Skip to main content

Archived Comments for: The distribution of probability values in medical abstracts: an observational study

Back to article

  1. Not p-hacking, just publication bias

    Daniel Lakens, Eindhoven University of Technology

    27 November 2015

    Despite the suggestion this might be p-hacking, this is just publication bias. The authors follow previous studies that fail to use a formal model of the distribution of p-values. If they had taken the effort to use a formal model, it would have become clear there is no evidence of p-hacking. I've explained this in detail in this article: https://peerj.com/articles/1142/. We don't need low quality arguments suggesting there might be p-hacking on a massive scale. There is a problem, but it is very difficult to identify (see also http://daniellakens.blogspot.nl/2014/09/what-p-hacking-really-looks-like.html). 

    Competing interests

    No competing interests

Advertisement

\