lew@ihuxr.UUCP (Lew Mammel, Jr.) (02/01/84)
A week or so ago, I posted a comment about the psychokinesis experiment by Helmut Schmidt which was described on the NOVA ESP show. I noted that a net displacement of 120 after 6000 random binary steps was not statistically exceptional. Dean Radin has convinced me that the figure of 120 is NOT the displacement but the number of steps-to-the-right above average. This is 1/2 the displacement. A displacement of 240 steps after 6000 total IS significant. About one trial in 10,000 would be expected to achieve this. It seems that the NOVA graph is taken from one in the book, MIND AT LARGE which shows deviations of +160 and -144 out of 6400 "trials". This adds a new confusion since (according to Dean) the cumulative deviations are shown "after each section of 256 trials". Presumably a set of 256 trials means one session of 256 binary events. Of course, the division into sets of 256 raises the old spectre of selected data. I have obtained the source article of another graph shown on the NOVA show. It is by R.G. Jahn (Dean of the Princeton School of Engineering) and appears in Proceedings of IEEE, Vol 70, no.2, February 1982. It is a long survey article which includes a description of a REG (Random Event Generator) experiment conducted by a Princeton undergrad. This was cited on NOVA as one of the corroborations of Schmidt's results. I'll post some more comments about this article to net.misc. Lew Mammel, Jr. ihnp4!ihuxr!lew
Barker.PA@PARC-MAXC.ARPA (02/07/84)
Lew, I have followed this subject with keen interest, I have even obtained a gieger counter to fool around with. There are a couple of confusiong points about your last message. First you mention that someone has stated that the data presented in the graph is movement to the right above average. However, stasticly average is zero. If however you mean the average number of steps away from center in either direction that number is given by the square root of the number of total steps and is therefore about 77. Even so the data that I saw on the nova program was compared to a curve of random data centered around zero, as if the were implying that the deviation was away from a random walk that resulted in no net average displacement. I agree with your earlier computer simulation and have generated the same datsa myself. I have also evaluated the probabilities for a random walk according to the following formula P(X) = (1/2)**N*N! / (X!*((NX)/2)!) where N is total number os steps and P(X) is the prob. of landing on the Xth step. which gives the following results, # P(X) total P(all steps below X) step= +/- 0 .010300216203 .010300216203 step= +/- 1 .000000000000 .010300216203 step= +/- 2 .020593567883 .030893784085 step= +/- 3 .000000000000 .030893784085 step= +/- 4 .020572988035 .051466772120 step= +/- 5 .000000000000 .051466772120 step= +/- 6 .020538733975 .072005506095 step= +/- 7 .000000000000 .072005506095 step= +/- 8 .020490874076 .092496380171 step= +/- 9 .000000000000 .092496380171 step= +/- 60 .015261862398 .569012209115 step= +/- 61 .000000000000 .569012209115 step= +/- 120 .006205475647 .881743699915 step= +/- 121 .000000000000 .881743699915 step= +/- 240 .000169454944 .998139688874 Points to be made are that there is almost a 12% chance that the data will come out with a displacement above 120. Not hard to reproduce in a lab. The chance that it will be beyond 240 is .2 % not 1/10000. 1/10000 is closer to the probability that the displacement is on step # 240 ONLY