Meta-analysis of ESP Evidence - Precognition. Charles Honorton and Diane Ferrari (1989) published a meta-analytic study of all English-language forced-choice experiments conducted between 1935 and 1987 on the topic of precognition.
In a typical forced-choice precognition study, a person is asked to guess which one of a fixed number of targets will be selected later. The targets could be colored lamps, ESP card symbols, or a die face. Later, one target is randomly selected, and if the person’s guess matches the selected symbol, this is counted as a “hit.” In many such studies, immediately after the person guesses a symbol the target is randomly generated and presented as feedback. (Radin, 1997, p. 113)
Meta-analysis included over 300 studies. The meta-analysis included 309 forced-choice precognition studies reported in 113 articles by 62 different investigators constituting a database of nearly 2 million individual trials by more than 50,000 subjects.
The combined results of the 309 studies produced odds against chance of … ten million billion billion to one [1025]…The possibility of a selective-reporting bias – the file-draw problem – was also eliminated by determining that the number of unpublished, unsuccessful studies required to eliminate these astronomical odds was 14,268. Further analysis showed that twenty-three studies of the sixty-two investigators (37 percent) had reported successful studies, so the overall results were not due to one or two wildly successful experiments. In other words, the precognition effect had been successfully replicated across many different experimenters. (Radin, 1997, p. 114)
|
Trimmed analysis. After trimming away the outliers (i.e., the 10% of studies producing the largest effects and the bottom 10% of studies producing the smallest effects), Honorton & Ferrari’s (1989) analysis of the remaining 248 studies conducted by fifty-seven different investigators indicated that
The combined effect of the remaining 80% of the data still produced odds against guessing that many “hits” was an astronomical billion to one.
A non-significant positive correlation was found between study quality and precognition performance with the better-designed studies showing greater effects than poor-designed studies.
Study quality tended to improve over time with later experiments showing better experimental controls than earlier experiments.
The effect size (i.e., the number of standard deviations the results fall above chance) of precognitive results across time remained remarkably stable despite changes in study quality.
Studies using selected (gifted) participants produced larger effects than unselected volunteers (p< .001).
Trial-by-trial feedback produced better results than no feedback, time-delayed feedback, or trial-series feedback (p< .01).
The shorter the time intervals between subjects’ response and target feedback the larger the effect produced (p < .01).
|
