“If you decide to write about this paper in your blog, please don’t mention my name. It is important to me to remain anonymous. . . . exposure might harm me or my students.”

So, this came in the email one day:

I am a social psychologist, and I was asked to review a PNAS paper . . . I don’t consider myself an expert in statistics, but I could easily detect three types of issues, which according to your blog are not uncommon in this journal: – Overstatement of findings
– The journal’s willingness to accept very weak findings (e.g., marginal effects, deviations from preregistration on the key analysis)
– Dishonest reporting: presenting exploratory analysis as if it was preregistered, presenting preregistered hypotheses that were not confirmed as if they were exploratory, deviation from preregistration that is not acknowledged (e.g., not using an important preregistered covariate that should have been used). . . . In my understanding, the paper creates the impression that a certain intervention . . . is effective, while in fact the evidence suggests otherwise. Personally, I spent a lot of time, effort, and money of funding agencies on an intervention that didn’t work, and I think that it’s important that as a field we move forward and abandon ideas that are wrong. . . .

My correspondent continued with details of various analyses in the paper that were not consistent with the preregistration and then added:

If you decide to write about this paper in your blog, please don’t mention my name. It is important to me to remain anonymous. . . . exposure might harm me or my students.

My reply: Yes, without looking at all the details, the general impression you give reminds me of other examples of this sort, unfortunately.

I’m not linking to the article here, partly to preserve anonymity but mostly because the point here is how commonplace this all is.