You are here

Case discussion: self-plagiarism and suspected salami publishing

Case Summary

Case 18-13 Self-plagiarism and suspected salami publishing

Paper A was published in journal A in January 2018 (accepted, June 2017), but the editors then found that a similar paper, paper B, was published by mostly the same authors in journal B in February 2018 (accepted, November 2017). Journal A’s editors suspected self-plagiarism and potential salami publication, because paper B had the same presentation format and study design as paper A, as well as recycled text in most of the Materials and Discussion sections.

Question(s) for the COPE Forum

• When self-plagiarism and suspected salami publishing is found in a published article, what can the editor do?

• Should the editor inform the other journal editor?

• In such cases, should the article be retracted from both journals?

Forum Advice and Follow-up

The Forum cited the COPE/BioMed Central text recycling guidelines but treated this as a case of redundant or salami publication, noting that journal B should consider retraction because paper A was published first. Journal A’s editor should contact journal B, following COPE’s guidance sharing of information among Editors-in-Chief regarding possible misconduct. The authors should be invited to explain, and an educational stance should be taken.

Journal A’s editor used COPE’s flowchart on suspected redundant publication in a published manuscript and asked the authors for an explanation. The editorial board concluded that this case did not represent redundant or salami publication, but did satisfy its criteria for text recycling. A relevant note was added to paper A and journal B was informed.

Case Discussion

This archived COPE Forum case is categorised under two COPE Core Practices, namely:

  • Allegations of Misconduct, which states: “Journals should have a clearly described process for handling allegations, however they are brought to the journal's or publisher’s attention. Journals must take seriously allegations of misconduct pre-publication and post-publication. Policies should include how to handle allegations from whistleblowers.”
  • Post-publication Discussions and Corrections, which states: “Journals must allow debate post publication either on their site, through letters to the editor, or on an external moderated site, such as PubPeer. They must have mechanisms for correcting, revising or retracting articles after publication.”

The scholarly community values originality, honesty, and transparency in the research record, and hence expects correct attribution of past work and valid claims of new work. An area of growing concern is ‘self-plagiarism’ of one’s past work to portray it as new (eg, unattributed paraphrasing, or unquoted copying with or without attribution). This COPE Forum case (18-13) was eventually treated as one of text recycling rather than redundant or salami publication. The three phenomena discussed in the case are closely related and may overlap. The COPE taxonomy of publication ethics cases lists the following relevant definitions:

  • Redundant/duplicate publication: The publication, or attempted publication, of whole or substantial parts of work/data/analysis that have already been published (or have been submitted elsewhere), without transparency or appropriate declaration/referencing.
  • Plagiarism: Taking/using/presenting others’ ideas, data/results, writings and inventions without giving due or appropriate credit to the originator
  • ‘Self-plagiarism’ / text recycling: Reusing one’s own previous writing without being transparent about this or appropriately referencing/quoting from the original.

The COPE/BioMed Central text recycling guidelines defines text recycling as “when sections of the same text appear (usually unattributed) in more than one of an author’s own publications.” Minor breaches may be remedied by publishing a correction that includes correct citation. The guidelines distinguish between text recycling and redundant/duplicate publication (larger-scale non-transparent republication of data or ideas). A paper containing recycled text could be treated as redundant/duplicate publication that warrants retraction if it presents previously published data in a Results section and there is not enough new material to warrant a new paper.

The definition of “salami publication” is less clear. It may include text/data recycling, but with just enough new data presented in a minimal (minimum/smallest/least) publishable unit. According to COPE’s flowchart suspected redundant publication in a published manuscript, salami publishing involves a lesser degree of redundancy that requires correction (eg, through quoting/paraphrasing and citation of the earlier data). Citation and avoidance of redundancy are especially important to reduce bias in future systematic reviews and meta-analyses.

In a different COPE Forum case (05-07 salami publication), the Forum defined salami publication as “where papers cover the same population, methods, and question” but divide up the outcomes. If overlapping papers have separate questions/hypotheses, are appropriately referenced, and submitted with necessary declarations, editors can make a judgement call. Interestingly, that Forum proposed that salami publication be treated as redundant publication if there is a two-thirds overlap of data.

Perhaps in the same line of thinking, the Forum in the present case recommended retraction of the later paper in journal B because of possible redundancy/salami involving substantial overlap. Text-matching software may be useful in detecting substantial textual overlap, either before or after publication, and journals should routinely screen submissions. However, staff should not rely solely on a certain cut-off percentage score (as advised in COPE Forum cases 15-16 profusion of copied text passages and 16-21 suspected unattributed text in a published article). Additionally, text-matching software cannot detect thorough but unattributed paraphrasing. An editor’s or reviewer’s judgement is ultimately needed in terms of assessing content, location, and extent of overlapping material, as well as checking for (mis)attribution, to decide whether text overlap between works constitutes text recycling, redundancy/duplication, partial redundancy, salami publication, or plagiarism. As for assessing intent, editors need to question the author/s in a neutral way, take an educational approach if authors are new to academia, or inform the institution if misconduct is suspected.

In the COPE/BioMed Central text recycling guidelines, suggested remedial steps for text recycling in a published paper are correction or, rarely, retraction of the later publication. Currently, no action is expected from the publisher of the first source. However, in the present case, journal A posted a notice in paper A about text recycling, presumably citing/linking to paper B. Perhaps the journal was drawing a parallel to the resolution step for confirmed redundancy in COPE’s flowchart on suspected redundant publication in a published manuscript. If a later paper is a duplicate of an earlier one, the earlier source adds a Statement of Redundant Publication, citing the later paper. The later source should publish a retraction, giving redundancy/duplication as the reason and citing the earlier paper.

In the present case, one expects that after investigation by journal B, paper B would at least need correction by indicating quotes or paraphrasing, together with relevant citations if needed. (It was not stated whether paper B cited the passages of recycled text from paper A.) It is generally agreed that, even with citation, simple copying of large portions of previously published text is poor scholarly practice. In a previous COPE Forum case (09-21 self plagiarism) involving text recycling in a well-cited review article, the presenting editor pointed out that a review should be a critical synthesis of works rather than direct unquoted reproduction of long passages of an author’s past work.

Substantive text recycling with or without attribution has added layers of technical and legal complexity, in terms of copyright ownership of the original work (which may be held by the publisher or shared by multiple authors) and whether the authors of earlier and later works are identical. In the case under discussion, the authors were not identical, so it could be argued that without the relevant permissions, the new author/s in paper B committed plagiarism of the affected passages.

Another COPE Core Practice is relevant to the issues in this case and to avoiding disputes in the future, namely:

  • Intellectual Property, which states: “All policies on intellectual property, including copyright and publishing licenses, should be clearly described. In addition, any costs associated with publishing should be obvious to authors and readers. Policies should be clear on what counts as prepublication that will preclude consideration. What constitutes plagiarism and redundant/overlapping publication should be specified.”

Journals should clearly define what counts and does not count as prior publication, as well as define (self-)plagiarism and redundancy. Having clear policies in the author instructions may have helped avoid the events leading up to the present case, as well as in two previous COPE Forum cases: one of alleged self-plagiarism of a student thesis (10-18 self plagiarism) and one of alleged self-plagiarism/redundancy because of a previously posted paper on a personal webpage (14-10 possible self-plagiarism and/or prior publication). Indeed, some journals require a declaration of originality of all submissions and request a statement plus copies of any potentially overlapping material in preparation or already published. Such a policy may help detect and prevent text recycling from an author’s unpublished work, which was alleged to have happened in COPE Forum case 16-21 suspected unattributed text in a published article.

It must be noted that most journals do not regard conference abstracts and trial registry records as prior publication, thereby avoiding the issue of self-plagiarism of those sources. Furthermore, the COPE/BioMed Central text recycling guidelines mention possible tolerance of a reasonable degree of text recycling, with citation, in the Methods section in papers in some fields. This is an evolving area and might be allowed by some jurisdictions as “fair use” without the need for copyright clearance. Clear journal policies about any allowed limited and cited text recycling in the Methods would help. However, policies also probably need to accommodate allowed reuse of material from newer, innovative types of advance methodological documents that aim to improve study transparency. Examples are preregistered protocols, registered reports, protocol papers, and cohort profile papers.

Finally, it was clear in the present case which of the two journals accepted and published the authors’ first paper. Proving which journal was the first source may at times be difficult. This case serves as a reminder that editorial offices should keep accurate and dated records at each stage of manuscript processing, including signed copyright transfer and/or publishing agreements.

Trevor Lane on behalf of the COPE Education Subcommittee