A Statistical Artifact in William Braud's (1990) Experiment on Remote Mental Influence of Hemolysis

In preparing to conduct a conceptual replication of a generally well-designed experiment by William Braud (1990) that apparently demonstrated remote mental influence on the hemolysis of red blood cells, I studied the report in some detail. This examination uncovered a previously undetected statistical artifact. Our replication is reported elsewhere (Palmer, Simmonds, & Baumann, 2006). Braud (1990) tested 32 volunteers in his study. Fourteen of the participants (Ps) donated their own blood to be used in the experiment, whereas the remaining 18 attempted to influence blood that was not their own (presumably supplied by the other 14). P was alone in a room separated from the room in which the hemolysis procedure would be performed by the experimenter (E). The procedure consisted of transferring blood to a test tube containing a concentration of .425% physiological saline. After the contents were mixed by shaking, the test tube was placed inside a spectrophotometer that measures the hemolysis. When red blood cells burst, the solution becomes more transparent. The spectrophotometer charts the progress of the hemolysis by passing a light beam through the test tube and recording the amount of light that is absorbed at specified time intervals. Sixty hemolysis measurements were recorded during each trial, i.e., 1 per s. These measurements can be graphed, demonstrating the time course of the hemolysis during the run. Each session consisted of four 15-min trials. Two of these trials, as determined by a random process, were experimental trials, during which Ps were to concentrate on retarding the hemolysis taking place in the test room while having the opportunity to observe a photo of healthy blood cells if they so chose. During the control trials, Ps were asked not to think about the red blood cells, and if that was impossible, to concentrate on having the hemolysis occur at the normal rate. E was blind as to which two trials were the experimental ones. Prior to the first trial, Ps listened to a progressive relaxation and guided imagery tape over headphones. During each of the four trials, E performed the hemolysis procedure on groups of either two or eight test tubes, resulting in 10 tubes being tested in the experimental and control conditions, respectively. This manipulation was introduced to allow a test of decision augmentation theory (May, Utts, & Spottiswoode, 1995), which maintains that apparent micro-PK effects (under which rubric the hemolysis task would fall) are in fact caused by ESP. For example, the experimenter responsible for randomizing the order of experimental and control trials might use precognition to assign as experimental those trials in which the hemolysis effect was going to be stronger anyway. The theory makes differential predictions as to the results in the two- and eight-tube cases, hence the manipulation. The main analysis indicated no significant difference between performance in the experimental and control trials, nor did it make any significant difference whether Ps attempted to influence their own blood or someone else's blood. It also made no difference whether the scores were based on two or eight test tubes. However, there was highly significant variability among the scores of individual subjects. Of the 32 subjects, 9 achieved independently significant scores, whereas only 1.6 would be expected by chance, assuming an alpha criterion of .05, two-tailed. Of the nine significant sessions, seven were in the psi-hitting direction and two were in the psi-missing direction. The author concluded from the excessive number of significant sessions that PK had been demonstrated in the experiment. The main control introduced in Braud's (1990) study was the random ordering of the experimental and control trials, an order to which E was blind. Although this control is effective with respect to directional findings, it is not effective with respect to variance or bidirectional effects of the type Braud found. …