(2020-01-16, 08:56 AM)Chris Wrote: The Parapsychological Association has made available another video of a presentation at its 2019 Annual Convention, also related to the methods developed by von Lucadou and co-workers. It's entitled "Micro-Psychokinetic Observer Effects on Addiction-Related Stimuli" and is presented by Moritz C. Dechamps of the Department of Psychology at the Ludwig Maximilian University of Munich. In the description on YouTube, Markus A. Maier is also credited:
These were experiments in which smokers and non-smokers were shown randomly selected cigarette-related and non-cigarette-related images. The initial hypothesis was that the proportion of smoking-related images shown to the smokers would depart from chance. A very strong decrease in this proportion for smokers (but not non-smokers) was indeed observed in the first study, but failed to be replicated in two subsequent studies.
This work looks at the question of whether the effect showed an oscillatory behaviour as the experiments proceeded. They do find indications of this, though perhaps that's inevitable to some extent, given a strong effect in the first experiment and non-significant effects in the others.
What surprises me is that all the results in the talk are couched in terms of Bayesian analysis, and that the psi hypothesis used is actually a two-sided one, because they say that initially they didn't know which direction any effect would take. (I don't think the fact that it's two-sided is even mentioned in the talk.) Given that they're analysing oscillations in an effect that was initially very strongly directional, and given that the plots shown in the talk are clearly labelled "Exploratory" in the paper, surely if they wanted to use a Bayesian analysis it would have been better to use a directional one?