations across studies yielded z = 1.01. The 90 subjects in Study 5 did the computer test with a complicated combination of RNGs. One may wonder whether there was any special RNG condition that correlated higher with the sheep-goat scale for the subjects in Study 5. As can be seen in Table 5, such was not the case. For instance, the PRNG1 condition (when the experimenter initiated the computer test), which is the same RNG condition as that used in Studies 1, 2, and 4, correlated non- significantly and in a negative direction with the sheep-goat scale, rs(43) = ? .18, z = 1.18, p = .24, two-tailed. There seem to be three elements that were different between Study 5 and the previous four studies: (1) The subjects in Study 5 Questionnaires as Predictors of PK Performance TABLE 5 SPEARMAN RHO (rs) CORRELATIONS FOR STUDY 5 BETWEEN SELECTED PAPEQ FACTORS AND PK SCORES, BROKEN DOWN INTO THEIR VARIOUS RNG COMPONENTS 135 Live-RNG (n = 90) PRNG2 (n = 90) PRNG1-E (n = 44) Sheep-goat scale (F1) rs ?.08 ?.002 ?.18 0.74 0.02 1.18 .47 .93 .24 Success on tasks (F6) rs ?.04 .04 ?.11 z 0.41 0.39 0.71 P .68 .70 .49 PK experience (Q15) rs .12 .24 .31 1.11 2.23 1.98 .27 .02 .04 PRNOV-S (n =46) m Note: All p values are reported as two-tailed. 0 were shown the PK computer test and had it described for Oiem after they completed the scales, whereas subjects in the otheiCfour studies had a demonstration of the test before they went thilugh the questionnaires. (2) The subjects in Study 5 had to conglete more questionnaires than subjects in the other studies did. (3The experimenter spent less time on subjects in Study 5 in compigison to the time he spent on subjects in the other studies. Study XI was conducted under time pressure, and the experimenter tend to run subjects quickly through the procedure. Future researcE will have to decide the significance, if any, these changes had mg sub- jects' responses and scoring on the PK test. F6: Success on tasks. The more subjects perceived themse1s as successful and able to influence the PK test on PAPEQ (F6; ues- tions 16 and 18), the higher PK scoring they tended to get incgtud- ies 1-3. As with Fl, the reason why this relationship disappeared in the remaining two studies is not clear and may simply reflect statis- tical regression. Combining z scores for the F6-PK correlations across studies yielded z = 0.63. The PRNG1 (when the experimen- ter initiated the computer test) condition in Study 5, which is the same RNG condition as that used in Studies 1, 2, and 4, correlated nonsignificantly and in the opposite direction with the F6 dimen- sion, rs(41) = ?.11, z = 0.71, p = .49, two-tailed (see Table 5). 136 The Journal of Parapsychology Question 15: Prior PK experience. The positive relationship be- tween Question 15 (whether people report having had a psychoki- netic experience in everyday life) and total PK scores was the only relationship that was consistent in all the studies. The weighted, combined composite z scores for this trend yielded z = 3.03, p = .001, one-tailed; and the weighted, combined estimate of the size of the effect yielded an overall r of .27, which is a decent correlation coefficient although not very high. As can be seen in Table 5, this relationship is also consistent across RNGs for Study 5, with the low- est correlation observed on the live-RNG. The PRNG1 (when the experimenter initiated the computer test) condition, which is the same RNG condition as that used in Studies 1, 2, and 4, correlated significantly and in the positive direction with Question 15, rs(41) = .31, z = 1.98, p < .05, two-tailed. This is perhaps the most prom- ising finding from all the studies, especially since it seems suffi- ciently robust to survive the diversity of conditions presented throughout the studies, including Study 5. Concluding Remarks Perhaps the most interesting and promising finding from the five studies reported here was that the more subjects reported hav- ing "had a psychokinetic experience" the higher their PK scores tended to be on "Synthia." It would be interesting if other research- ers attempted to replicate this finding. In the future, researchers could also follow this question through by developing other ques- tionnaire items in order to inquire more about these PK experi- ences, and perhaps gradually build up an effective self-report in- ventory in predicting PK performance. The apparent consistency of the relationship between prior PK experience and experimental PK success across a variety of RNG conditions suggests that there may not be radical differences in the psi process from condition to con- dition. It should be noted, however, that although this finding is encouraging, no correction has been made for selection, and the fact remains that it could be a statistical fluke. In the absence of any clear-cut findings, the implications of this for the different theoret- ical approaches to RNG-PK will remain unclear for the present. Questionnaires as Predictors of PK Performance 137 APPENDIX A PK ATTITUDE AND PERCEIVED EXPERIENCE QUESTIONNAIRE (PAPEQ) The proportion of subjects responding to each item is given in rounded percentages after rating categories. 0 Rating 0: g% 1: 24-% 2. Do you think that some people may be able to affect physical (7 conditions (or move objects or influence other people) with their "min*? a. Definitely yes. Rating 3: i% b. Yes, I think so. 2: fig% c. Probably not. 1: d. No. 0: @% 3. Do you believe that you can demonstrate the psychokinesis effect (lb., affect physical conditions or move objects or influence others with your) 1. Do you think that the existence of psychokinesis is: a. Impossible, b. Unlikely, c. Likely, d. Certain. "mind")? a. No, definitely. b. No, I don't think so. c. Yes, perhaps. d. Yes, definitely. Rating 0: 1: 2: 3: 5: g% 0 4. Do you experience your hopes or wishes about the future coming co true? a. b. C. d. Never, Seldom, Now and then, Often. 5. Do you consider yourself lucky? a. Not at all, b. Slightly, c. Fairly, d. Very. Rating 0: 1: 2: 3: 2% Rating 0: 718% 1: in6% 2: 49% 3: 28% 8% 43% 4% 6. Have you previously had experience of some sort of mind power training? a. Never, b. Once, c. Twice, Rating 0: 77% 1: 9% 2: 0% 138 The Journal of Parapsychology d. Three times or more. 3: 13% 7. Which of the following alternatives do you consider to be the best description of your luckiness/unluckiness? a. I am lucky in terms of getting what I want. b. I am lucky in terms of receiving unexpected gifts. c. I am very rarely lucky. d. I am not lucky at all. Rating 3: 66% 2: 19% 1: 12% 0: 3% 8. Would you be satisfied with yourself (or feel comfortable) if you were personally responsible for a PK event (for instance, if you were to break glass with your "mind")? a. Not at all, b. Unlikely, c. Likely, d. Certain. Rating 0: 11% 1: 24% 2: 38% 3: 28% 9. If you get the opportunity, do you then watch films like Poltergeist or read articles or books about people that have extraordinarily powerful influence/effect upon others with their "minds"? a. Never, b. Seldom, c. Now and then, d. Often. 10. Do you read books about psychic phenomena? a. Often, b. Seldom, c. Never. 11. Do you read books or articles on a. Never, b. Seldom, c. Now and then, d. Often. Rating 0: 17% 1: 24% 2: 41% 3: 18% Rating 2: 33% 1: 43% 0: 25% mind power training? Rating 0: 1: 2: 3: 12. Would you be afraid of possessing psychokinetic abilities? a. Yes, b. Probably yes, c. Probably not, d. No. Rating 0: 2: 3: 51% 20% 25% 4% 3% 24% 32% 41% 13. Would it bother you to directly witness a PK event (for instance, a table levitation)? Questionnaires as Predictors of PK Performance a. No, b. Perhaps, c. Yes. 139 Rating 2: 57% 1: 31% 0: 12% 14. Do you think you could easily get over it (and not be concerned agibut it in the future)? -cs Rating 0: O a. No, :08% b. Unlikely, c. Likely, 31:: :38: 2: g:41% d. Certain. -s X 15. Have you had a psychokinetic experience? CD a. Never, Rating 0: Fl% b. Rarely, 231: :: t c 808! c. Likely, d. Now and then, e. Often. 4: 03% Co 16. How successful in general do you consider yourself to be? a. I am definitely not a very successful person. b. I am not as successful as the others. c. I think I am a rather successful person. d. I am definitely a very successful person. 17. Which of the following statements best describes you? a. I am definitely strong-willed. b. I am moderately strong-willed. c. I am fairly weak-willed. d. I am very weak-willed. Which of the following statements best describes how task that you are about to participate in? a. I will definitely not be able to influence the test. b. I will probably not be able to influence the test. c. I will probably be able to influence the test. d. I will definitely be able to influence the test. 18. the Rating 0: "2% 1: ;-11. 6% 2: 3% 3:CD i310% Rating 3: iT4% 2: &6% 1: 00% 0: K.) 0% iyou feel alut Rating 0:g 3% 1:-467% 2:o28% 3: Es 2% Cr) 140 The Journal of Parapsychology APPENDIX B FACTOR LOADINGS, MEANS, AND STANDARD DEVIATIONS FOR THE 18 PAPEQ ITEMS Items 10 and 13 contain three options, and Item 15 has five. All others have four. Items Factor loadings Means SD 1 .56 1.95 0.69 2 .59 1.99 0.65 3 .71 1.42 0.74 4 .32 2.15 0.60 5 .23 1.97 0.86 6 .47 0.49 1.02 7 .22 2.49 0.81 8 .58 1.82 0.96 9 .36 1.60 0.97 10 .50 1.08 0.76 11 .46 0.83 0.96 12 .62 2.11 0.88 13 .61 1.45 0.70 14 .47 2.08 0.92 15 .58 0.64 1.14 16 .24 1.90 0.57 17 .13 2.14 0.57 18 .64 1.29 0.55 REFERENCES BATCHELDOR, K. J. (1984). Recent phenomena at Exeter. PK Messenger, No. 4, April. BE-rrs, G. H. (1909). The distribution and functions of mental imagery (Contri- butions to Education Series, No. 26, pp. 1-99). New York: Columbia University Teachers' College. BRAUD, W. G. (1980). Lability and inertia in psychic functioning. In B. Shapin & L. Coly (Eds.), Concepts and theories of parapsychology: Proceed- ings of an international conference (pp. 1-36). New York: Parapsychology Foundation. BROUGHTON, R. S. (1979). An experiment with the Head of Jut. European Journal of Parapsychology, 2, 337-357. DALE, L. A. (1946). The psychokinetic effect: The first A.S.P.R. experiment. Journal of the American Society for Psychical Research, 40, 123-151. Questionnaires as Predictors of PK Performance 141 DI VESTA, F. J., INGERSOLL, G., & SUNSHINE, P. (1971). A factor analysis of imagery tests. Journal of Verbal Learning and Verbal Behavior, 10, 471- 479. FORISHA, B. D. (1975). Mental imagery verbal processes: A developmen> tal study. Developmental Psychology, 11, 259-267. FoRwALD, H. (1969). Mind, matter and gravitation: A theoretical and expel-ren- tal study (Parapsychological Monographs No. 11). New York: Par&sy- chology Foundation. -n GEORGE, L. (1981). A survey of research into the relationships betweer'gim- agery and psi. Journal of Parapsychology, 45, 121-146. GISSURARSON, L. R. (1989). Psychokinetic attempts on a random event baseEmi- crocomputer test using imagery strategies. Unpublished Ph.D. thesis in$sy- chology, University of Edinburgh, Edinburgh. GISSURARSON, L. R. (1991a). Reported auditory imagery and its relationship with visual imagery. Journal of Mental Imagery, in press. 0 GISSURARSON, L. R. (1991b). Studies into methods of enhancing and pan- tially training psychokinesis: A review. Journal of the American Socia@for Psychical Research, in press. GISSURARSON, L. R., & MORGAN, K. (1988). Comments on the Wichmann- Hill random number generator. SRU Bulletin, 13, 99-100, 107- 1 on GISSURARSON, L. R., & MORRIS, R. L. (1990). Volition and psychokinFsis: Attempts to enhance PK performance through the practice of imagry strategies. Journal of Parapsychology, 54, 331-370. -0 GISSURARSON, L. R., & MORRIS, R. L. (1991). An experimental investigatit of assigned volitional mentation. Manuscript submitted for publication. 6 GORDON, R. (1949). An investigation into some of the factors that faeur the formation of stereotyped images. British Journal of Psychology,, 9, 156-167. GREENE, F. M. (1960). The feeling of luck and its effect on PK. J ourng of Parapsychology, 24, 129-141. 0 GUR, R. C., & HILGARD, E. R. (1975). Visual imagery and the discriminaxson of differences between altered pictures simultaneously and successimely presented. British Journal of Psychology, 66, 341-345. 0 HARALDSSON, E. (1981). Some determinants of belief in psychical phen@n- ena. Journal of the American Society for Psychical Research, 75, 297-30?t HONORTON, C. (1975). Psi and mental imagery: Keeping score on the Pktts scale. Journal of the American Society for Psychical Research, 69, 327-332. JACKSON, D. N., & PAUNONEN, S. V. (1980). Personality structure and as- sessment. Annual Review of Psychology, 31, 503-551. JAcoss, J. C. (1985). PK experiments with a true and a pseudo random number generator. Journal of the Society for Psychical Research, 53, 18-25. JAcoas, J. C. (1987). A compound pseudo random number generator with an extremely long cycle period. SRU Bulletin, 19, 92-96. JAHN, R., & DUNNE, B. J. (1987). Margins of reality: The role of consciousness in the physical world. San Diego: Harcourt Brace Jovanovich.