j/ -% f 'f 1 'I Michael Shermei's ESP Testing Score Sheet.
Before concluding that high scores indicate a high degree of ESP ability, you have to know what kind of scores people would get purely by chance. The scores expected by chance can be predicted by probability theory and statistical analysis. Scientists use comparisons between statistically predicted test results and actual test results to determine whether results are significant, that is, better than what would be expected by chance. The ESP test results clearly matched the expected pattern for random results.
I explained to the group, "In the first set, three got 2, three got 8, and everyone else [twenty-nine people] scored between 3 and 7. In the second set, there was one 9, two 2s, and one 1, all scored by different people than those who scored high and low in the first test Doesn't that sound like a normal distribution around an average of 5?" The instructor turned and said, with a smile, "Are you an engineer or one of those statisticians or something?" The group laughed, and he went back to lecturing about how to improve your ESP with practice.
When he asked for questions, I waited until no one else had any and then inquired, "You say you've been working with A.R.E. for several decades, correct?" He nodded. "And you say that with experience one can improve ESP, right?" He immediately saw where I was going and said, "Well. . .," at which point I jumped in and drew the conclusion, "By now you must be very good at this sort of test. How about we send the signals to you at the machine. I'll bet you could get at least 15 out of the 25." He was not amused at my suggestion and explained to the group that he had not practiced ESP in a long time and, besides, we were out of time for the experiment. He quickly dismissed the group, upon which a handful of people surrounded me and asked for an explanation of what I meant by "a normal distribution around an average of 5."
On a piece of scrap paper, I drew a crude version of the normal frequency curve, more commonly known as the bell curve (see figure 6). I explained that the mean, or average number, of correct responses ("hits") is expected by chance to be 5 (5 out of 25). The amount that the number of hits will deviate from the standard mean of 5, by chance, is 2. Thus, for a group this size, we should not put any special significance on the fact that someone got 8 correct or someone scored only 1 or 2 correct hits. This is exactly what is expected to happen by chance.
So these test results suggest that nothing other than chance was operating. The deviation from the mean for this experiment was nothing more than what we would expect. If the audience were expanded into the millions, say on a television show, there would be an even bigger opportunity
Was this article helpful?