Have retraction notices improved over time?

Evelyne Decullier
Hervé Maisonneuve

Evelyne Decullier & Hervé Maisonneuve have been studying retractions for a long time. They’ve looked at how long retractions take to show up in PubMed, and five years ago they published a paper on the quality of retraction notices — and how well they were disseminated — in 2008. Now, they’ve repeated that analysis for papers retracted in 2016, and in a new paper in BMC Research Notes, conclude that “management of retraction has improved.” We asked them some questions — one of which, about an Elsevier policy, as noted below, led to a re-examination of the conclusions — about their findings:

Retraction Watch (RW): Why do you think it’s important for journals to provide a reason for retraction?

Evelyne Decullier & Hervé Maisonneuve (ED and HM): Correcting the literature is key for ensuring the quality of data and that the scientific method is respected. Readers should at least be able to differentiate retractions for honest errors from retractions for fraud or plagiarism.

Their confidence in the authors’ integrity is linked to the kind of retraction. We insist on the importance of the reader knowing whether the data, and which data, are valid.

Indeed, readers must be able to distinguish retracted articles, for which the information is completely false (ex: fabrication, falsification), from cases where information is scientifically valid (ex: authorship conflicts). When articles are retracted for honest error, readers should know if all or only part of the data are involved.

RW: Did anything about your findings surprise you?

ED and HM: We were positively impressed by the quality of retraction notices in 2016, as nearly all (99.2%) provided a reason for the retractions. Using the same methods, we found 91% for a sample of notices extracted in 2008. We did not expect to get such a high percentage and that is encouraging.  This result means that editors are concerned about quality and are paying more attention their retraction notices.

RW: You conclude that “management of retraction has improved with time.” What do you think led to that improvement?

ED and HM: We can suggest several possible reasons: a greater awareness by authors and editors that correcting the literature is a sign of integrity; the science community, including stakeholders, understands that hiding reasons for retraction is short sighted, and can damage the image of researchers, institutions and journals; and editors don’t want to be criticized for poor management of retractions. We also think that this improvement is a result of articles pointing out the growing number of retractions, the number of journals having no policy for retraction and citing poor research management.

RW: You excluded Elsevier’s “withdrawals” they were of in-press papers. But these papers have a DOI, and can be cited, so it’s unclear how different they really are from published papers. Our concern with Elsevier’s policy is that it allows papers to effectively be retracted without giving any reason. If the 16 withdrawals had been included, your study would have had 139 retractions, and 17 — or 12% — would have lacked a reason, which would have been even worse than the findings from 2008 (8%). What are your thoughts?

ED and HM: Good observation!

We spent a lot of time discussing how to manage such cases: “withdrawal” does not seem to fall into the usual categories: erratum, expression of concern, or retraction.

“Withdrawal” is an Elsevier policy. Therefore, there would seem to be no scientific ground for this category named “withdrawal”. Elsevier withdrawals occur before any final publication, and “withdrawn” in-press papers are simply removed from the journal’s website without any trace of the original article. These are the reasons why we decided not to include “withdrawals” as a standard retraction category. Moreover, we did not have withdrawals in our 2008 research.

The “withdrawal” status usually corresponds to the following notice that replaces the paper on the journal website: “This article has been withdrawn at the demand of the authors and editor. The Publisher apologizes for any inconvenience this may cause. The full Elsevier Policy on Article Withdrawal can be found at https://www.elsevier.com/about/policies/article-withdrawal

We don’t see any reason why this policy should be continued, and fully agree with your above-mentioned 2013 post on RW. Yes, retracting a paper is acceptable for in-press papers. But hiding the reason for “withdrawal” seems contrary to the transparency objectives of the scientific community.

Being disturbed by your question, we checked the current (31 July 2018) status of the 16 withdrawals (issued between 1 March to 30 June 2016), and we observed that:

  • 14 of the 16 articles still have the withdrawn status. One was a withdrawal of a correction, and not of a paper. One was a temporary removal that was later qualified as withdrawn;
  • one paper is currently published with no more mention of the withdrawal (we archived the withdrawal notice);
  • one is still (31 July 2018) withdrawn by an Elsevier journal (Comput Med Imaging Graph), but a paper (same title, same authors) was later published by PLoS One without any mention of the previous withdrawal. We cannot compare the two original drafts, as withdrawn papers are removed.

We will submit a detailed comment to our BMC Res Notes paper.

Your suggestion to include withdrawals as an ‘others’ class may have changed our ‘positive’ conclusion. If that’s the case, then a conservative conclusion should be that the quality of retraction notices (including “withdrawals”) did not improve, and this is a legitimate concern. However, withdrawals were not retrieved from our previous search, and maybe Elsevier withdrew papers without providing any information before publicizing their “withdrawal” policy. If that was the case, we cannot compare the results of our 2 studies.

The number of “withdrawals” was significant (16 / 123 retractions), and this policy should be abandoned. We do not see any difference between retracted in-press and retracted published papers, as both are accepted papers, and both have a DOI. Other publishers don’t seem to follow Elsevier’s withdrawal policy.

Nevertheless, we still believe that management of retractions has improved with time.

RW: What else could improve handling of retractions?

ED and HM: Scientific journals published by small publishing houses do not have the support of lawyers, and need help to better manage retractions. We propose that retraction notices should be standardized to better inform readers, although we recognize that this need may not be felt by major publishers. Our central concern is the validity of data: which data are still valid in a retracted paper? The annual number of retractions is very low compared to the volume of questionable research practices; therefore we would expect the number of retractions to be higher.

Retractions should be better recognized as an honorable process. However, because the word “retraction” is so strong, carrying a negative connotation, researchers discovering “honest errors” may be reluctant to declare them. Should we perhaps introduce a different wording for retractions of this nature? When publishing retraction’ notices, we should not focus on the misconduct of researchers: rather, all stakeholders have responsibilities and a role to play in furthering scientific integrity.

Like Retraction Watch? You can make a tax-deductible contribution to support our growth, follow us on Twitter, like us on Facebook, add us to your RSS reader, sign up for an email every time there’s a new post (look for the “follow” button at the lower right part of your screen), or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at team@retractionwatch.com.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.