Against conventional wisdom

[Pierre Dragicevic]

Hi all,

I like the conclusion of this recent pre-print by a statistician who responds to the redefine statistical significance proposal:

"Against conventional wisdom, I contend that the reproducibility crisis cannot be fixed by better statistical education, increased awareness, patronizing ‘how to’ guides (e.g., [10]), or enhanced oversight by journals and intellectual bureaucrats. These measures have been tried, and they have failed. Such steps only accelerate bureaucracy, which in turn calcifies the status quo and further promotes a collective inability to conceive of what constitutes ‘knowledge’ independently of the bureaucratic policies used to evaluate ‘findings’. The only way to reverse course is to loosen—not tighten—the restrictions on what makes an analysis scientific and a finding significant."

I find it's in line with some aspects of the transparent statistics proposal, in particular with Matthew's call for not admonishing researchers, and my call for emphasizing the many ways we can do decent stats, and why reviewers need guidelines to help them stop rejecting papers for silly reasons.

Pierre