How well do journals publicize retractions?
A new paper in BMC Research Notes looks at the retraction class of 2008, and finds journals’ handling of them less than optimal.
Evelynne Decullier and colleagues — including Hervé Maisonneuve, who was helpful to us for a recent post — found:
Overall, 244 retractions were considered for analysis. Formal retraction notices could not be retrieved for 9. Of the 235 retractions available (96%), the reason was not detailed for 21 articles (9%). The most cited reasons were mistakes (28%), plagiarism (20%), fraud (14%) and overlap (11%). The original paper or its location was found for 233 retractions (95%). Of these, 22% were available with no mention of the retraction.
In 2009 — the year after the retractions Decullier and colleagues studied — the Committee on Publication Ethics (COPE) issued retraction guidelines. As the new paper’s authors note:
COPE’s guidelines also state that notices of retraction should clearly identify the retracted article (title and authors) and be linked with the retracted article. The notices should be available freely and state the reason for retraction (without being defamatory) and who is retracting.
It would be interesting, then, to do the same kind of analysis for papers retractions published in 2010, 2011, or 2012, to see if adherence to COPE’s suggestions for transparency had improved. When Grant Steen looked at a larger dataset of retractions over a decade, he found that 31% of retracted papers were not watermarked or otherwise flagged as withdrawn.
We also asked Ferric Fang, who co-authored a paper that revealed how poorly many retraction notices describe what had actually led to a withdrawal, for his take on the given reasons for retractions in the class of 2008:
The results of this study do not agree with ours, but this is not surprising since these authors relied upon retraction notices to ascertain the reason for retraction.
Our databases of articles were not precisely the same but should be similar. I have looked at articles listed in our database that were retracted in 2008 and compared our results with those of Decullier et al.. Those authors report the distribution of reasons for retraction to be fraud (14%), error (28%), plagiarism (20%) and duplicate publication (11%). In contrast, during the same year we found fraud or suspected fraud (38%), error (17%), plagiarism (18%) and duplicate publication (14%).
This supports our conclusion that a reliance on retraction notices will result in an overestimation of error and an underestimation of fraud as reasons for retraction. It also underscores the importance of sources such as Retraction Watch for understanding the true reason that an article has been retracted.