Now it is time to move from PK to ESP. Perhaps the most controversial of all types of ESP is precognition, ESP of the future. Honorton and Ferrari (1989) reported a meta-analysis of 309 precognition experiments published between 1935 and 1987.
The analysis summarized the effort of 62 separate experimenters and over 50,000 subjects. The overall p-value was a miniscule 6.3 x 10^-25 and the file drawer ratio was 46 to 1. Removing the 10% most extreme outcomes from each end of the distribution reduced the sample size to 248 and raised the p-value to 1.1 x 10^-9. There was no significant relationship between outcome and study quality (measured on the reduced sample), and the methodological quality was shown to have improved over the years. Thus the authors concluded that precognition is a real effect.
Blocking analysis confirmed the effect reported for PK dice experiments that subjects selected by prior testing performed better than unselected subjects. It was also found that the precognition effect was limited to subjects tested individually rather than in groups. Finally, results were better the more quickly subjects recived feedback on their scores and the shorter the time interval between their guesses and the random selection of the targets.