Suspicions were raised on 20 September 2012 by a reviewer who commented that some of the passages in a submission from Dr J were similar to an earlier paper published in our journal by the same author. An iThenticate check indicated a similarity index of 60%: however, the overlap was not from that earlier paper but from another source by a different author which had contributed 41% of the material.
This prompted an iThenticate check of the published paper, which gave a similarity index of 57%, with 45% of the material from three papers by other authors. (It should be noted that this paper was reviewed and accepted before iThenticate was available for checking incoming submissions.)
It was clear that the new submission should be rejected. The key issue was the action to be taken about the paper that had already been published.
The editor of the journal in which two of the key sources had been published kindly provided copies and the published paper was checked by hand against these two earlier papers. This check established that the iThenticate report was reasonably accurate. It appeared that one of the plagiarized papers had been used as a means to improve the quality of the English while the other had provided a framework for the reporting of the statistical results: Dr J had substituted new figures in the running text of the earlier paper.
COPE guidelines were followed and a carefully worded letter was sent asking Dr J for an explanation. In summary, his reply said that: (a) he was building on the work of the earlier authors, (b) he did not understand or mean to do it and (c) he was very sorry and would not do it again.
Dr J had made six other submissions to our journal, all of which had been rejected on the grounds of quality. iThenticate checks on these revealed similarity indexes between 66% and 77%. Typically up to three sources had been plagiarized to contribute up to 63% of the material. A search using Google Scholar identified that Dr J had published over 20 other papers in different journals since 2005.
In the light of this information, Dr J’s explanation of naivety was considered to be implausible and the decision was taken to retract the published paper. Dr J was given a final opportunity to respond and gave the same explanation for the overlap. The retraction will be published in the next issue of our journal and on the journal website. In view of the extent of the plagiarism, the decision was also taken to inform the president of his institution.
There remains the question of whether the editors of the other journals in which Dr J has published work should also be informed of this case.
The editor would welcome the comments of the Forum on this issue.
Following this incident, the journal has reviewed its policy to detect and discourage plagiarism in submitted work.
- As a matter of routine, the journal now checks all of the work submitted for publication using iThenticate.
- Submissions that appear to include a significant amount of previously published material are investigated further to establish whether that material has been referenced and attributed appropriately.
- Where the overlap is found to exceed an acceptable level, we write to the author(s) providing a link to the full report and inviting them to withdraw the submission, or alternatively to revise it extensively to reduce the overlap and to indicate where they are quoting the work of others (or their own previously published work). We also ask for their comments on the overlap.
- If the author cannot provide an acceptable explanation or where the overlap is very significant, then we will immediately reject the submission.
- The issue of plagiarism is being included in the Journal Reviewer Development Programme to heighten awareness of the problem within the Reviewer Panel.
- We are seeking to engage in discussions and the exchange of information on plagiarism with editors of other journals in the field.
We have also retrospectively checked the overlap of all submissions currently in process and identified several others with unacceptably high similarity indexes. We are asking those authors to withdraw their submissions or to revise them to eliminate the overlap.
Of the 231 submissions that have been checked to date, 71% have an iThenticate similarity index of less than 30%. Over 12% have a similarity index in excess of 40%—the level at which iThenticate gives a plagiarism alert. Excluding the eight submissions from Dr J, there were 9% falling into this category. The remaining 16% fall in the range 30–39% and have been investigated. In all of these cases, the overlap was in acceptable quotations and in the bibliography and no further action has been taken. The, as yet unanswered, question is whether these figures are typical for an international journal.
The Forum agreed with the editor’s course of action and also agreed that the editor could contact the editors of the other journals. The editor could inform the other editors that he is retracting the paper, stating the reasons (eg sending the retraction notice), and saying that he noticed that their journals have also published papers by this author. One suggestion was that the editor could run the other papers (from the other journals) through the plagiarism detection software, but this is very time consuming.
COPE does not recommend using percentages as cut-offs for detecting plagiarism. COPE believes that each paper should be judged individually and by eye after an initial screen. Percentages can mean very different things in different disciplines and in different sections of papers. COPE is considering developing a flowchart for what to do about plagiarism detected using plagiarism detection software.
A retraction was published in the journal. There has been no further communication from the authors. As suggested by the COPE Forum, the editor contacted the editors of the other journals who had published work by the author in the past, drawing their attention to the retraction but without further comment. Most of them have understood the implications of the email. The editor considers the case closed.
The journal continues to get a significant number of papers with high similarity indices. About half are understandable (eg, a paper that makes accessible a report which received very limited circulation, derivatives of theses, etc): others are naivety.