Skip to main content

Correcting the Past: Failures to Replicate Psi

πŸ›‘οΈ Critical replication β†—
Galak, Jeff, LeBoeuf, Robyn A, Nelson, Leif D, Simmons, Joseph P β€’ 2012 Modern Era β€’ skeptical

πŸ“Œ Appears in:

Plain English Summary

Can people somehow study for a test *after* taking it and still boost their scores? That's essentially what Daryl Bem claimed in his famous 2011 experiments on precognition. A team of four researchers decided to put this to the ultimate test, running seven tightly controlled experiments with over 3,200 participants β€” no peeking at results early, no wiggling the methods. The verdict? Six out of seven experiments found absolutely nothing. The combined effect was essentially zero. A broader look at all 19 known replication attempts told the same story. Here's the real kicker, though: the only thing that predicted positive results was whether Bem himself ran the experiment. When he did, effects appeared; when anyone else tried, they vanished. This became a landmark study in the replication crisis β€” raising hard questions about whether subtle researcher choices, rather than psychic powers, drove the original findings.

Research Notes

The most comprehensive replication and meta-analytic challenge to Bem's retroactive recall paradigm, with the largest combined sample of any single replication effort. Discusses researcher degrees of freedom as a possible explanation for original positive findings. A cornerstone of Controversy #2 (Bem FTF) and the broader replication crisis.

Across seven experiments (N=3,289), the retroactive facilitation of recall paradigm from Bem's (2011) Experiments 8 and 9 was replicated using computer-standardized delivery, predetermined sample sizes, and no data inspection before stopping. Six of seven experiments found no evidence of precognition; the combined effect was dβ‰ˆ0.01 with Bayesian BF=70.48 providing 'extreme' support for the null. A meta-analysis of all 19 known replication attempts (N=4,091) yielded an overall effect of d=0.04, 95% CI [-0.00, 0.09], indistinguishable from zero. The only significant moderator was whether Bem himself conducted the experiment (d=0.29 vs. d=0.02 for all others).

Links

Related Papers

More in Skeptical

πŸ“‹ Cite this paper
APA
Galak, Jeff, LeBoeuf, Robyn A, Nelson, Leif D, Simmons, Joseph P (2012). Correcting the Past: Failures to Replicate Psi. Journal of Personality and Social Psychology. https://doi.org/10.1037/a0029709
BibTeX
@article{galak_2012_correcting,
  title = {Correcting the Past: Failures to Replicate Psi},
  author = {Galak, Jeff and LeBoeuf, Robyn A and Nelson, Leif D and Simmons, Joseph P},
  year = {2012},
  journal = {Journal of Personality and Social Psychology},
  doi = {10.1037/a0029709},
}