Why is PSI so elusive? A review and proposed model

Journal of Parapsychology, The, Sept, 2001 by J.E. Kennedy

After decades of involvement with parapsychological research, the founding statisticians expressed doubts that the experimental results were sufficiently consistent to meet the requirements for valid statistical analysis (Greenwood & Greville, 1979).

Braud (1985) commented that experiments designed to provide useful comparative information about the operation of psi seem less likely to get significant results than simple experiments designed primarily to provide evidence for the existence of psi. This observation is implicitly supported by the pervasive and otherwise inexplicable lack of control groups in prominent psi research such as studies investigating ostensible psi-enhancing properties of the ganzfeld procedure.

These findings and observations are more consistent with the hypothesis that the results are due to methodological artifacts than with the hypothesis that they indicate lawful but unexplained information transfer as conceptualized in parapsychological research.

Although specific methodological flaws that could account for recent meta-analysis findings have not been identified, the history of psychical research suggests that methodological problems are a likely explanation for the effects. For over a century, psychical research has been a continuing sequence of new test techniques heralded as definitive proof by their advocates and subsequently dropped from investigation when unrecognized methodological problems were found and attempted replications failed (Hyman, 1995). It can take years of detailed data analysis and attempted replications before the methodological problems are fully recognized. In addition, the fact that parapsychological findings are not cumulative and convergent suggests that there is not a meaningful underlying phenomena. For individual researchers and the field as a whole, each new research interest replaces previous interests rather than builds on them.

The lack of convergence is most apparent in the inconsistent results for specific research techniques. For example, the initial ganzfeld studies obtained significant results with pictures as targets. In a later series of studies that investigated the hypothesis that videotapes would produce better results than still pictures, the trials with videotapes were significant and the trials with still pictures were not significant, even though there was adequate statistical power based on the previous findings (Bem & Honorton, 1994; Hyman, 1994). These inconsistent findings suggest that the different results are produced by different factors, which raises the likelihood that the results are due to methodological problems rather than a common underlying phenomenon of information transfer (Hyman, 1995).

Other problematic issues related to the lack of cumulative findings include the lack of a theory or coherent explanation for the findings, the lack of an ability to predict when the alleged phenomenon will occur, claims for a wide diversity of manifestations (displacement effects, psi missing, position effect, etc.) that are unpredictable and difficult to distinguish from random fluctuations, widely differing results among investigators, and a negative definition of psi that is based on what it is not rather than what it is (Hyman, 1995).