Why is the statistical revolution not progressing? Vicious cycle of the scientific reform

To analyze the results of the research, behavioral scientists widely use a statistical rule that sets the significance level to 0.05. Recently, two recommendations on how to improve statistical inference were published: to redefine statistical significance to 0.005, and to select and justify the alpha. We analyzed the empirical work that cited the original recommendation papers, as well as the papers published by the scientist that co-authored the publications. About half of the numerous papers citing these recommendations adhered to them already in the first year since their publication. What is striking, the original authors that proposed the recommendations followed their own recommendations only in 6% of their empirical work. We surveyed the authors asking them to identify major obstacles they experienced while trying to implement their own recommendations, and obstacles they think others could expect or experience.