As a professional statistician, you are called by a colleague to examine and "bless" a biomedical experimental report. You are urged to do it quickly because the report has already been submitted and accepted for publication in a prestigious journal in the author's field. One of the reviewers, however, had suggested that a quick review by a statistician might be in order. To your horror, the report appears to be utter statistical nonsense. The data were not sampled according to any plan, but rather were drawn from various similar experiments done for different purposes. There is no reason to assume the observations were random or independent within or among data sets. There was no definition of how many data points had been originally available or how those used had been selected. The scatter plots within the paper were plainly skewed, but the computer statistical tests which had been run would have presumed a normal distribution. You explain gently that the statistical work is not an asset to the paper and could prove embarrassing to the author and the institution if published. You suggest that he eliminate the statistical portions and describe his work based on the qualitative reasoning which he obviously used. Initially very angry, he calms down and says, "I'll leave the contents alone, but I will add you as a coauthor. How's that?"

How do you reply? How is your reply conditioned by the relative power positions you may hold? If you are unable to reach an accommodation with the author, under what conditions, if any, would you write to the journal editor to preclude publication? Under what conditions, if any, would you decline to comment on the paper yourself, but refer the author to another colleague whose statistical expertise you consider to be so minimal that he or she might approve the paper as written?
 
 

Posted on Thursday, May 14, 1998, by John Gardenier, drgarden@erols.com. This is taken from his 1997 Joint Statistical Meetings paper, Toward a Statistical Ethics Casebook, in a session sponsored by the section on Teaching of Statistics in the Health Sciences.