From: Rick Anderson <[log in to unmask]>
Date: Fri, 11 Mar 2016 01:49:29 +0000

>I would question this view of numerical significance, Rick. First, if
>you are going to say that three isn't enough, then someone will
>inevitably come back with the question "well, what is enough, then?"
>Whatever answer you give will be just as open to the charge of
>arbitrariness.

Not at all. That’s what sampling is for. Take a random sample of, say,
200 articles from PLOS One and see how many of them exhibit the same
issues that were found with the three problematic articles we’ve been
talking about. If you’ve taken a good sample you should get a valid
result, and no one would be able (reasonably) to accuse you of being
arbitrary.

Rick Anderson
Assoc. Dean for Collections & Scholarly Communication
Marriott Library, University of Utah
[log in to unmask]