Following a complaint from a reader, editors at the U.S.-based publisher Institute of Electrical and Electronic Engineers (IEEE) determined the researchers behind two decade-old papers had committed “self-plagiarism,” charges the authors deny, Retraction Watch has learned.
However, IEEE passed the buck on to Elsevier, which published one of the articles a month after IEEE had published the other. Elsevier, in turn, said it is wrapping up its investigation and will make the conclusions public “once final.” And one of the authors said a corrigendum is in the works.
The studies share three authors, including last author Li-Qun Zhang, a professor in the Department of Physical Therapy and Rehabilitation Science at University of Maryland School of Medicine, and both focused on movement of the knee as it relates to people with osteoarthritis.
Real-Time Knee Adduction Moment Feedback Training Using an Elliptical Trainer was published online November 14, 2013, in IEEE Transactions on Neural Systems and Rehabilitation Engineering, whereas Real-time tracking of knee adduction moment in patients with knee osteoarthritis appeared online December 19 of that same year in Elsevier’s Journal of Neuroscience Methods. Only the latter title states when it first received the manuscript, which was on June 10, 2013.
In September 2023, someone who identified himself to Retraction Watch as a friend of someone who has worked with Zhang in the past raised concerns about “self-plagiarism” to Daniel Ferris, the former editor-in-chief of IEEE Transactions on Neural Systems and Rehabilitation Engineering, after noticing “substantial overlapped contents, contexts, and conclusions” between the two articles.
The friend told Retraction Watch:
The overlaps are significant, so I did not elaborate the details one by one. Before reporting, we posted those two papers on a public website (ResearchGate) to see how people thought about them. Most of them said like the authors merely changed the title.
In December 2023, Ferris responded to the friend, stating that an independent committee of associate editors reviewed the complaint and the authors had been informed. After examining the evidence, “the committee came to the conclusion that self-plagiarism had been committed as some data had been re-used and there was considerable overlap in the papers,” Ferris told the friend.
Ferris also stated the journal would take an “appropriate action” as a result of the situation. But when Retraction Watch contacted Ferris about what this action might be, he told us:
The journal followed IEEE policy as outlined in the IEEE operations manual (https://www.ieee.org/publications/rights/section-822.html). After following procedures to get a response from the authors and forming a committee to review the allegations, a report was sent to the IEEE Publications office. They determined that because the IEEE TNSRE paper was first, there was no misconduct under the jurisdiction of IEEE.
Giuseppe Di Giovanni, editor-in-chief of the Journal of Neuroscience Methods, told Retraction Watch the “case is still in progress.”
Megan Monachino, Senior Publisher of Life Sciences for Elsevier, added:
We are in the final stages of our investigation. At this stage, the details of the case are confidential; however, the conclusion will be made part of the publication record once final.
Meanwhile, Zhang denies the allegations, but said a corrigendum is in the works. He told Retraction Watch:
The first paper with IEEE (Kang et al.,2013) was on the methodology development of real-time knee moment determination with experimental corroborations on healthy subjects, while the second JNM paper was on subsequent clinical study of patients with knee osteoarthritis and biomechanical changes with knee OA. The differences can be seen in the Results figures of the two papers. The lists of authors for the two papers were also different. However, the Methods section of the second paper was similar to that of the first (Kang et al.,2013). Since the computations involved in the studies were complex, and to facilitate readers, the key method and procedure were described in a simplified manner in the second JNM paper. To avoid confusion, we did not change the frequently occurring terminologies and symbols in this paper, which also contributed to the overlap (overall 31% similarity between the two papers). Moreover, the first paper (Kang et al.,2013) paper was cited throughout the second paper.
The first paper with IEEE was published electronically in 2013, before the second JNM paper published in 2014. So the self-plagiarism issue is with the second JNM paper. The JNM editor investigated the issue carefully and concluded that there is no case of duplicate publication. However, there is 31% similarity in the text. We have submitted a corrigendum accordingly.
Update, 2/15/24, 0300 UTC:
Update: On February 1, two weeks after we requested more details about Elsevier’s investigation, the publisher issued a lengthy corrigendum to Zhang et al’s paper. It reads
The authors regret the following:
Figure 1 on the experimental setup was modified but not completely redrawn to make it different from figure 1 in (Kang et al., 2013c). The authors’ previous study (Kang et al., 2013c) was a new method at the time of publication of this study and the computation involved was complex. Thus, we provided a similar figure explaining the method including the many variables/symbols used in the paper. Despite the differences between the two setup figures, a copyright clearance for Figure 1 in (Kang et al., 2013c) was obtained from IEEE and appears below:
© 2014 IEEE. Reprinted, with permission, from S. H. Kang, S. J. Lee, Y. Ren and L. -Q. Zhang, “Real-Time Knee Adduction Moment Feedback Training Using an Elliptical Trainer,” in IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 22, no. 2, pp. 334–343, March 2014, doi: 10.1109/TNSRE.2013.2291203.
The (Kang et al., 2013c) paper was on the methodology development of real-time knee moment determination with experimental corroborations on healthy subjects, while this JNM study was on subsequent clinical application on patients with knee osteoarthritis and biomechanical changes with knee OA. However, the Methods section of this study is similar to that of (Kang et al., 2013c). Since the computations involved in the studies were complex, and to facilitate readers, the key method and procedure were paraphrased and described in a simplified manner in this paper. To avoid confusion, we did not change the frequently occurring symbols and terminologies in this paper, which contributed to the text overlap. Moreover, the (Kang et al., 2013c) paper was cited in this study as follows.
Introduction
2nd page, 1st column, 2nd, 3rd and 4th paragraphs
2nd page, 2nd column, 1st, 2nd and 3rd paragraphs
– 2.2 Experimental procedure
3rd page, 1st column, 3rd paragraph
– 2.3 Method for estimating external knee adduction moment in real-time
3rd page, 2nd column, 1st paragraph
– 2.3.1 Modified elliptical trainer with the 6-DOF goniometer
3rd page, 2nd column, 2nd paragraph
– 2.3.2.1 Frames of reference
3rd page, 2nd column, 4th paragraph
– 2.3.2.2 Kinematic measurements
4th page, 1st column, 2nd, 3rd and 4th paragraphs
4th page, 2nd column, 1st paragraph
– 2.3.3 Modified 3-D inverse dynamics
4th page, 2nd column, 2nd and 3rd paragraphs
– 2.4 Data analysis
5th page, 2nd column, 2nd paragraph
– 3.1 3-D ankle angles
6th page, 1st column, 2nd paragraph
– 4. Discussion and conclusion
7th page, 1st column, 1st and 2nd paragraphs
– 4.1. Measurement methods of kinematic variables
7th page, 1st column, 4th paragraph
– 4.2. Potential clinical and practical benefits of real-time EKAM estimation
7th page, 2nd column, 4th paragraph
The authors would like to apologise for any inconvenience caused.
The publisher first alerted us to the publication of the corrigendum today, nearly a month after our request for details.
Like Retraction Watch? You can make a tax-deductible contribution to support our work, subscribe to our free daily digest or paid weekly update, follow us on Twitter, like us on Facebook, or add us to your RSS reader. If you find a retraction that’s not in The Retraction Watch Database, you can let us know here. For comments or feedback, email us at [email protected].
Dear Retraction Watch team do realize that there is someone out there that makes it a personal crusade to slander Li-Qun Zhang https://li-qunzhangadeceitfulresearche.godaddysites.com and all over the place (like in RG) sound like a broken record. In for example the following discussions on ResearchGate https://www.researchgate.net/post/About_discussions_on_politics/2 and https://www.researchgate.net/post/Lets_talk_about_some_bad_practices_in_Research two names pop-up:
Dongwon Kim (his profile is removed in the meantime) and (a fake profile?) David Johnson.
All I see is that this whole long story is fueled by personal frustration and enormous unsubstantiated accusations towards respected scientists (and their research group).
As a statistician/methodologist, the explanation of the overlap in the papers makes perfect sense to me. There are times that I’ll work on a study where we have to develop new methodology, and end up writing two papers aimed at two different audiences: (1) clinicians, and (2) those interested in the methodology.
In this case, it looks like the IEEE paper was aimed at those interested in learning more details of the new instrumentation, while the Transactions paper was aimed at clinicians interested in using this instrumentation in clinical practice. It makes sense that the methods sections of the two papers would overlap and that there might be some data overlap, while the focuses of the two papers would differ as would their audiences.
Furthermore, it is common for Methods sections to be close to identical if multiple papers come out of a study. I’ve worked on clinical trials where there were multiple papers with virtually the same methods section because most of it described details of the trial. There’s no reason to re-write sections describing the trial and its participants.
There’s enough real plagiarism out there to keep journals and Retraction Watch busy.
I agree that some degree of self-plagiarism in decade-old papers that only seem to have a combined total of 15 citations on Google Scholar doesn’t feel like the most important thing to focus on.
The scenario you are painting is not typical for engineering papers. We typically don’t have a methods section at all. And we do not have different papers for different audiences.
How do you do science without methods?
The absence of a section named Methods does not imply the lack of methods. Sigh.
So replication is not a big issue in engineering?
This pair of papers fall in the cusp between engineering and medicine. It is a common bifurcation – clinical journal readers want to know the parts that are relevant to their patients, and neither care about nor are competent to understand the engineering aspects. However, it is important to share the engineering methodology and that should go in a journal for people who will understand and potentially make use of it.
One might forgive some degree of text overlap in methods sections for the reasons Meredith W. states, though I note that it wouldn’t be that much trouble to be transparent with the reader by adding a simple sentence such as ‘the following method is taken verbatim or near verbatim from [citation]. Be that as it may, it seems to me that data overlap is an entirely different story. The provenance of data should never, ever come into question. If the same data appear twice or more times, there should be some indication as to their earlier publication. If the data will be appearing in two or more simultaneous publications then there must be some cross-referencing indicating their duplicate appearance. Of course, exceptions may occur as when the data come from a longitudinal project (e.g., Framingham Heart study) where baseline control data appear repeatedly across studies. In such cases and in the absence of a direct cross-reference citation, a reasonable reader, especially someone familiar with the literature, would understand that some of the (control?) data were also part of earlier presentations of the same long-running study.
Who does “self plagarism” actually harm? Publishers? Who cares enough to tattle tale to the publishers?
Self-plagiarism indeed harms publishers, but also the more honest scientists. Publishers are harmed because citations to papers give the journal more credibility, so if the same information is published several places, citations may get diluted between those different outlets. Honest scientists are harmed because the self-plagiarist gets more publications with (almost) the same amount of work. However much we dislike all the metrics being used, they sadly are, meaning that some decisions are made based on scientist A having more publications than scientist B. In some countries it is even worse, as publication in internationals gives the authors money. That money comes from somewhere and is generally not infinite.
Marco, I think when someone asks “Who cares enough” this is a signal that you don’t need to reply. The actual answer is that they should spend more time reading things that do interest them, and less time on things that don’t. Every comment section on any forum will have instances of people claiming to have no interest in the subject of their comments, and they add no value.
Perhaps that the original commenter wasn’t interested, but sometimes you respond to the (hopefully) benefit of others who might wonder how self-plagiarism can be a problem.
It is not a matter of “harm” but a matter of fraud. Authors must attest that the work submitted is new and not published elsewhere. If they lifted text wholesale from anywhere (including their own previously published word), then the attestation of novelty is false. By submitting a false attestation, the authors have committed fraud. No fraud is not acceptable in the scientific enterprise, even if some readers might consider it “harmless”.