The Ganzfeld Meta-Analyses and the 1986 Joint Communique

By the early 1980s Honorton had accumulated several dozen ganzfeld studies and reported a pooled hit rate above the twenty-five percent expected by chance. Ray Hyman examined the same set and produced a critique organised around specific defects rather than general suspicion: inadequate randomisation of the target order, the possibility of sensory leakage between the sender's and receiver's rooms, multiple analyses of single datasets, and selective reporting of series that had worked. He also argued that the flaws he identified correlated with study outcome, which if true would explain the effect without invoking anything unusual. Honorton replied with a reanalysis disputing that correlation and arguing that the studies with the tightest procedures had not produced smaller effects.

At that point the exchange could have continued indefinitely, as such exchanges usually do, with each side re-scoring the same studies against its own criteria. Instead the two men wrote a joint document. That decision, rather than either meta-analysis, is what makes the episode worth an entry. It is worth noting how unusual the starting position was. Both men had already published their analyses, both had readerships that expected them to win, and neither had any professional incentive to sign a document conceding ground. That they did so anyway is the reason this exchange is cited in methodology courses that have no interest in psi at all.

The 1986 document does two things. It records an agreement, that the accumulated ganzfeld effect could not be explained by chance or by the file drawer alone, and it records the continuing disagreement about cause, with Hyman holding that an unidentified artefact remained the more probable explanation. Then it sets standards. Future studies were to use automated and documented randomisation, physical arrangements that made acoustic leakage between the rooms impossible rather than merely unlikely, full specification of the analysis before data collection, and complete reporting of every series begun rather than only those completed.

Read now, that list is unremarkable, because it describes ordinary practice in any pre-registered psychology experiment. In 1986 it was ahead of the mainstream discipline, which would not adopt comparable norms for another twenty-five years and did so partly under pressure from its own replication problems. The awkward legacy is that no ganzfeld series since has been agreed by both sides to have met the standard, and the two later meta-analyses that bear on the question, Milton and Wiseman in 1999 and Storm, Tressoldi and Di Risio in 2010, reach opposite conclusions on different study pools.