dir (10/30/82)
Considering the interest in the sugar-sand experiment, I'll give a few more details on the experimental design. The two bottles of sugar and sand weighed the same (1 ounce), were in identical opaque plastic bottles, and were packed with cotton to prevent the subjects from guessing the contents by shaking the bottles. The design was a counterbalanced, repeated-measures experiment, for those of you who understand the jargon. And the analysis was performed by a computer-based ANOVA package. The significance level of the main effect (sugar less than sand in strength) was p < 0.03, which is not astounding, but within the conventional probability level for significance. Another interesting result (at p = 0.008) was an interaction which essentially said that people who got the bottles in a sand-sugar order showed a much greater decrease in strength (35 pounds) than people in a sugar-sand order. I interpret this to mean that the effects of fatigue compounded the possible effects of sugar for people in the sand-sugar order. Since the results were somewhat unexpected, I'm having a colleague replicate the experiment with tighter controls on recording the data from the dynamometer (grip strength meter). I'll keep the net posted on results. Dean Radin - Bell Labs