3

I would like to create a list of known problematic studies that misused statistics.
I tried searching google and read a few papers on the topic, but none of them contained specific examples of bad studies.

Is there a list of such papers? is anyone aggregating them?

Note: if there is a reason not to reference them I would also like to know why.

9
  • I would think "mistakes" are different than "misuse." which one are you looking for? Dec 19, 2020 at 18:29
  • @AzorAhai-him- At this point both (I couldn't find any), I want studies that have some kind of problematic statistical misuse or mistake in them that renders them false.
    – funerr
    Dec 19, 2020 at 18:32
  • I assume that no one has an incentive to keep such a list. Finding studies with such problems will likely be quite difficult. The only way to identify them that I can think of is by finding other studies pointing out statistical errors (here is an example), but finding those will probably also be hard.
    – cheersmate
    Dec 19, 2020 at 19:19
  • @cheersmate how did you find that study? cataloging them can help others see where mistakes were made and to be wary of references to those studies no?
    – funerr
    Dec 19, 2020 at 19:45
  • 3
    you could browse the archives of Andrew Gelman's blog (this will not be very efficient, although it could be informative).
    – Ben Bolker
    Dec 19, 2020 at 23:40

1 Answer 1

5

The closest I would know of are Retraction Watch and PubPeer.

Retraction Watch

First, Retraction Watch has a database comprising retractions of scientific papers. Click on the arrow next to the search field "Reason(s) for Retraction", and you will see possible candidates for statistical mistakes.

A list of reasons behind retractions, as shown in the Retraction Watch Database

PubPeer

Secondly, PubPeer allows anyone to write a 'post-publication review' responding to any scientific paper. Search for "statistical mistake" and you will find many cases in which readers raised concerns about published findings - some of them having led to formal errata, corrigenda or retractions. Here is one example, where the reader starts an extensive response with "I have no comments to make about the data, but I do have concerns about some of the statistical analyses", and goes on to list issues about the methodical approach of that paper.

Not the answer you're looking for? Browse other questions tagged .