Experimenter effects in laboratory tests of ESP and PK using a common protocol

This paper describes the fourth in a series of studies that explores the relationship between ESP and PK performance by testing for both using a common protocol so as to control for expectancy effects and experimental artifacts. Following earlier work (Roe, Davey & Stevens, 2003, 2004, 2005) we were particularly concerned to look for evidence of experimenter effects. Forty participants completed a computer-based greyhound racing game. Races occurred in two blocks of 12, with one block presented as an ESP task and the other as a PK task, though in fact each block included equal numbers of ESP and PK trials presented in random order. Roe and Davey each served as experimenter for 20 sessions and, after briefing each participant, would rate the interaction for warmth, spontaneity and positivity. Performance was non-significantly better than chance overall, but was significantly so for disguised ESP trials (p = .011). As predicted, participants who had been briefed by Roe performed better overall than those briefed by Davey; suggestively so overall (p = .085) and significantly so for Disguised ESP (p = .002). Some interaction measures gave promising correlations with task performance, particularly the experimenter’s confidence of success (rs = -.431, p = .007).