Remember last week, when JAMA replaced an article about the impact of moving homes on kids’ mental health after discovering some errors in the analysis? We’re going to see more of these somewhat unusual notices coming out of JAMA journals in the near future – the JAMA Network journals may issue more “retract and replace” decisions for papers, in which it pulls an old version of an article and replaces it with a corrected one. But it’s not a correction — we spoke with Annette Flanagin, the Executive Managing Editor for The JAMA Network, to learn more.
Retraction Watch: We’ve spotted three “retract and replace” notices in JAMA journals, including one issued earlier this year for a highly cited paper that contained “pervasive errors,” and the one last week about the impact of moving on kids’ mental health. How do you decide whether a paper will be retracted and replaced, or just retracted?
Annette Flanagin: For articles with confirmed research misconduct (eg, fabrication, falsification), we will publish a Notice of Retraction and retract the article. However, several studies have shown that about 20% of retractions are due to some type of major error, not misconduct. And we need a mechanism to address honest pervasive error (ie, unintentional human or programmatic errors that result in the need to correct numerous data and text in the abstract, text, tables and figures, such as a coding error) without the current stigma that is associated with retraction. Thus, for articles with honest pervasive error, in which the corrections needed to address the errors result in statistically significant changes to the findings, interpretations, or conclusions and for which the methodology/science is still valid, we will consider publication of a Notice of Retraction and Replacement. In such cases, we would also publish a Letter of explanation from the authors. We hope this will allow readers/users to access and use the work (in its replaced version) – without being stopped cold by a Do Not Use Retraction notice and Retraction watermark on the article.
RW: Why not simply correct the original paper with the updated material, and issue a correction notice?
AF: We would indeed publish a Correction Notice, with a Letter of explanation from the authors, if an article had pervasive error but the corrections did not result in statistically significant changes to the findings, interpretations, or conclusions. I believe that some have labeled this a “mega correction.”
RW: Are you worried readers will be confused about whether or not the paper has been retracted?
AF: A formal Notice of Retraction and Replacement is published and cross-linked with the retracted and replaced article. This is a new model and article type. We have developed and tested new XML coding and display of the notices and linking on the html and PDF versions of the articles with the intent of making this clear to readers/users of the content. For those interested, we also republish a version of the original retracted article with the errors highlighted and another version of the replacement article with the corrections highlighted as an online supplement to the replaced article. The Lancet journals also published 2 similar Notices of Retraction and Replacement last year.
RW: Do you expect authors to count this as a retraction on their record, or a correction? Are you concerned they could just cite the new paper and no one would ever know the original was retracted?
AF: What authors do is their decision, but they do want to make sure that their CVs are correct. In this regard, authors should include a note or citation that includes Notice of Retraction and Replacement – as they would do with a Correction notice. In both situations, we hope there is less stigma associated with errors, minor and pervasive, and the need to correct or retract and replace the literature. A citation might follow this format:
“Authors. Title. Journal. Year; Volume(Issue): page numbers or e-locator. Retracted and replaced on month, day, year. doi”
RW: Earlier this year, Nature published a piece by Daniele Fanelli suggesting there should be a separate mechanism for retracting papers due to “honest error.” Does the JAMA retract/replace function in this capacity?
AF: Absolutely. Often, authors come to us directly to report a pervasive error and work with us as we review the original publication and the circumstances of the errors to best determine whether a Correction or Retraction and Replacement is needed. We have published many Corrections, and thus far, have only processed three Notices of Retraction and Replacement.
In each of these cases, the authors have been forthcoming and transparent and have thanked us for the approach we have taken. In addition, I agree with Fanelli’s comment that authors “who do the right thing” after reporting such mistakes should not experience “citation penalty” or other stigma as a result. With this in mind, for articles that are retracted and replaced, we have decided to retain the DOI of the original article for the replacement article, as we believe the authors should not be penalized and that any citations or other metrics associated with the article before replacement should still be counted.
Like Retraction Watch? Consider making a tax-deductible contribution to support our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, sign up on our homepage for an email every time there’s a new post, or subscribe to our new daily digest. Click here to review our Comments Policy. For a sneak peek at what we’re working on, click here.
“we also republish a version of the original retracted article with the errors highlighted and another version of the replacement article with the corrections highlighted as an online supplement to the replaced article.”
This is a significant positive advance in the evolution of post-publication peer review, so congratulations to JAMA for successfully implementing this, even if only for two papers for now. I hope that this model can be widely applied to other publishers and journals and should be widely disseminated.
I like this model because:
a) it allows the literature to be corrected as many times as necessary, provided that the authors are alive, and able, to do so.
b) it represents the true state of science, i.e., in constant evolution.
Two suggestions to JAMA. I looked at the two retracted+replaced papers:
http://jama.jamanetwork.com/article.aspx?articleid=1835504
http://archpsyc.jamanetwork.com/article.aspx?articleid=1897300
1) Why not make “retracted” papers open access? The site says “FREE”, but one needs to register to access JAMA papers. Why is that?
2) The indication that supplementary content is the actual “retracted” or “corrected” paper is not immediately clear to the reader (I guess this is a period of conditioning while readers get used to the idea). But it wouldn’t hurt to differentiate a regular supplementary file from a file that shows the errors/retraction, perhaps with different terminology to avoid confusion, such as “Errors/retraction supplement”. That also avoids ambiguities.
Finally, how does JAMA prevent authors from “sneaking in” unpublished data that perhaps they couldn’t get published in another paper, into their retracted+corrected paper? For example, a small “supporting” data set, table or figure?
Any word from IDF on this obliteration of the remaining DOI semantics? What will SK do? Enquiring minds want to know.
As an addendum, I think PLOS is still assigning DOIs to figures, which was part of the original Grand Plan to Assign Long URLs Forever to All the Content Sands in the Hourglass.
Thanks to all, including the comments by Jaime A. Teixeira da Silva and Narad. It might just be that I’m fatigued due to travelling, but I found the conversation between RW and AF very difficult to follow. I hope that it is, or will become, an easily understood practice.