The replication record is unambiguous and is the most important fact on this page. Stuart Ritchie, Richard Wiseman and Christopher French ran three direct replications of the retroactive recall experiment in 2012, found nothing, and had considerable difficulty publishing the null result, an experience that itself became evidence in the wider argument about what journals accept. Galak, LeBoeuf, Nelson and Simmons reported seven further failed replications together with a meta-analysis showing no effect.
Alongside the original paper, Wagenmakers and colleagues argued that the statistical approach was inappropriate for a claim of this kind and that a Bayesian reanalysis of Bem's own data gave no evidential support. Bem, with Tressoldi, Rabeyron and Duggan, answered in 2015 with a meta-analysis of ninety experiments reporting a small positive overall effect; critics read that as an accumulation of the same biases rather than as evidence.
What followed was larger than the dispute. Because the methods were standard, the paper became the discipline's worked example of how ordinary practice could yield a false positive, and it is cited throughout the literature that produced pre-registration, larger samples, open data and organised replication efforts. The claim did not survive. The reforms did.