,
7 tweets,
2 min read
Read on Twitter
@GSCollins brought this up yesterday (citeseerx.ist.psu.edu/viewdoc/downlo…), so I decided to give this another read, and was struck by how accurate Altman's list of five troublesome questions at the very end was. These all spoke to me very much as a reviewer:
1) How much of a purist should the referee be (especially if it is unlikely that the 'correct' analysis will alter the findings)?
2) How much detail should the referee require of highly complex analyses that would not be understood by most readers?
3) Should the referee take account of the expectation that the authors have no access to expert statistical help? If so, how?
4) How should the referee deal with papers using methods that are (widely) used by other statisticians but which he/she does not approve of? (I have been feeling this burn HARD with the number of papers submitted using stepwise selection & propensity-score matching)
5) When is a definite error a fatal flaw?
That's a great list of challenging questions. I think all statistical referees wrestle with these. I certainly do.
